WIP Add AES-XTS client encryption support

Maybe I should remove all of this and use libwebsockets :)

.gitea/workflows/buildenv.Dockerfile

@@ -1,29 +1,28 @@
-FROM node:16-bookworm
+FROM node:16-bullseye
 
 WORKDIR /root
 
-ADD ./docker/etc/apt/trusted.gpg.d /etc/apt/trusted.gpg.d
+ADD ./docker/vitastor.gpg /etc/apt/trusted.gpg.d
 
-RUN echo 'deb http://deb.debian.org/debian bookworm-backports main' >> /etc/apt/sources.list; \
+RUN echo 'deb http://deb.debian.org/debian bullseye-backports main' >> /etc/apt/sources.list; \
-    echo 'deb http://vitastor.io/debian bookworm main' >> /etc/apt/sources.list; \
+    echo 'deb http://vitastor.io/debian bullseye main' >> /etc/apt/sources.list; \
     echo >> /etc/apt/preferences; \
     echo 'Package: *' >> /etc/apt/preferences; \
-    echo 'Pin: release n=bookworm-backports' >> /etc/apt/preferences; \
+    echo 'Pin: release a=bullseye-backports' >> /etc/apt/preferences; \
     echo 'Pin-Priority: 500' >> /etc/apt/preferences; \
     echo >> /etc/apt/preferences; \
     echo 'Package: *' >> /etc/apt/preferences; \
     echo 'Pin: origin "vitastor.io"' >> /etc/apt/preferences; \
     echo 'Pin-Priority: 1000' >> /etc/apt/preferences; \
-    perl -i -pe 's/Types: deb$/Types: deb deb-src/' /etc/apt/sources.list.d/debian.sources; \
     grep '^deb ' /etc/apt/sources.list | perl -pe 's/^deb/deb-src/' >> /etc/apt/sources.list; \
     echo 'APT::Install-Recommends false;' >> /etc/apt/apt.conf; \
     echo 'APT::Install-Suggests false;' >> /etc/apt/apt.conf
 
 RUN apt-get update
-RUN apt-get -y install etcd qemu-system-x86 qemu-block-extra qemu-utils fio libasan8 \
+RUN apt-get -y install etcd qemu-system-x86 qemu-block-extra qemu-utils fio libasan5 \
     libgoogle-perftools-dev devscripts libjerasure-dev cmake libibverbs-dev libisal-dev
 RUN apt-get -y build-dep fio qemu=`dpkg -s qemu-system-x86|grep ^Version:|awk '{print $2}'`
-RUN apt-get update && apt-get -y install jq lp-solve sudo nfs-common fdisk parted libc-ares-dev udev
+RUN apt-get update && apt-get -y install jq lp-solve sudo nfs-common fdisk parted
 RUN apt-get --download-only source fio qemu=`dpkg -s qemu-system-x86|grep ^Version:|awk '{print $2}'`
 
 RUN set -ex; \

.gitea/workflows/test.yml

@@ -63,7 +63,7 @@ jobs:
     container: ${{env.TEST_IMAGE}}:${{github.sha}}
     steps:
     # leak sanitizer sometimes crashes
-    - run: cd /root/vitastor/build && ASAN_OPTIONS=detect_leaks=0 make -j16 build_tests test
+    - run: cd /root/vitastor/build && ASAN_OPTIONS=detect_leaks=0 make -j16 test
 
   npm_lint:
     runs-on: ubuntu-latest
@@ -234,6 +234,60 @@ jobs:
           echo ""
         done
 
+  test_etcd_fail_https:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 10
+      run: ETCD_SCHEME=https /root/vitastor/tests/test_etcd_fail.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
+  test_etcd_fail_https_antietcd:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 10
+      run: ETCD_SCHEME=https ANTIETCD=1 /root/vitastor/tests/test_etcd_fail.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
+  test_snapshot_https:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 3
+      run: ETCD_SCHEME=https /root/vitastor/tests/test_snapshot.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
   test_interrupted_rebalance:
     runs-on: ubuntu-latest
     needs: build
@@ -1224,6 +1278,24 @@ jobs:
           echo ""
         done
 
+  test_checksum_xxhash:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 3
+      run: TEST_NAME=xxhash OSD_ARGS="--data_csum_type xxh3_32" /root/vitastor/tests/test_checksum.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
   test_old_checksum:
     runs-on: ubuntu-latest
     needs: build

.gitea/workflows/tests-to-yaml.pl

@@ -38,6 +38,10 @@ for my $line (<>)
             {
                 $test_name .= '_antietcd';
             }
+            elsif ($1 eq 'ETCD_SCHEME' && $2 eq 'https')
+            {
+                $test_name .= '_https';
+            }
             elsif ($1 eq 'OLD')
             {
                 $test_name =~ s/^test_/test_old_/s;

CMakeLists.txt

@@ -1,20 +1,20 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
-set(VITASTOR_VERSION "3.0.12")
+set(VITASTOR_VERSION "3.0.3")
 
 include(CTest)
 
 add_custom_target(build_tests)
-set_property(TEST PROPERTY ENVIRONMENT LSAN_OPTIONS=suppressions=${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt)
+add_custom_target(test
-add_test(gen_lsan_suppress
+	COMMAND
-    ${CMAKE_COMMAND} -E echo leak:tcmalloc > "${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt"
+	echo leak:tcmalloc > ${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt &&
+	env LSAN_OPTIONS=suppressions=${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt ${CMAKE_CTEST_COMMAND}
 )
-set_tests_properties(gen_lsan_suppress PROPERTIES FIXTURES_SETUP f_lsan_suppress)
-set_property(TEST PROPERTY FIXTURES_REQUIRED f_lsan_suppress)
 # make -j16 -C ../../build test_heap && ../../build/src/test/test_heap
 # make -j16 -C ../../build test_heap && rm -f $(find ../../build -name '*.gcda') && ctest -V -T test -T coverage -R heap --test-dir ../../build && (cd ../../build; gcovr -f ../src --html --html-nested -o coverage/index.html; cd ../src/test)
 # make -j16 -C ../../build test_blockstore && rm -f $(find ../../build -name '*.gcda') && ctest -V -T test -T coverage -R blockstore --test-dir ../../build && (cd ../../build; gcovr -f ../src --html --html-nested -o coverage/index.html; cd ../src/test)
 # kcov --include-path=../../../src ../../kcov ./test_blockstore
+add_dependencies(test build_tests)
 add_subdirectory(src)

csi/Dockerfile

@@ -25,8 +25,8 @@ RUN apt-get update && \
         # NFS mount dependencies
         nfs-common netbase \
         # dependencies of qemu-storage-daemon
-        libaio1t64 libc6 libfuse3-4 libglib2.0-0t64 libgmp10 libgnutls30t64 \
+        libnuma1 liburing2 libglib2.0-0 libfuse3-3 libaio1 libzstd1 libnettle8 \
-        libhogweed6t64 libnettle8t64 libnuma1 libselinux1 liburing2 libzstd1 zlib1g && \
+        libgmp10 libhogweed6 libp11-kit0 libidn2-0 libunistring2 libtasn1-6 libpcre2-8-0 libffi8 && \
     apt-get clean && \
     (echo options nbd nbds_max=128 > /etc/modprobe.d/nbd.conf)
 

csi/Dockerfile.local

@@ -1,5 +1,5 @@
 # Compile stage
-FROM golang:trixie AS build
+FROM golang:bookworm AS build
 
 ADD go.sum go.mod /app/
 RUN cd /app; CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go mod download -x
@@ -9,7 +9,7 @@ RUN perl -i -e '$/ = undef; while(<>) { s/\n\s*(\{\s*\n)/$1\n/g; s/\}(\s*\n\s*)e
     CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go build -o vitastor-csi
 
 # Final stage
-FROM debian:trixie
+FROM debian:bookworm
 
 LABEL maintainers="Vitaliy Filippov <vitalif@yourcmc.ru>"
 LABEL description="Vitastor CSI Driver"
@@ -36,8 +36,8 @@ ADD deb /deb
 
 RUN apt-get update && \
     apt-get -y install /deb/vitastor-client_*.deb && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-9.2.2%2Bds-1%2Bvitastor4/qemu-utils_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-utils_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-9.2.2%2Bds-1%2Bvitastor4/qemu-block-extra_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-block-extra_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
     dpkg -x qemu-utils*.deb tmp1 && \
     dpkg -x qemu-block-extra*.deb tmp1 && \
     cp -a tmp1/usr/bin/qemu-storage-daemon /usr/bin/ && \

csi/Makefile

@@ -1,4 +1,4 @@
-VITASTOR_VERSION ?= v3.0.12
+VITASTOR_VERSION ?= v3.0.3
 
 all: build push
 

csi/deploy/004-csi-nodeplugin.yaml

@@ -49,7 +49,7 @@ spec:
             capabilities:
               add: ["SYS_ADMIN"]
             allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v3.0.12
+          image: vitalif/vitastor-csi:v3.0.3
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/deploy/007-csi-provisioner.yaml

@@ -121,7 +121,7 @@ spec:
             privileged: true
             capabilities:
               add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v3.0.12
+          image: vitalif/vitastor-csi:v3.0.3
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/src/config.go

@@ -5,7 +5,7 @@ package vitastor
 
 const (
     vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "3.0.12"
+    vitastorCSIDriverVersion = "3.0.3"
 )
 
 // Config struct fills the parameters of request or user input

debian/build-vitastor-ubuntu-resolute.sh

@@ -1,5 +0,0 @@
-#!/bin/bash
-# 26.04 Resolute Raccoon
-
-docker build --build-arg DISTRO=ubuntu --build-arg REL=resolute -t vitastor-buildenv:resolute -f vitastor-buildenv.Dockerfile .
-docker run -it --rm -e REL=resolute -v `dirname $0`/../:/root/vitastor vitastor-buildenv:resolute /root/vitastor/debian/vitastor-build.sh

debian/changelog

@@ -1,4 +1,4 @@
-vitastor (3.0.12-1) unstable; urgency=medium
+vitastor (3.0.3-1) unstable; urgency=medium
 
   * Bugfixes
 

debian/vitastor-build.sh

@@ -44,7 +44,7 @@ curl -s https://git.yourcmc.ru/vitalif/antietcd/archive/master.tar.gz | tar -zx
 curl -s https://git.yourcmc.ru/vitalif/tinyraft/archive/master.tar.gz | tar -zx
 
 cd /root/vitastor/packages/vitastor-$REL
-if [[ ( "$REL" = "trixie" || "$REL" = "resolute" ) && -e ../vitastor-bookworm/vitastor_$VER.orig.tar.xz ]]; then
+if [[ "$REL" = "trixie" && -e ../vitastor-bookworm/vitastor_$VER.orig.tar.xz ]]; then
     # Fucking shit, archives differ between bookworm (xz 5.4.1) and trixie (xz 5.8.1)
     cp ../vitastor-bookworm/vitastor_$VER.orig.tar.xz .
 else

docker/Makefile

@@ -1,4 +1,4 @@
-VITASTOR_VERSION ?= v3.0.12
+VITASTOR_VERSION ?= v3.0.3
 
 all: build push
 

docker/etc/apt/preferences

@@ -1,3 +1,3 @@
 Package: *
-Pin: release n=trixie-backports
+Pin: release n=bookworm-backports
 Pin-Priority: 500

docker/etc/apt/sources.list.d/vitastor.list

@@ -1,2 +1,2 @@
-deb http://vitastor.io/debian trixie main
+deb http://vitastor.io/debian bookworm main
-#deb http://http.debian.net/debian/ trixie-backports main
+deb http://http.debian.net/debian/ bookworm-backports main

docker/etc/systemd/system/vitastor-host.service

@@ -7,7 +7,7 @@ PartOf=vitastor.target
 [Service]
 Restart=always
 EnvironmentFile=/etc/vitastor/docker.conf
-ExecStart=bash -c 'docker run --rm -i -v /etc/vitastor:/etc/vitastor -v /dev:/dev -v /run:/run -e SYSTEMD_IN_CHROOT=0 \
+ExecStart=bash -c 'docker run --rm -i -v /etc/vitastor:/etc/vitastor -v /dev:/dev -v /run:/run \
     --security-opt seccomp=unconfined --privileged --pid=host --log-driver none --network host --name vitastor vitastor:$VITASTOR_VERSION \
     sleep.sh'
 ExecStartPost=udevadm trigger

docker/etc/vitastor/docker.conf

@@ -4,7 +4,7 @@
 #
 
 # Desired Vitastor version
-VITASTOR_VERSION=v3.0.12
+VITASTOR_VERSION=v3.0.3
 
 # Additional arguments for all containers
 # For example, you may want to specify a custom logging driver here

docker/install.sh

@@ -2,7 +2,8 @@
 
 set -e
 
-cp -urv /etc/systemd/system/vitastor* /host-etc/systemd/system/
+cp -urv /etc/default /host-etc/
-cp -urv /etc/udev/rules.d /host-etc/udev/
+cp -urv /etc/systemd /host-etc/
+cp -urv /etc/udev /host-etc/
 cp -urnv /etc/vitastor /host-etc/
 cp -urnv /opt/scripts/* /host-bin/

docs/config/osd.en.md

@@ -38,7 +38,6 @@ with an OSD restart or, for some of them, even without restarting by updating co
 - [journal_io](#journal_io)
 - [journal_sector_buffer_count](#journal_sector_buffer_count)
 - [journal_no_same_sector_overwrites](#journal_no_same_sector_overwrites)
-- [skip_corrupted_meta_entries](#skip_corrupted_meta_entries)
 - [throttle_small_writes](#throttle_small_writes)
 - [throttle_target_iops](#throttle_target_iops)
 - [throttle_target_mbs](#throttle_target_mbs)
@@ -70,7 +69,6 @@ with an OSD restart or, for some of them, even without restarting by updating co
 - [use_atomic_flag](#use_atomic_flag)
 - [pg_reshard_chunk_size](#pg_reshard_chunk_size)
 - [pg_reshard_chunk_pause_ms](#pg_reshard_chunk_pause_ms)
-- [gc_on_start](#gc_on_start)
 
 ## bind_address
 
@@ -281,19 +279,13 @@ Maximum number of journal flushers (see above min_flusher_count).
 - Type: boolean
 - Default: true
 
-Only for the old store ([meta_format](layout-osd.en.md#meta_format) 2).
+This parameter makes Vitastor always keep metadata area of the block device
-
+in memory. It's required for good performance because it allows to avoid
-This parameter makes Vitastor keep a copy of metadata area in memory as it is
+additional read-modify-write cycles during metadata modifications. Metadata
-on disk, in addition to the metadata database. When the option is enabled, every
+area size is currently roughly 224 MB per 1 TB of data. You can turn it off
-metadata entry is effectively stored in RAM twice. It's required for good performance
+to reduce memory usage by this value, but it will hurt performance. This
-because it allows to avoid additional read-modify-write cycles during metadata
+restriction is likely to be removed in the future along with the upgrade
-modifications. Metadata area size with the old store is roughly 224 MB per 1 TB
+of the metadata storage scheme.
-of data. You can turn the option off to reduce memory usage by this value, but
-it will reduce performance.
-
-For the new store ([meta_format](layout-osd.en.md#meta_format) 3), the option
-may be changed in the future to support operation without loading full metadata
-database in memory.
 
 ## inmemory_journal
 
@@ -372,8 +364,6 @@ blocks. The only situation when you should increase it to a larger value
 is when you enable journal_no_same_sector_overwrites. In this case set
 it to, for example, 1024.
 
-Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-
 ## journal_no_same_sector_overwrites
 
 - Type: boolean
@@ -387,17 +377,6 @@ journal after writing it instead of possibly overwriting it the second time.
 
 Most (99%) other SSDs don't need this option.
 
-Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-
-## skip_corrupted_meta_entries
-
-- Type: boolean
-- Default: false
-
-Only for the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-Allow OSD to start when some metadata entries or blocks are corrupted by
-skipping them. Should be only used as an emergency measure.
-
 ## throttle_small_writes
 
 - Type: boolean
@@ -754,9 +733,3 @@ This option sets the maximum number of object is a chunk. Moving 100k objects us
 - Default: 100
 
 This option sets the interval between handling two PG count change chunks.
-
-## gc_on_start
-
-- Type: boolean
-
-Forcibly clean all garbage entries in the new store on every OSD restart.

docs/config/osd.ru.md

@@ -39,7 +39,6 @@
 - [journal_io](#journal_io)
 - [journal_sector_buffer_count](#journal_sector_buffer_count)
 - [journal_no_same_sector_overwrites](#journal_no_same_sector_overwrites)
-- [skip_corrupted_meta_entries](#skip_corrupted_meta_entries)
 - [throttle_small_writes](#throttle_small_writes)
 - [throttle_target_iops](#throttle_target_iops)
 - [throttle_target_mbs](#throttle_target_mbs)
@@ -71,7 +70,6 @@
 - [use_atomic_flag](#use_atomic_flag)
 - [pg_reshard_chunk_size](#pg_reshard_chunk_size)
 - [pg_reshard_chunk_pause_ms](#pg_reshard_chunk_pause_ms)
-- [gc_on_start](#gc_on_start)
 
 ## bind_address
 
@@ -289,19 +287,13 @@ Flusher - это микро-поток (корутина), которая коп
 - Тип: булево (да/нет)
 - Значение по умолчанию: true
 
-Только для старого хранилища ([meta_format](layout-osd.en.md#meta_format) 2).
+Данный параметр заставляет Vitastor всегда держать область метаданных диска
-
+в памяти. Это нужно, чтобы избегать дополнительных операций чтения с диска
-Данный параметр заставляет Vitastor всегда держать копию области метаданных
+при записи. Размер области метаданных на данный момент составляет примерно
-в памяти в том же виде, как она лежит на диске, в дополнение к БД метаданных.
+224 МБ на 1 ТБ данных. При включении потребление памяти снизится примерно
-То есть, с включённой опцией каждая запись метаданных хранится в памяти дважды.
+на эту величину, но при этом также снизится и производительность. В будущем,
-Это нужно, чтобы избегать дополнительных операций чтения с диска при записи.
+после обновления схемы хранения метаданных, это ограничение, скорее всего,
-Размер области метаданных в старом хранилище составляет примерно 224 МБ на
+будет ликвидировано.
-1 ТБ данных. Вы можете отключить опцию, чтобы снизить потребление памяти
-примерно на эту величину, но при этом также снизится и производительность.
-
-Для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3) опция,
-возможно, будет переработана в будущем для поддержки работы без полной
-загрузки метаданных в памяти.
 
 ## inmemory_journal
 
@@ -384,8 +376,6 @@ fsync небезопасным даже с режимом "directsync".
 нужно менять - это если вы включаете journal_no_same_sector_overwrites. В
 этом случае установите данный параметр, например, в 1024.
 
-Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-
 ## journal_no_same_sector_overwrites
 
 - Тип: булево (да/нет)
@@ -401,18 +391,6 @@ fsync небезопасным даже с режимом "directsync".
 
 Почти все другие SSD (99% моделей) не требуют данной опции.
 
-Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-
-## skip_corrupted_meta_entries
-
-- Тип: булево (да/нет)
-- Значение по умолчанию: false
-
-Только для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3).
-Разрешить OSD запускаться, даже если часть блоков или записей метаданных
-повреждена, пропуская их. Опция предназначена для использования только в
-целях аварийного восстановления.
-
 ## throttle_small_writes
 
 - Тип: булево (да/нет)
@@ -794,9 +772,3 @@ pg_minsize OSD во время переключений, что может по
 - Значение по умолчанию: 100
 
 Данная опция задаёт интервал между обработкой двух порций изменения числа PG пулов.
-
-## gc_on_start
-
-- Тип: булево (да/нет)
-
-Принудительно очищать все мусорные записи в новом хранилище при каждом запуске OSD.

docs/config/src/osd.yml

@@ -253,33 +253,21 @@
   type: bool
   default: true
   info: |
-    Only for the old store ([meta_format](layout-osd.en.md#meta_format) 2).
+    This parameter makes Vitastor always keep metadata area of the block device
-
+    in memory. It's required for good performance because it allows to avoid
-    This parameter makes Vitastor keep a copy of metadata area in memory as it is
+    additional read-modify-write cycles during metadata modifications. Metadata
-    on disk, in addition to the metadata database. When the option is enabled, every
+    area size is currently roughly 224 MB per 1 TB of data. You can turn it off
-    metadata entry is effectively stored in RAM twice. It's required for good performance
+    to reduce memory usage by this value, but it will hurt performance. This
-    because it allows to avoid additional read-modify-write cycles during metadata
+    restriction is likely to be removed in the future along with the upgrade
-    modifications. Metadata area size with the old store is roughly 224 MB per 1 TB
+    of the metadata storage scheme.
-    of data. You can turn the option off to reduce memory usage by this value, but
-    it will reduce performance.
-
-    For the new store ([meta_format](layout-osd.en.md#meta_format) 3), the option
-    may be changed in the future to support operation without loading full metadata
-    database in memory.
   info_ru: |
-    Только для старого хранилища ([meta_format](layout-osd.en.md#meta_format) 2).
+    Данный параметр заставляет Vitastor всегда держать область метаданных диска
-
+    в памяти. Это нужно, чтобы избегать дополнительных операций чтения с диска
-    Данный параметр заставляет Vitastor всегда держать копию области метаданных
+    при записи. Размер области метаданных на данный момент составляет примерно
-    в памяти в том же виде, как она лежит на диске, в дополнение к БД метаданных.
+    224 МБ на 1 ТБ данных. При включении потребление памяти снизится примерно
-    То есть, с включённой опцией каждая запись метаданных хранится в памяти дважды.
+    на эту величину, но при этом также снизится и производительность. В будущем,
-    Это нужно, чтобы избегать дополнительных операций чтения с диска при записи.
+    после обновления схемы хранения метаданных, это ограничение, скорее всего,
-    Размер области метаданных в старом хранилище составляет примерно 224 МБ на
+    будет ликвидировано.
-    1 ТБ данных. Вы можете отключить опцию, чтобы снизить потребление памяти
-    примерно на эту величину, но при этом также снизится и производительность.
-
-    Для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3) опция,
-    возможно, будет переработана в будущем для поддержки работы без полной
-    загрузки метаданных в памяти.
 - name: inmemory_journal
   type: bool
   default: true
@@ -398,15 +386,11 @@
     blocks. The only situation when you should increase it to a larger value
     is when you enable journal_no_same_sector_overwrites. In this case set
     it to, for example, 1024.
-
-    Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
   info_ru: |
     Максимальное число буферов, разрешённых для использования под записываемые
     в журнал блоки метаданных. Единственная ситуация, в которой этот параметр
     нужно менять - это если вы включаете journal_no_same_sector_overwrites. В
     этом случае установите данный параметр, например, в 1024.
-
-    Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
 - name: journal_no_same_sector_overwrites
   type: bool
   default: false
@@ -418,8 +402,6 @@
     journal after writing it instead of possibly overwriting it the second time.
 
     Most (99%) other SSDs don't need this option.
-
-    Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
   info_ru: |
     Включайте данную опцию для SSD вроде Intel D3-S4510 и D3-S4610, которые
     ОЧЕНЬ не любят, когда ПО перезаписывает один и тот же сектор несколько раз
@@ -430,20 +412,6 @@
     самого сектора.
 
     Почти все другие SSD (99% моделей) не требуют данной опции.
-
-    Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-- name: skip_corrupted_meta_entries
-  type: bool
-  default: false
-  info: |
-    Only for the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-    Allow OSD to start when some metadata entries or blocks are corrupted by
-    skipping them. Should be only used as an emergency measure.
-  info_ru: |
-    Только для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3).
-    Разрешить OSD запускаться, даже если часть блоков или записей метаданных
-    повреждена, пропуская их. Опция предназначена для использования только в
-    целях аварийного восстановления.
 - name: throttle_small_writes
   type: bool
   default: false
@@ -938,7 +906,3 @@
     This option sets the interval between handling two PG count change chunks.
   info_ru: |
     Данная опция задаёт интервал между обработкой двух порций изменения числа PG пулов.
-- name: gc_on_start
-  type: bool
-  info: Forcibly clean all garbage entries in the new store on every OSD restart.
-  info_ru: Принудительно очищать все мусорные записи в новом хранилище при каждом запуске OSD.

docs/installation/docker.en.md

@@ -26,37 +26,13 @@ at Vitastor Kubernetes operator: https://github.com/Antilles7227/vitastor-operat
 The instruction is very simple.
 
 1. Download a Docker image of the desired version: \
-   `docker pull vitalif/vitastor:v3.0.12`
+   `docker pull vitalif/vitastor:v3.0.3`
 2. Install scripts to the host system: \
-   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.12 install.sh`
+   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.3 install.sh`
 3. Reload udev rules: \
    `udevadm control --reload-rules`
-4. Enable the vitastor-host service: \
-   `systemctl enable --now vitastor-host`
 
-After these steps, you can return to [Quick Start](../intro/quickstart.en.md).
+And you can return to [Quick Start](../intro/quickstart.en.md).
-
-## Podman
-
-If you use Podman, run the following commands as root before installing Vitastor containers:
-
-```
-ln -s podman /usr/bin/docker
-
-mkdir -p /etc/systemd/system/systemd-udevd.service.d
-
-cat >/etc/systemd/system/systemd-udevd.service.d/override.conf <<EOF
-[Service]
-CapabilityBoundingSet=~
-SystemCallFilter=@mount capset
-EOF
-
-systemctl daemon-reload
-
-systemctl restart systemd-udevd
-```
-
-Without it, udev fails to do calls into a Podman container and Vitastor disk detection doesn't work.
 
 ## Upgrading Containers
 

docs/installation/docker.ru.md

@@ -25,39 +25,14 @@ Vitastor можно установить в Docker/Podman. При этом etcd,
 Инструкция по установке максимально простая.
 
 1. Скачайте Docker-образ желаемой версии: \
-   `docker pull vitalif/vitastor:v3.0.12`
+   `docker pull vitalif/vitastor:v3.0.3`
 2. Установите скрипты в хост-систему командой: \
-   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.12 install.sh`
+   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.3 install.sh`
 3. Перезагрузите правила udev: \
    `udevadm control --reload-rules`
-4. Включите сервис vitastor-host: \
-   `systemctl enable --now vitastor-host`
 
 После этого вы можете возвращаться к разделу [Быстрый старт](../intro/quickstart.ru.md).
 
-## Podman
-
-Если вы используете Podman, перед установкой контейнеров Vitastor выполните следующие
-команды от имени суперпользователя:
-
-```
-ln -s podman /usr/bin/docker
-
-mkdir -p /etc/systemd/system/systemd-udevd.service.d
-
-cat >/etc/systemd/system/systemd-udevd.service.d/override.conf <<EOF
-[Service]
-CapabilityBoundingSet=~
-SystemCallFilter=@mount capset
-EOF
-
-systemctl daemon-reload
-
-systemctl restart systemd-udevd
-```
-
-Без этих настроек udev не может делать вызовы внутрь Podman-контейнеров и определение дисков Vitastor не работает.
-
 ## Обновление контейнеров
 
 Сначала обязательно проверьте раздел [Обновление Vitastor](../usage/admin.ru.md#обновление-vitastor),

docs/installation/packages.en.md

@@ -17,7 +17,6 @@
   - Debian 10 (Buster): `deb https://vitastor.io/debian buster main`
   - Ubuntu 22.04 (Jammy): `deb https://vitastor.io/debian jammy main`
   - Ubuntu 24.04 (Noble): `deb https://vitastor.io/debian noble main`
-  - Ubuntu 26.04 (Resolute): `deb https://vitastor.io/debian resolute main`
   - Add `-oldstable` to bookworm/bullseye/buster in this line to install the last
     stable version from 0.9.x branch instead of 1.x
 - To always prefer vitastor-patched QEMU and Libvirt versions, add the following to `/etc/apt/preferences`:
@@ -34,17 +33,15 @@
   - CentOS 7: `yum install https://vitastor.io/rpms/centos/7/vitastor-release.rpm`
   - CentOS 8: `dnf install https://vitastor.io/rpms/centos/8/vitastor-release.rpm`
   - AlmaLinux 9 and other RHEL 9 clones (Rocky, Oracle...): `dnf install https://vitastor.io/rpms/centos/9/vitastor-release.rpm`
-  - AlmaLinux 10 and other RHEL 10 clones: `dnf install https://vitastor.io/rpms/centos/10/vitastor-release.rpm`
 - Enable EPEL: `yum/dnf install epel-release`
 - Enable additional CentOS repositories:
   - CentOS 7: `yum install centos-release-scl`
   - CentOS 8: `dnf install centos-release-advanced-virtualization`
-  - RHEL 9/10 clones: not required
+  - RHEL 9 clones: not required
 - Enable elrepo-kernel:
   - CentOS 7: `yum install https://www.elrepo.org/elrepo-release-7.el7.elrepo.noarch.rpm`
   - CentOS 8: `dnf install https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm`
   - RHEL 9 clones: `dnf install https://www.elrepo.org/elrepo-release-9.el9.elrepo.noarch.rpm`
-  - RHEL 10 clones: not required
 - Install packages: `yum/dnf install vitastor lpsolve etcd kernel-ml qemu-kvm`
 
 ## Installation requirements

docs/installation/packages.ru.md

@@ -17,7 +17,6 @@
   - Debian 10 (Buster): `deb https://vitastor.io/debian buster main`
   - Ubuntu 22.04 (Jammy): `deb https://vitastor.io/debian jammy main`
   - Ubuntu 24.04 (Noble): `deb https://vitastor.io/debian noble main`
-  - Ubuntu 26.04 (Resolute): `deb https://vitastor.io/debian resolute main`
   - Добавьте `-oldstable` к слову bookworm/bullseye/buster в этой строке, чтобы
     установить последнюю стабильную версию из ветки 0.9.x вместо 1.x
 - Чтобы всегда предпочитались версии пакетов QEMU и Libvirt с патчами Vitastor, добавьте в `/etc/apt/preferences`:
@@ -34,17 +33,15 @@
   - CentOS 7: `yum install https://vitastor.io/rpms/centos/7/vitastor-release.rpm`
   - CentOS 8: `dnf install https://vitastor.io/rpms/centos/8/vitastor-release.rpm`
   - AlmaLinux 9 и другие клоны RHEL 9 (Rocky, Oracle...): `dnf install https://vitastor.io/rpms/centos/9/vitastor-release.rpm`
-  - AlmaLinux 10 и другие клоны RHEL 10: `dnf install https://vitastor.io/rpms/centos/10/vitastor-release.rpm`
 - Включите EPEL: `yum/dnf install epel-release`
 - Включите дополнительные репозитории CentOS:
   - CentOS 7: `yum install centos-release-scl`
   - CentOS 8: `dnf install centos-release-advanced-virtualization`
-  - Клоны RHEL 9/10: не нужно
+  - Клоны RHEL 9: не нужно
 - Включите elrepo-kernel:
   - CentOS 7: `yum install https://www.elrepo.org/elrepo-release-7.el7.elrepo.noarch.rpm`
   - CentOS 8: `dnf install https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm`
   - Клоны RHEL 9: `dnf install https://www.elrepo.org/elrepo-release-9.el9.elrepo.noarch.rpm`
-  - Клоны RHEL 10: не нужно
 - Установите пакеты: `yum/dnf install vitastor lpsolve etcd kernel-ml qemu-kvm`
 
 ## Установочные требования

docs/installation/source.en.md

@@ -16,7 +16,8 @@
   designated initializers support from C++20
 - CMake
 - jerasure headers and libraries
-- ISA-L, libibverbs, librdmacm, libnl3 headers and libraries (optional)
+- ISA-L, libibverbs and librdmacm headers and libraries (optional)
+- tcmalloc (google-perftools-dev)
 
 ## Basic instructions
 

docs/installation/source.ru.md

@@ -16,7 +16,8 @@
   назначенных инициализаторов (designated initializers) из C++20
 - CMake
 - Заголовки и библиотеки jerasure
-- Опционально - заголовки и библиотеки ISA-L, libibverbs, librdmacm, libnl3
+- Опционально - заголовки и библиотеки ISA-L, libibverbs, librdmacm
+- tcmalloc (google-perftools-dev)
 
 ## Базовая инструкция
 

docs/usage/nfs.en.md

@@ -262,4 +262,3 @@ Options:
 | `--logfile <FILE>` | log to the specified file                                |
 | `--enforce 1`      | enforce permissions at the server side (no by default)   |
 | `--foreground 1`   | stay in foreground, do not daemonize                     |
-| `--trace`          | trace all NFS requests                                   |

docs/usage/nfs.ru.md

@@ -274,4 +274,3 @@ VitastorFS из GPUDirect.
 | `--logfile <FILE>` | записывать логи в заданный файл                         |
 | `--enforce 1`      | проверять права доступа на стороне сервера (по умолчанию нет) |
 | `--foreground 1`   | не уходить в фон после запуска                          |
-| `--trace`          | логгировать все запросы NFS                             |

mon/antietcd_adapter.js

@@ -18,7 +18,7 @@ class AntiEtcdAdapter
                 cluster = cluster ? (''+(cluster||'')).split(/,+/) : [];
             cluster = Object.keys(cluster.reduce((a, url) =>
             {
-                a[url.toLowerCase().replace(/^(https?:\/\/)/, '').replace(/\/.*$/, '')] = true;
+                a[url.toLowerCase().replace(/^(https?:\/\/)?(.*?)(\/.*)?$/, (m, m1, m2) => (m1||'http://')+m2)] = true;
                 return a;
             }, {}));
             const cfg_port = config.antietcd_port;
@@ -26,7 +26,8 @@ class AntiEtcdAdapter
             is_local['0.0.0.0'] = true;
             is_local['::'] = true;
             is_local[''] = true;
-            const selected = cluster.map(s => s.split(':', 2)).filter(ip => is_local[ip[0]] && (!cfg_port || ip[1] == cfg_port));
+            // split :, 3 -> <schema>:<//ip>:<port>
+            const selected = cluster.map(s => s.split(':', 3)).filter(ip => is_local[ip[1].substr(2)] && (!cfg_port || ip[2] == cfg_port));
             if (selected.length > 1)
             {
                 console.error('More than 1 etcd_address matches local IPs, please specify port');
@@ -35,12 +36,15 @@ class AntiEtcdAdapter
             else if (selected.length == 1)
             {
                 const antietcd_config = {
-                    ip: selected[0][0],
+                    ip: selected[0][1].substr(2),
-                    port: selected[0][1],
+                    port: selected[0][2],
-                    data: config.antietcd_data_file || ((config.antietcd_data_dir || '/var/lib/vitastor') + '/mon_'+selected[0][1]+'.json.gz'),
+                    cert: config.antietcd_cert,
+                    key: config.antietcd_key,
+                    ca: config.etcd_ca,
+                    data: config.antietcd_data_file || ((config.antietcd_data_dir || '/var/lib/vitastor') + '/mon_'+selected[0][2]+'.json.gz'),
                     persist_filter: vitastor_persist_filter({ vitastor_prefix: config.etcd_prefix || '/vitastor' }),
-                    node_id: selected[0][0]+':'+selected[0][1], // node_id = ip:port
+                    node_id: selected[0][1].substr(2)+':'+selected[0][2], // node_id = ip:port
-                    cluster: (cluster.length == 1 ? null : cluster.reduce((a, c) => { a[c] = "http://"+c; return a; }, {})),
+                    cluster: (cluster.length == 1 ? null : cluster.reduce((a, c) => { a[c.replace(/^(https?:\/\/)/, '')] = c; return a; }, {})),
                     cluster_key: (config.etcd_prefix || '/vitastor'),
                     stale_read: 1,
                     log_level: 1,

mon/etcd_adapter.js

@@ -1,7 +1,9 @@
 // Copyright (c) Vitaliy Filippov, 2019+
 // License: VNPL-1.1 (see README.md for details)
 
+const fs = require('fs');
 const http = require('http');
+const https = require('https');
 const WebSocket = require('ws');
 const { b64, local_ips } = require('./utils.js');
 
@@ -15,11 +17,18 @@ class EtcdAdapter
         this.ws = null;
         this.ws_alive = false;
         this.ws_keepalive_timer = null;
+        this.opts = {};
     }
 
     parse_config(config)
     {
         this.parse_etcd_addresses(config.etcd_address||config.etcd_url);
+        if (config.etcd_client_cert)
+            this.opts.cert = fs.readFileSync(config.etcd_client_cert, { encoding: 'utf-8' });
+        if (config.etcd_client_key)
+            this.opts.key = fs.readFileSync(config.etcd_client_key, { encoding: 'utf-8' });
+        if (config.etcd_ca)
+            this.opts.ca = fs.readFileSync(config.etcd_ca, { encoding: 'utf-8' });
     }
 
     parse_etcd_addresses(addrs)
@@ -39,7 +48,7 @@ class EtcdAdapter
         for (let url of addrs)
         {
             let scheme = 'http';
-            url = url.trim().replace(/^(https?):\/\//, (m, m1) => { scheme = m1; return ''; });
+            url = url.trim().replace(/^(https?):\/\//i, (m, m1) => { scheme = m1.toLowerCase(); return ''; });
             const slash = url.indexOf('/');
             const colon = url.indexOf(':');
             const is_local = is_local_ip[colon >= 0 ? url.substr(0, colon) : (slash >= 0 ? url.substr(0, slash) : url)];
@@ -130,7 +139,7 @@ class EtcdAdapter
                     }
                     ok(false);
                 }, this.mon.config.etcd_mon_timeout);
-                this.ws = new WebSocket(base+'/watch');
+                this.ws = new WebSocket(base+'/watch', this.opts);
                 this.ws_used_url = cur_addr;
                 const fail = () =>
                 {
@@ -272,7 +281,7 @@ class EtcdAdapter
             {
                 throw new Error(MON_STOPPED);
             }
-            const res = await POST(base+path, body, timeout);
+            const res = await POST(base+path, body, timeout, this.opts);
             if (this.mon.stopped)
             {
                 throw new Error(MON_STOPPED);
@@ -298,7 +307,7 @@ class EtcdAdapter
     }
 }
 
-function POST(url, body, timeout)
+function POST(url, body, timeout, opts)
 {
     return new Promise(ok =>
     {
@@ -310,10 +319,10 @@ function POST(url, body, timeout)
             req = null;
             ok({ error: 'timeout' });
         }, timeout) : null;
-        let req = http.request(url, { method: 'POST', headers: {
+        let req = (url.substr(0, 5) == 'https' ? https : http).request(url, { method: 'POST', headers: {
             'Content-Type': 'application/json',
             'Content-Length': body_text.length,
-        } }, (res) =>
+        }, ...(opts||{}) }, (res) =>
         {
             if (!req)
             {

mon/etcd_schema.js

@@ -45,7 +45,10 @@ const etcd_tree = {
             config_path: "/etc/vitastor/vitastor.conf",
             etcd_prefix: "/vitastor",
             // etcd connection - configurable online
-            etcd_address: "10.0.115.10:2379/v3",
+            etcd_address: "http://10.0.115.10:2379/v3",
+            etcd_client_cert: "",
+            etcd_client_key: "",
+            etcd_ca: "",
             // mon
             etcd_mon_ttl: 5, // min: 1
             etcd_mon_timeout: 1000, // ms. min: 0

mon/http_server.js

@@ -16,7 +16,7 @@ async function create_http_server(cfg, handler)
         };
         if (cfg.mon_https_ca)
         {
-            tls.ca = await fsp.readFile(cfg.mon_https_ca);
+            tls.mon_https_ca = await fsp.readFile(cfg.mon_https_ca);
         }
         if (cfg.mon_https_client_auth)
         {

mon/mon.js

@@ -627,7 +627,7 @@ class Mon
                 if (this.state.pg.history[pool_id] &&
                     this.state.pg.history[pool_id][pg])
                 {
-                    pg_history[pg-1] = JSON.parse(JSON.stringify(this.state.pg.history[pool_id][pg]));
+                    pg_history[pg-1] = this.state.pg.history[pool_id][pg];
                 }
             }
             const real_prev_pgs = [];
@@ -719,7 +719,7 @@ class Mon
                 this.next_recheck_timer = null;
                 this.next_recheck_at = 0;
                 this.schedule_recheck();
-            }, (this.next_recheck_at-now)*1000);
+            }, now-this.next_recheck_at);
         }
     }
 

mon/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vitastor-mon",
-  "version": "3.0.12",
+  "version": "3.0.3",
   "description": "Vitastor SDS monitor service",
   "main": "mon-main.js",
   "scripts": {

mon/pg_utils.js

@@ -84,7 +84,7 @@ function scale_pg_history(prev_pg_history, prev_pgs, new_pgs)
         finish_pg_history(merged_history[1]);
         for (let i = 0; i < new_pg_count; i++)
         {
-            new_pg_history[i] = JSON.parse(JSON.stringify(merged_history[1]));
+            new_pg_history[i] = { ...merged_history[1] };
         }
     }
     // Mark history keys for removed PGs as removed
@@ -102,7 +102,7 @@ function scale_pg_count(prev_pgs, new_pg_count)
     {
         for (let i = prev_pgs.length; i < new_pg_count; i++)
         {
-            prev_pgs[i] = [ ...prev_pgs[i % prev_pgs.length] ];
+            prev_pgs[i] = prev_pgs[i % prev_pgs.length];
         }
     }
     else if (prev_pgs.length > new_pg_count)

mon/stats.js

@@ -37,7 +37,6 @@ function derive_osd_stats(st, prev, prev_diff)
         const n = c.count - BigInt(pr && pr.count||0);
         diff.recovery_stats[op] = { ...c, bps: n > 0 ? b*1000n/timediff : 0n, iops: n > 0 ? n*1000n/timediff : 0n };
     }
-    diff.inode_stats = {};
     for (const pool_id in st.inode_stats||{})
     {
         diff.inode_stats[pool_id] = {};

node-binding/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vitastor",
-  "version": "3.0.12",
+  "version": "3.0.3",
   "description": "Low-level native bindings to Vitastor client library",
   "main": "index.js",
   "keywords": [

patches/cinder-vitastor.py

@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils
 
-VITASTOR_VERSION = '3.0.12'
+VITASTOR_VERSION = '3.0.3'
 
 LOG = logging.getLogger(__name__)
 
@@ -275,7 +275,7 @@ class VitastorDriver(driver.CloneableImageVD,
             LOG.exception('error getting vitastor pool stats: '+str(e))
 
         self._stats = stats
-
+        
     def get_volume_stats(self, refresh=False):
         """Get volume stats.
         If 'refresh' is True, run update the stats first.
@@ -291,14 +291,6 @@ class VitastorDriver(driver.CloneableImageVD,
         else:
             return (1 + resp['kvs'][0]['value'], resp['kvs'][0]['mod_revision'])
 
-    def _cli(self, descr, *args):
-        args = [ 'vitastor-cli', *args, *(self._vitastor_args()) ]
-        try:
-            self._execute(*args)
-        except processutils.ProcessExecutionError as exc:
-            LOG.error("Failed to "+descr+": "+exc)
-            raise exception.VolumeBackendAPIException(data = exc.stderr)
-
     def create_volume(self, volume):
         """Creates a logical volume."""
 
@@ -310,7 +302,7 @@ class VitastorDriver(driver.CloneableImageVD,
 
         LOG.debug("creating volume '%s'", vol_name)
 
-        self._cli('create volume', 'create', vol_name, '--size', size)
+        self._create_image(vol_name, { 'size': size })
 
         if volume.encryption_key_id:
             self._create_encrypted_volume(volume, volume.obj_context)
@@ -354,7 +346,7 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = utils.convert_str(snapshot.name)
         if snap_name.find('@') >= 0 or snap_name.find('/') >= 0:
             raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
-        self._cli('create snapshot', 'snap-create', vol_name+'@'+snap_name)
+        self._create_snapshot(vol_name, vol_name+'@'+snap_name)
 
     def snapshot_revert_use_temp_snapshot(self):
         """Disable the use of a temporary snapshot on revert."""
@@ -367,8 +359,21 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = utils.convert_str(snapshot.name)
 
         # Delete the image and recreate it from the snapshot
-        self._cli('delete image', 'rm', vol_name)
+        args = [ 'vitastor-cli', 'rm', vol_name, *(self._vitastor_args()) ]
-        self._cli('recreate image', 'create', '--parent', vol_name+'@'+snap_name, vol_name)
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to delete image "+vol_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
+        args = [
+            'vitastor-cli', 'create', '--parent', vol_name+'@'+snap_name,
+            vol_name, *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to recreate image "+vol_name+" from "+vol_name+"@"+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
 
     def delete_snapshot(self, snapshot):
         """Deletes a snapshot."""
@@ -376,7 +381,15 @@ class VitastorDriver(driver.CloneableImageVD,
         vol_name = utils.convert_str(snapshot.volume_name)
         snap_name = utils.convert_str(snapshot.name)
 
-        self._cli('remove snapshot', 'rm', vol_name+'@'+snap_name)
+        args = [
+            'vitastor-cli', 'rm', vol_name+'@'+snap_name,
+            *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to remove snapshot "+vol_name+'@'+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
 
     def _child_count(self, parents):
         children = 0
@@ -414,7 +427,13 @@ class VitastorDriver(driver.CloneableImageVD,
         if src_vref.admin_metadata.get('readonly') == 'True':
             # source volume is a volume-image cache entry or other readonly volume
             # clone without intermediate snapshot
-            self._cli('create clone', 'create', '--parent', src_name, '--size', size, dest_name)
+            src = self._get_image(src_name)
+            LOG.debug("creating image '%s' from '%s'", dest_name, src_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': src['idx']['id'],
+                'parent_pool_id': src['idx']['pool_id'],
+            })
             return {}
 
         clone_snap = "%s@%s.clone_snap" % (src_name, dest_name)
@@ -427,12 +446,15 @@ class VitastorDriver(driver.CloneableImageVD,
             clone_snap = dest_name
             make_img = False
 
-        LOG.debug("creating snapshot '%s'", clone_snap)
+        LOG.debug("creating layer '%s' under '%s'", clone_snap, src_name)
-        self._cli('create base snapshot', 'snap-create', '--allow-existing', '1', clone_snap)
+        new_cfg = self._create_snapshot(src_name, clone_snap, True)
-
         if make_img:
             # Then create a clone from it
-            self._cli('create clone', 'create', '--parent', clone_snap, '--size', size, dest_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': new_cfg['parent_id'],
+                'parent_pool_id': new_cfg['parent_pool_id'],
+            })
 
         return {}
 
@@ -442,8 +464,7 @@ class VitastorDriver(driver.CloneableImageVD,
         vol_name = utils.convert_str(volume.name)
         snap_name = utils.convert_str(snapshot.name)
 
-        src_snap = 'volume-'+snapshot.volume_id+'@'+snap_name
+        snap = self._get_image('volume-'+snapshot.volume_id+'@'+snap_name)
-        snap = self._get_image(src_snap)
         if not snap:
             raise exception.SnapshotNotFound(snapshot_id = snap_name)
         snap_inode_id = int(resp['responses'][0]['kvs'][0]['value']['id'])
@@ -452,8 +473,12 @@ class VitastorDriver(driver.CloneableImageVD,
         size = snap['cfg']['size']
         if int(volume.size):
             size = int(volume.size) * units.Gi
+        new_cfg = self._create_image(vol_name, {
+            'size': size,
+            'parent_id': snap['idx']['id'],
+            'parent_pool_id': snap['idx']['pool_id'],
+        })
 
-        self._cli('create clone', 'create', vol_name, '--size', size, '--parent', src_snap)
         return {}
 
     def _vitastor_args(self):
@@ -480,7 +505,49 @@ class VitastorDriver(driver.CloneableImageVD,
         """Deletes a logical volume."""
 
         vol_name = utils.convert_str(volume.name)
-        self._cli('delete volume', 'rm', '--matching', vol_name, vol_name+'@*', '--progress', '0')
+
+        # Find the volume and all its snapshots
+        range_end = b'index/image/' + vol_name.encode('utf-8')
+        range_end = range_end[0 : len(range_end)-1] + six.int2byte(range_end[len(range_end)-1] + 1)
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name, 'range_end': range_end } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            # already deleted
+            LOG.info("volume %s no longer exists in backend", vol_name)
+            return
+        layers = resp['responses'][0]['kvs']
+        layer_ids = {}
+        for kv in layers:
+            inode_id = int(kv['value']['id'])
+            pool_id = int(kv['value']['pool_id'])
+            inode_pool_id = (pool_id << 48) | (inode_id & 0xffffffffffff)
+            layer_ids[inode_pool_id] = True
+
+        # Check if the volume has clones and raise 'busy' if so
+        children = self._child_count(layer_ids)
+        if children > 0:
+            raise exception.VolumeIsBusy(volume_name = vol_name)
+
+        # Clear data
+        for kv in layers:
+            args = [
+                'vitastor-cli', 'rm-data', '--pool', str(kv['value']['pool_id']),
+                '--inode', str(kv['value']['id']), '--progress', '0',
+                *(self._vitastor_args())
+            ]
+            try:
+                self._execute(*args)
+            except processutils.ProcessExecutionError as exc:
+                LOG.error("Failed to remove layer "+kv['key']+": "+exc)
+                raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+        # Delete all layers from etcd
+        requests = []
+        for kv in layers:
+            requests.append({ 'request_delete_range': { 'key': kv['key'] } })
+            requests.append({ 'request_delete_range': { 'key': 'config/inode/'+str(kv['value']['pool_id'])+'/'+str(kv['value']['id']) } })
+        self._etcd_txn({ 'success': requests })
 
     def retype(self, context, volume, new_type, diff, host):
         """Change extra type specifications for a volume."""
@@ -500,6 +567,98 @@ class VitastorDriver(driver.CloneableImageVD,
         """Removes an export for a logical volume."""
         pass
 
+    def _create_image(self, vol_name, cfg):
+        pool_s = str(self.cfg['pool_id'])
+        image_id = 0
+        while image_id == 0:
+            # check if the image already exists and find a free ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/maxid/'+pool_s } },
+            ] })
+            if len(resp['responses'][0]['kvs']) > 0:
+                # already exists
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' already exists')
+            image_id, id_mod = self._next_id(resp['responses'][1])
+            # try to create the image
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+pool_s },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+pool_s+'/'+str(image_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+pool_s, 'value': image_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': image_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+pool_s+'/'+str(image_id), 'value': json.dumps({
+                    **cfg, 'name': vol_name,
+                }) } },
+            ] })
+            if not resp.get('succeeded'):
+                # repeat
+                image_id = 0
+
+    def _create_snapshot(self, vol_name, snap_vol_name, allow_existing = False):
+        while True:
+            # check if the image already exists and snapshot doesn't
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/image/'+snap_vol_name } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            if len(resp['responses'][1]['kvs']) > 0:
+                if allow_existing:
+                    snap_idx = resp['responses'][1]['kvs'][0]['value']
+                    resp = self._etcd_txn({ 'success': [
+                        { 'request_range': { 'key': 'config/inode/'+str(snap_idx['pool_id'])+'/'+str(snap_idx['id']) } },
+                    ] })
+                    if len(resp['responses'][0]['kvs']) == 0:
+                        raise exception.VolumeBackendAPIException(data =
+                            'Volume '+snap_vol_name+' is already indexed, but does not exist'
+                        )
+                    return resp['responses'][0]['kvs'][0]['value']
+                raise exception.VolumeBackendAPIException(
+                    data = 'Volume '+snap_vol_name+' already exists'
+                )
+            vol_idx = resp['responses'][0]['kvs'][0]['value']
+            vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            # get image inode config and find a new ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+                { 'request_range': { 'key': 'index/maxid/'+str(self.cfg['pool_id']) } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            vol_cfg = resp['responses'][0]['kvs'][0]['value']
+            vol_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            new_id, id_mod = self._next_id(resp['responses'][1])
+            # try to redirect image to the new inode
+            new_cfg = {
+                **vol_cfg, 'name': vol_name, 'parent_id': vol_idx['id'], 'parent_pool_id': vol_idx['pool_id']
+            }
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol_idx_mod, 'key': 'index/image/'+vol_name },
+                { 'target': 'MOD', 'mod_revision': vol_mod, 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) },
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+str(self.cfg['pool_id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+snap_vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+str(self.cfg['pool_id']), 'value': new_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': new_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id), 'value': json.dumps(new_cfg) } },
+                { 'request_put': { 'key': 'index/image/'+snap_vol_name, 'value': json.dumps({
+                    'id': vol_idx['id'], 'pool_id': vol_idx['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']), 'value': json.dumps({
+                    **vol_cfg, 'name': snap_vol_name, 'readonly': True
+                }) } }
+            ] })
+            if resp.get('succeeded'):
+                return new_cfg
+
     def initialize_connection(self, volume, connector):
         data = {
             'driver_volume_type': 'vitastor',
@@ -538,9 +697,13 @@ class VitastorDriver(driver.CloneableImageVD,
                     size = int(volume.size) * units.Gi
                     dest_name = utils.convert_str(volume.name)
                     # Find or create the base snapshot
-                    self._cli('create base snapshot', 'create', '--allow-existing', '1', base_vol.name+'@.clone_snap')
+                    snap_cfg = self._create_snapshot(base_vol.name, base_vol.name+'@.clone_snap', True)
                     # Then create a clone from it
-                    self._cli('create clone', 'create', dest_name, '--size', size, '--parent', base_vol.name+'@.clone_snap')
+                    new_cfg = self._create_image(dest_name, {
+                        'size': size,
+                        'parent_id': snap_cfg['parent_id'],
+                        'parent_pool_id': snap_cfg['parent_pool_id'],
+                    })
                     return ({}, True)
         return ({}, False)
 
@@ -607,8 +770,26 @@ class VitastorDriver(driver.CloneableImageVD,
     def extend_volume(self, volume, new_size):
         """Extend an existing volume."""
         vol_name = utils.convert_str(volume.name)
-        size = int(new_size) * units.Gi
+        while True:
-        self._cli('extend volume', 'modify', vol_name, '--resize', new_size)
+            vol = self._get_image(vol_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            # change size
+            size = int(new_size) * units.Gi
+            if size == vol['cfg']['size']:
+                break
+            resp = self._etcd_txn({ 'compare': [ {
+                'target': 'MOD',
+                'mod_revision': vol['cfg_mod'],
+                'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+            } ], 'success': [
+                { 'request_put': {
+                    'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'size': size }),
+                } },
+            ] })
+            if resp.get('succeeded'):
+                break
         LOG.debug(
             "Extend volume from %(old_size)s GB to %(new_size)s GB.",
             {'old_size': volume.size, 'new_size': new_size}
@@ -681,7 +862,28 @@ class VitastorDriver(driver.CloneableImageVD,
         """
         from_name = self._get_existing_name(existing_ref)
         to_name = utils.convert_str(volume.name)
-        self._cli('rename', 'modify', from_name, '--rename', to_name)
+        self._rename(from_name, to_name)
+
+    def _rename(self, from_name, to_name):
+        while True:
+            vol = self._get_image(from_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+from_name+' does not exist')
+            to = self._get_image(to_name)
+            if to:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+to_name+' already exists')
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol['idx_mod'], 'key': 'index/image/'+vol['cfg']['name'] },
+                { 'target': 'MOD', 'mod_revision': vol['cfg_mod'], 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+to_name },
+            ], 'success': [
+                { 'request_delete_range': { 'key': 'index/image/'+vol['cfg']['name'] } },
+                { 'request_put': { 'key': 'index/image/'+to_name, 'value': json.dumps(vol['idx']) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'name': to_name }) } },
+            ] })
+            if resp.get('succeeded'):
+                break
 
     def unmanage(self, volume):
         pass
@@ -754,7 +956,7 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = self._get_existing_name(existing_ref)
         from_name = vol_name+'@'+snap_name
         to_name = vol_name+'@'+utils.convert_str(snapshot.name)
-        self._cli('rename', 'modify', from_name, '--rename', to_name)
+        self._rename(from_name, to_name)
 
     def unmanage_snapshot(self, snapshot):
         """Removes the specified snapshot from Cinder management."""

patches/libvirt-12.0-vitastor.diff

@@ -1,637 +0,0 @@
-diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
-index aaad4a3da1..5f5daa8341 100644
---- a/include/libvirt/libvirt-storage.h
-+++ b/include/libvirt/libvirt-storage.h
-@@ -326,6 +326,7 @@ typedef enum {
-     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17, /* (Since: 1.2.8) */
-     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18, /* (Since: 3.1.0) */
-     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19, /* (Since: 5.6.0) */
-+    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20, /* (Since: 5.0.0) */
- } virConnectListAllStoragePoolsFlags;
- 
- int                     virConnectListAllStoragePools(virConnectPtr conn,
-diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
-index 9ca5c2450c..cc52f00c0c 100644
---- a/src/conf/domain_conf.c
-+++ b/src/conf/domain_conf.c
-@@ -7453,7 +7453,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
-     src->configFile = virXPathString("string(./config/@file)", ctxt);
- 
-     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
--        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
-+        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
-+        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
-         src->query = virXMLPropString(node, "query");
- 
-     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
-@@ -32187,6 +32188,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSource *src,
- 
-     case VIR_STORAGE_POOL_MPATH:
-     case VIR_STORAGE_POOL_RBD:
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_SHEEPDOG:
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_LAST:
-diff --git a/src/conf/domain_validate.c b/src/conf/domain_validate.c
-index 7346a61731..83e94d762e 100644
---- a/src/conf/domain_validate.c
-+++ b/src/conf/domain_validate.c
-@@ -520,6 +520,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-@@ -592,7 +593,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-         }
-     }
- 
--    /* internal snapshots and config files are currently supported only with rbd: */
-+    /* internal snapshots are currently supported only with rbd: */
-     if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
-         src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD) {
-         if (src->snapshot) {
-@@ -600,10 +601,14 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-                            _("<snapshot> element is currently supported only with 'rbd' disks"));
-             return -1;
-         }
--
-+    }
-+    /* config files are currently supported only with rbd and vitastor: */
-+    if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
-+        src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD &&
-+        src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR) {
-         if (src->configFile) {
-             virReportError(VIR_ERR_XML_ERROR, "%s",
--                           _("<config> element is currently supported only with 'rbd' disks"));
-+                           _("<config> element is currently supported only with 'rbd' and 'vitastor' disks"));
-             return -1;
-         }
-     }
-diff --git a/src/conf/schemas/domaincommon.rng b/src/conf/schemas/domaincommon.rng
-index 114dd3f96f..c71f9a3277 100644
---- a/src/conf/schemas/domaincommon.rng
-+++ b/src/conf/schemas/domaincommon.rng
-@@ -2093,6 +2093,35 @@
-     </element>
-   </define>
- 
-+  <define name="diskSourceNetworkProtocolVitastor">
-+    <element name="source">
-+      <interleave>
-+        <attribute name="protocol">
-+          <value>vitastor</value>
-+        </attribute>
-+        <ref name="diskSourceCommon"/>
-+        <optional>
-+          <attribute name="name"/>
-+        </optional>
-+        <optional>
-+          <attribute name="query"/>
-+        </optional>
-+        <zeroOrMore>
-+          <ref name="diskSourceNetworkHost"/>
-+        </zeroOrMore>
-+        <optional>
-+          <element name="config">
-+            <attribute name="file">
-+              <ref name="absFilePath"/>
-+            </attribute>
-+            <empty/>
-+          </element>
-+        </optional>
-+        <empty/>
-+      </interleave>
-+    </element>
-+  </define>
-+
-   <define name="diskSourceNetworkProtocolISCSI">
-     <element name="source">
-       <attribute name="protocol">
-@@ -2443,6 +2472,7 @@
-       <ref name="diskSourceNetworkProtocolSimple"/>
-       <ref name="diskSourceNetworkProtocolVxHS"/>
-       <ref name="diskSourceNetworkProtocolNFS"/>
-+      <ref name="diskSourceNetworkProtocolVitastor"/>
-     </choice>
-   </define>
- 
-diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
-index 1dc9365bf2..a8a736be81 100644
---- a/src/conf/storage_conf.c
-+++ b/src/conf/storage_conf.c
-@@ -56,7 +56,7 @@ VIR_ENUM_IMPL(virStoragePool,
-               "logical", "disk", "iscsi",
-               "iscsi-direct", "scsi", "mpath",
-               "rbd", "sheepdog", "gluster",
--              "zfs", "vstorage",
-+              "zfs", "vstorage", "vitastor",
- );
- 
- VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
-@@ -242,6 +242,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
-           .formatToString = virStorageFileFormatTypeToString,
-       }
-     },
-+    {.poolType = VIR_STORAGE_POOL_VITASTOR,
-+     .poolOptions = {
-+         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
-+                   VIR_STORAGE_POOL_SOURCE_NETWORK |
-+                   VIR_STORAGE_POOL_SOURCE_NAME),
-+      },
-+      .volOptions = {
-+          .defaultFormat = VIR_STORAGE_FILE_RAW,
-+          .formatFromString = virStorageVolumeFormatFromString,
-+          .formatToString = virStorageFileFormatTypeToString,
-+      }
-+    },
-     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
-      .poolOptions = {
-          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
-@@ -538,6 +550,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
-                        _("element 'name' is mandatory for RBD pool"));
-         return -1;
-     }
-+    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
-+        virReportError(VIR_ERR_XML_ERROR, "%s",
-+                       _("element 'name' is mandatory for Vitastor pool"));
-+        return -1;
-+    }
- 
-     if (options->formatFromString) {
-         g_autofree char *format = NULL;
-@@ -1127,6 +1144,7 @@ virStoragePoolDefFormatBuf(virBuffer *buf,
-     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
-      * files, so they don't have a target */
-     if (def->type != VIR_STORAGE_POOL_RBD &&
-+        def->type != VIR_STORAGE_POOL_VITASTOR &&
-         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
-         def->type != VIR_STORAGE_POOL_GLUSTER &&
-         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
-diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
-index fc67957cfe..720c07ef74 100644
---- a/src/conf/storage_conf.h
-+++ b/src/conf/storage_conf.h
-@@ -103,6 +103,7 @@ typedef enum {
-     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
-     VIR_STORAGE_POOL_ZFS,      /* ZFS */
-     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
-+    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
- 
-     VIR_STORAGE_POOL_LAST,
- } virStoragePoolType;
-@@ -454,6 +455,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
-                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
-+                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
-diff --git a/src/conf/storage_source_conf.c b/src/conf/storage_source_conf.c
-index d7b9bdfecb..38aefd0dd4 100644
---- a/src/conf/storage_source_conf.c
-+++ b/src/conf/storage_source_conf.c
-@@ -90,6 +90,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
-               "ssh",
-               "vxhs",
-               "nfs",
-+              "vitastor",
- );
- 
- 
-@@ -1317,6 +1318,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-             return 24007;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-             /* we don't provide a default for RBD */
-             return 0;
-diff --git a/src/conf/storage_source_conf.h b/src/conf/storage_source_conf.h
-index 22c35d420d..f1e32ea83d 100644
---- a/src/conf/storage_source_conf.h
-+++ b/src/conf/storage_source_conf.h
-@@ -131,6 +131,7 @@ typedef enum {
-     VIR_STORAGE_NET_PROTOCOL_SSH,
-     VIR_STORAGE_NET_PROTOCOL_VXHS,
-     VIR_STORAGE_NET_PROTOCOL_NFS,
-+    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
- 
-     VIR_STORAGE_NET_PROTOCOL_LAST
- } virStorageNetProtocol;
-diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
-index 59fa5da372..4739167f5f 100644
---- a/src/conf/virstorageobj.c
-+++ b/src/conf/virstorageobj.c
-@@ -1438,6 +1438,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
-             return 1;
-         break;
- 
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_ISCSI_DIRECT:
-     case VIR_STORAGE_POOL_RBD:
-     case VIR_STORAGE_POOL_LAST:
-@@ -1921,6 +1922,8 @@ virStoragePoolObjMatch(virStoragePoolObj *obj,
-                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
-                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
-+              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
-+               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
-                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
-diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
-index db7660aac4..561df34709 100644
---- a/src/libvirt-storage.c
-+++ b/src/libvirt-storage.c
-@@ -94,6 +94,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
-  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
-  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
-  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
-+ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
-  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
-  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
-  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
-diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
-index 2b988157fa..9d0eb47b25 100644
---- a/src/libxl/libxl_conf.c
-+++ b/src/libxl/libxl_conf.c
-@@ -1069,6 +1069,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSource *src,
-     case VIR_STORAGE_NET_PROTOCOL_SSH:
-     case VIR_STORAGE_NET_PROTOCOL_VXHS:
-     case VIR_STORAGE_NET_PROTOCOL_NFS:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-         virReportError(VIR_ERR_NO_SUPPORT,
-diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
-index e72e7d7f44..8482c21805 100644
---- a/src/libxl/xen_xl.c
-+++ b/src/libxl/xen_xl.c
-@@ -1461,6 +1461,7 @@ xenFormatXLDiskSrcNet(virStorageSource *src)
-     case VIR_STORAGE_NET_PROTOCOL_SSH:
-     case VIR_STORAGE_NET_PROTOCOL_VXHS:
-     case VIR_STORAGE_NET_PROTOCOL_NFS:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-         virReportError(VIR_ERR_NO_SUPPORT,
-diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
-index 9b43279797..459d8e8a65 100644
---- a/src/qemu/qemu_block.c
-+++ b/src/qemu/qemu_block.c
-@@ -743,6 +743,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSource *src,
- }
- 
- 
-+static virJSONValue *
-+qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
-+{
-+    virJSONValue *ret = NULL;
-+    virStorageNetHostDef *host;
-+    size_t i;
-+    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
-+    g_autofree char *etcd = NULL;
-+
-+    for (i = 0; i < src->nhosts; i++) {
-+        host = src->hosts + i;
-+        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
-+            return NULL;
-+        }
-+        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
-+    }
-+    if (src->nhosts > 0) {
-+        etcd = virBufferContentAndReset(&buf);
-+    }
-+
-+    if (virJSONValueObjectAdd(&ret,
-+                              "S:etcd-host", etcd,
-+                              "S:etcd-prefix", src->query,
-+                              "S:config-path", src->configFile,
-+                              "s:image", src->path,
-+                              NULL) < 0)
-+        return NULL;
-+
-+    return ret;
-+}
-+
-+
- static virJSONValue *
- qemuBlockStorageSourceGetSshProps(virStorageSource *src)
- {
-@@ -1094,6 +1126,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSource *src,
-                 return NULL;
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+            driver = "vitastor";
-+            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
-+                return NULL;
-+            break;
-+
-         case VIR_STORAGE_NET_PROTOCOL_SSH:
-             driver = "ssh";
-             if (!(fileprops = qemuBlockStorageSourceGetSshProps(src)))
-@@ -1997,6 +2035,7 @@ qemuBlockGetBackingStoreString(virStorageSource *src,
- 
-             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-             case VIR_STORAGE_NET_PROTOCOL_RBD:
-+            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-             case VIR_STORAGE_NET_PROTOCOL_VXHS:
-             case VIR_STORAGE_NET_PROTOCOL_NFS:
-             case VIR_STORAGE_NET_PROTOCOL_SSH:
-@@ -2377,6 +2416,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSource *src,
-                 return -1;
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+            driver = "vitastor";
-+            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
-+                return -1;
-+            break;
-+
-         case VIR_STORAGE_NET_PROTOCOL_SSH:
-             if (srcPriv->nbdkitProcess) {
-                 /* disk creation not yet supported with nbdkit, and even if it
-diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
-index ac56fc7cb4..9e407b4aab 100644
---- a/src/qemu/qemu_domain.c
-+++ b/src/qemu/qemu_domain.c
-@@ -4677,7 +4677,8 @@ qemuDomainValidateStorageSource(virStorageSource *src,
-     if (src->query &&
-         (actualType != VIR_STORAGE_TYPE_NETWORK ||
-          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
--          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
-+          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
-+          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
-         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
-                        _("query is supported only with HTTP(S) protocols"));
-         return -1;
-@@ -9103,6 +9104,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSource *src,
-         break;
- 
-     case VIR_STORAGE_NET_PROTOCOL_RBD:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
-index e738afffc3..37d64f469b 100644
---- a/src/qemu/qemu_snapshot.c
-+++ b/src/qemu/qemu_snapshot.c
-@@ -665,6 +665,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDef *snapdisk,
-         case VIR_STORAGE_NET_PROTOCOL_NONE:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-@@ -893,6 +894,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDef *disk,
-         case VIR_STORAGE_NET_PROTOCOL_NONE:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
-index e19e032427..59f91f4710 100644
---- a/src/storage/storage_driver.c
-+++ b/src/storage/storage_driver.c
-@@ -1626,6 +1626,7 @@ storageVolLookupByPathCallback(virStoragePoolObj *obj,
- 
-         case VIR_STORAGE_POOL_GLUSTER:
-         case VIR_STORAGE_POOL_RBD:
-+        case VIR_STORAGE_POOL_VITASTOR:
-         case VIR_STORAGE_POOL_SHEEPDOG:
-         case VIR_STORAGE_POOL_ZFS:
-         case VIR_STORAGE_POOL_LAST:
-diff --git a/src/storage_file/storage_source_backingstore.c b/src/storage_file/storage_source_backingstore.c
-index 821378883c..2211f6891b 100644
---- a/src/storage_file/storage_source_backingstore.c
-+++ b/src/storage_file/storage_source_backingstore.c
-@@ -264,6 +264,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
- }
- 
- 
-+static int
-+virStorageSourceParseVitastorColonString(const char *colonstr,
-+                                         virStorageSource *src)
-+{
-+    char *p, *e, *next;
-+    g_autofree char *options = NULL;
-+
-+    /* optionally skip the "vitastor:" prefix if provided */
-+    if (STRPREFIX(colonstr, "vitastor:"))
-+        colonstr += strlen("vitastor:");
-+
-+    options = g_strdup(colonstr);
-+
-+    p = options;
-+    while (*p) {
-+        /* find : delimiter or end of string */
-+        for (e = p; *e && *e != ':'; ++e) {
-+            if (*e == '\\') {
-+                e++;
-+                if (*e == '\0')
-+                    break;
-+            }
-+        }
-+        if (*e == '\0') {
-+            next = e;    /* last kv pair */
-+        } else {
-+            next = e + 1;
-+            *e = '\0';
-+        }
-+
-+        if (STRPREFIX(p, "image=")) {
-+            src->path = g_strdup(p + strlen("image="));
-+        } else if (STRPREFIX(p, "etcd-prefix=")) {
-+            src->query = g_strdup(p + strlen("etcd-prefix="));
-+        } else if (STRPREFIX(p, "config-path=")) {
-+            src->configFile = g_strdup(p + strlen("config-path="));
-+        } else if (STRPREFIX(p, "etcd-host=")) {
-+            char *h, *sep;
-+
-+            h = p + strlen("etcd-host=");
-+            while (h < e) {
-+                for (sep = h; sep < e; ++sep) {
-+                    if (*sep == '\\' && (sep[1] == ',' ||
-+                                         sep[1] == ';' ||
-+                                         sep[1] == ' ')) {
-+                        *sep = '\0';
-+                        sep += 2;
-+                        break;
-+                    }
-+                }
-+
-+                if (virStorageSourceRBDAddHost(src, h) < 0)
-+                    return -1;
-+
-+                h = sep;
-+            }
-+        }
-+
-+        p = next;
-+    }
-+
-+    if (!src->path) {
-+        return -1;
-+    }
-+
-+    return 0;
-+}
-+
-+
- static int
- virStorageSourceParseNBDColonString(const char *nbdstr,
-                                     virStorageSource *src)
-@@ -379,6 +448,11 @@ virStorageSourceParseBackingColon(virStorageSource *src,
-             return -1;
-         break;
- 
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+        if (virStorageSourceParseVitastorColonString(path, src) < 0)
-+            return -1;
-+        break;
-+
-     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-@@ -953,6 +1027,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSource *src,
-     return 0;
- }
- 
-+static int
-+virStorageSourceParseBackingJSONVitastor(virStorageSource *src,
-+                                         virJSONValue *json,
-+                                         const char *jsonstr G_GNUC_UNUSED,
-+                                         int opaque G_GNUC_UNUSED)
-+{
-+    const char *filename;
-+    const char *image = virJSONValueObjectGetString(json, "image");
-+    const char *conf = virJSONValueObjectGetString(json, "config-path");
-+    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd-prefix");
-+    virJSONValue *servers = virJSONValueObjectGetArray(json, "server");
-+    size_t nservers;
-+    size_t i;
-+
-+    src->type = VIR_STORAGE_TYPE_NETWORK;
-+    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
-+
-+    /* legacy syntax passed via 'filename' option */
-+    if ((filename = virJSONValueObjectGetString(json, "filename")))
-+        return virStorageSourceParseVitastorColonString(filename, src);
-+
-+    if (!image) {
-+        virReportError(VIR_ERR_INVALID_ARG, "%s",
-+                       _("missing image name in Vitastor backing volume "
-+                         "JSON specification"));
-+        return -1;
-+    }
-+
-+    src->path = g_strdup(image);
-+    src->configFile = g_strdup(conf);
-+    src->query = g_strdup(etcd_prefix);
-+
-+    if (servers) {
-+        nservers = virJSONValueArraySize(servers);
-+
-+        src->hosts = g_new0(virStorageNetHostDef, nservers);
-+        src->nhosts = nservers;
-+
-+        for (i = 0; i < nservers; i++) {
-+            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
-+                                                                  virJSONValueArrayGet(servers, i)) < 0)
-+                return -1;
-+        }
-+    }
-+
-+    return 0;
-+}
-+
- static int
- virStorageSourceParseBackingJSONRaw(virStorageSource *src,
-                                     virJSONValue *json,
-@@ -1130,6 +1252,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
-     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
-     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
-     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
-+    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
-     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
-     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
-     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
-diff --git a/src/test/test_driver.c b/src/test/test_driver.c
-index 1165689de7..bba846351c 100644
---- a/src/test/test_driver.c
-+++ b/src/test/test_driver.c
-@@ -7345,6 +7345,7 @@ testStorageVolumeTypeForPool(int pooltype)
-     case VIR_STORAGE_POOL_ISCSI_DIRECT:
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_RBD:
-+    case VIR_STORAGE_POOL_VITASTOR:
-         return VIR_STORAGE_VOL_NETWORK;
-     case VIR_STORAGE_POOL_LOGICAL:
-     case VIR_STORAGE_POOL_DISK:
-diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-index eee75af746..8bd0a57bdd 100644
---- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-+++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-@@ -204,4 +204,11 @@
-       </enum>
-     </volOptions>
-   </pool>
-+  <pool type='vitastor' supported='no'>
-+    <volOptions>
-+      <defaultFormat type='raw'/>
-+      <enum name='targetFormatType'>
-+      </enum>
-+    </volOptions>
-+  </pool>
- </storagepoolCapabilities>
-diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
-index 805950a937..852df0de16 100644
---- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
-+++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
-@@ -204,4 +204,11 @@
-       </enum>
-     </volOptions>
-   </pool>
-+  <pool type='vitastor' supported='yes'>
-+    <volOptions>
-+      <defaultFormat type='raw'/>
-+      <enum name='targetFormatType'>
-+      </enum>
-+    </volOptions>
-+  </pool>
- </storagepoolCapabilities>
-diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
-index d5c2531ab8..b19308ac38 100644
---- a/tests/storagepoolxml2argvtest.c
-+++ b/tests/storagepoolxml2argvtest.c
-@@ -57,6 +57,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_ZFS:
-     case VIR_STORAGE_POOL_VSTORAGE:
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_LAST:
-     default:
-         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
-diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
-index 2010ef1356..072e2ff9e8 100644
---- a/tools/virsh-pool.c
-+++ b/tools/virsh-pool.c
-@@ -1187,6 +1187,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
-             case VIR_STORAGE_POOL_VSTORAGE:
-                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
-                 break;
-+            case VIR_STORAGE_POOL_VITASTOR:
-+                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
-+                break;
-             case VIR_STORAGE_POOL_LAST:
-                 break;
-             }

patches/qemu-10.1-vitastor.patch

@@ -1,172 +1,29 @@
-diff --git a/block/meson.build b/block/meson.build
+diff --git a/src/client/qemu_driver.c b/src/client/qemu_driver.c
-index 34b1b2a306..24ca0f1e52 100644
+index d8356dab..5f4cd50d 100644
---- a/block/meson.build
+--- a/src/client/qemu_driver.c
-+++ b/block/meson.build
++++ b/src/client/qemu_driver.c
-@@ -114,6 +114,7 @@ foreach m : [
+@@ -974,14 +974,21 @@ static void vitastor_co_read_bitmap_cb(void *opaque, long retval, uint8_t *bitma
-   [libnfs, 'nfs', files('nfs.c')],
+ #endif
-   [libssh, 'ssh', files('ssh.c')],
+ }
-   [rbd, 'rbd', files('rbd.c')],
-+  [vitastor, 'vitastor', files('vitastor.c')],
- ]
-   if m[0].found()
-     module_ss = ss.source_set()
-diff --git a/meson.build b/meson.build
-index 50c774a195..e5c7a3a4b1 100644
---- a/meson.build
-+++ b/meson.build
-@@ -1652,6 +1652,26 @@ if not get_option('rbd').auto() or have_block
-   endif
- endif
  
-+vitastor = not_found
+-static int coroutine_fn vitastor_co_block_status(
-+if not get_option('vitastor').auto() or have_block
+-    BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes,
-+  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
+-    int64_t *pnum, int64_t *map, BlockDriverState **file)
-+    required: get_option('vitastor'))
++static int coroutine_fn vitastor_co_block_status(BlockDriverState *bs,
-+  if libvitastor_client.found()
++#if QEMU_VERSION_MAJOR > 10 || QEMU_VERSION_MAJOR == 10 && QEMU_VERSION_MINOR >= 1
-+    if cc.links('''
++    unsigned int mode,
-+      #include <vitastor_c.h>
++#else
-+      int main(void) {
++    bool want_zero,
-+        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
++#endif
-+        return 0;
++    int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file)
-+      }''', dependencies: libvitastor_client)
+ {
-+      vitastor = declare_dependency(dependencies: libvitastor_client)
+     // Allocated => return BDRV_BLOCK_DATA|BDRV_BLOCK_OFFSET_VALID
-+    elif get_option('vitastor').enabled()
+     // Not allocated => return 0
-+      error('could not link libvitastor_client')
+     // Error => return -errno
-+    else
+     // Set pnum to length of the extent, `*map` = `offset`, `*file` = `bs`
-+      warning('could not link libvitastor_client, disabling')
++#if QEMU_VERSION_MAJOR > 10 || QEMU_VERSION_MAJOR == 10 && QEMU_VERSION_MINOR >= 1
-+    endif
++    int want_zero = (mode == BDRV_WANT_PRECISE);
-+  endif
++#endif
-+endif
+     VitastorRPC task;
-+
+     VitastorClient *client = bs->opaque;
- glusterfs = not_found
+     uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;
- glusterfs_ftruncate_has_stat = false
- glusterfs_iocb_has_stat = false
-@@ -2547,6 +2567,7 @@ endif
- config_host_data.set('CONFIG_OPENGL', opengl.found())
- config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
- config_host_data.set('CONFIG_RBD', rbd.found())
-+config_host_data.set('CONFIG_VITASTOR', vitastor.found())
- config_host_data.set('CONFIG_RDMA', rdma.found())
- config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
- config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
-@@ -4972,6 +4993,7 @@ summary_info += {'fdt support':       fdt_opt == 'internal' ? 'internal' : fdt}
- summary_info += {'libcap-ng support': libcap_ng}
- summary_info += {'bpf support':       libbpf}
- summary_info += {'rbd support':       rbd}
-+summary_info += {'vitastor support':  vitastor}
- summary_info += {'smartcard support': cacard}
- summary_info += {'U2F support':       u2f}
- summary_info += {'libusb':            libusb}
-diff --git a/meson_options.txt b/meson_options.txt
-index fff1521e58..f0844c0e00 100644
---- a/meson_options.txt
-+++ b/meson_options.txt
-@@ -202,6 +202,8 @@ option('pvg', type: 'feature', value: 'auto',
-        description: 'macOS paravirtualized graphics support')
- option('rbd', type : 'feature', value : 'auto',
-        description: 'Ceph block device driver')
-+option('vitastor', type : 'feature', value : 'auto',
-+       description: 'Vitastor block device driver')
- option('opengl', type : 'feature', value : 'auto',
-        description: 'OpenGL support')
- option('rdma', type : 'feature', value : 'auto',
-diff --git a/qapi/block-core.json b/qapi/block-core.json
-index dc6eb4ae23..d043f4340e 100644
---- a/qapi/block-core.json
-+++ b/qapi/block-core.json
-@@ -3280,7 +3280,7 @@
-             'parallels', 'preallocate', 'qcow', 'qcow2', 'qed', 'quorum',
-             'raw', 'rbd',
-             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
--            'ssh', 'throttle', 'vdi', 'vhdx',
-+            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
-             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
-@@ -4363,6 +4363,28 @@
-             '*key-secret': 'str',
-             '*server': ['InetSocketAddressBase'] } }
- 
-+##
-+# @BlockdevOptionsVitastor:
-+#
-+# Driver specific block device options for vitastor
-+#
-+# @image:       Image name
-+# @inode:       Inode number
-+# @pool:        Pool ID
-+# @size:        Desired image size in bytes
-+# @config-path: Path to Vitastor configuration
-+# @etcd-host:   etcd connection address(es)
-+# @etcd-prefix: etcd key/value prefix
-+##
-+{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { '*inode': 'uint64',
-+            '*pool': 'uint64',
-+            '*size': 'uint64',
-+            '*image': 'str',
-+            '*config-path': 'str',
-+            '*etcd-host': 'str',
-+            '*etcd-prefix': 'str' } }
-+
- ##
- # @ReplicationMode:
- #
-@@ -4831,6 +4853,7 @@
-       'throttle':   'BlockdevOptionsThrottle',
-       'vdi':        'BlockdevOptionsGenericFormat',
-       'vhdx':       'BlockdevOptionsGenericFormat',
-+      'vitastor':   'BlockdevOptionsVitastor',
-       'virtio-blk-vfio-pci':
-                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
-                       'if': 'CONFIG_BLKIO' },
-@@ -5304,6 +5327,20 @@
-             '*cluster-size' :   'size',
-             '*encrypt' :        'RbdEncryptionCreateOptions' } }
- 
-+##
-+# @BlockdevCreateOptionsVitastor:
-+#
-+# Driver specific image creation options for Vitastor.
-+#
-+# @location: Where to store the new image file.  This location cannot
-+#     point to a snapshot.
-+#
-+# @size: Size of the virtual disk in bytes
-+##
-+{ 'struct': 'BlockdevCreateOptionsVitastor',
-+  'data': { 'location':         'BlockdevOptionsVitastor',
-+            'size':             'size' } }
-+
- ##
- # @BlockdevVmdkSubformat:
- #
-@@ -5526,6 +5563,7 @@
-       'ssh':            'BlockdevCreateOptionsSsh',
-       'vdi':            'BlockdevCreateOptionsVdi',
-       'vhdx':           'BlockdevCreateOptionsVhdx',
-+      'vitastor':       'BlockdevCreateOptionsVitastor',
-       'vmdk':           'BlockdevCreateOptionsVmdk',
-       'vpc':            'BlockdevCreateOptionsVpc'
-   } }
-diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
-index 0ebe6bc52a..2c37ad3892 100644
---- a/scripts/meson-buildoptions.sh
-+++ b/scripts/meson-buildoptions.sh
-@@ -175,6 +175,7 @@ meson_options_help() {
-   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
-   printf "%s\n" '  qpl             Query Processing Library support'
-   printf "%s\n" '  rbd             Ceph block device driver'
-+  printf "%s\n" '  vitastor        Vitastor block device driver'
-   printf "%s\n" '  rdma            Enable RDMA-based migration'
-   printf "%s\n" '  replication     replication support'
-   printf "%s\n" '  rust            Rust support'
-@@ -459,6 +460,8 @@ _meson_option_parse() {
-     --disable-qpl) printf "%s" -Dqpl=disabled ;;
-     --enable-rbd) printf "%s" -Drbd=enabled ;;
-     --disable-rbd) printf "%s" -Drbd=disabled ;;
-+    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
-+    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
-     --enable-rdma) printf "%s" -Drdma=enabled ;;
-     --disable-rdma) printf "%s" -Drdma=disabled ;;
-     --enable-relocatable) printf "%s" -Drelocatable=true ;;

patches/qemu-10.2-vitastor.patch

@@ -1,172 +0,0 @@
-diff --git a/block/meson.build b/block/meson.build
-index 34b1b2a306..24ca0f1e52 100644
---- a/block/meson.build
-+++ b/block/meson.build
-@@ -114,6 +114,7 @@ foreach m : [
-   [libnfs, 'nfs', files('nfs.c')],
-   [libssh, 'ssh', files('ssh.c')],
-   [rbd, 'rbd', files('rbd.c')],
-+  [vitastor, 'vitastor', files('vitastor.c')],
- ]
-   if m[0].found()
-     module_ss = ss.source_set()
-diff --git a/meson.build b/meson.build
-index d9293294d8..776a5becc6 100644
---- a/meson.build
-+++ b/meson.build
-@@ -1665,6 +1665,26 @@ if not get_option('rbd').auto() or have_block
-   endif
- endif
- 
-+vitastor = not_found
-+if not get_option('vitastor').auto() or have_block
-+  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
-+    required: get_option('vitastor'))
-+  if libvitastor_client.found()
-+    if cc.links('''
-+      #include <vitastor_c.h>
-+      int main(void) {
-+        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
-+        return 0;
-+      }''', dependencies: libvitastor_client)
-+      vitastor = declare_dependency(dependencies: libvitastor_client)
-+    elif get_option('vitastor').enabled()
-+      error('could not link libvitastor_client')
-+    else
-+      warning('could not link libvitastor_client, disabling')
-+    endif
-+  endif
-+endif
-+
- glusterfs = not_found
- glusterfs_ftruncate_has_stat = false
- glusterfs_iocb_has_stat = false
-@@ -2509,6 +2529,7 @@ endif
- config_host_data.set('CONFIG_OPENGL', opengl.found())
- config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
- config_host_data.set('CONFIG_RBD', rbd.found())
-+config_host_data.set('CONFIG_VITASTOR', vitastor.found())
- config_host_data.set('CONFIG_RDMA', rdma.found())
- config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
- config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
-@@ -4948,6 +4969,7 @@ summary_info += {'fdt support':       fdt_opt == 'internal' ? 'internal' : fdt}
- summary_info += {'libcap-ng support': libcap_ng}
- summary_info += {'bpf support':       libbpf}
- summary_info += {'rbd support':       rbd}
-+summary_info += {'vitastor support':  vitastor}
- summary_info += {'smartcard support': cacard}
- summary_info += {'U2F support':       u2f}
- summary_info += {'libusb':            libusb}
-diff --git a/meson_options.txt b/meson_options.txt
-index 2836156257..148086cc6f 100644
---- a/meson_options.txt
-+++ b/meson_options.txt
-@@ -206,6 +206,8 @@ option('pvg', type: 'feature', value: 'auto',
-        description: 'macOS paravirtualized graphics support')
- option('rbd', type : 'feature', value : 'auto',
-        description: 'Ceph block device driver')
-+option('vitastor', type : 'feature', value : 'auto',
-+       description: 'Vitastor block device driver')
- option('opengl', type : 'feature', value : 'auto',
-        description: 'OpenGL support')
- option('rdma', type : 'feature', value : 'auto',
-diff --git a/qapi/block-core.json b/qapi/block-core.json
-index b82af74256..f25a6f5ce8 100644
---- a/qapi/block-core.json
-+++ b/qapi/block-core.json
-@@ -3351,7 +3351,7 @@
-             'parallels', 'preallocate', 'qcow', 'qcow2', 'qed', 'quorum',
-             'raw', 'rbd',
-             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
--            'ssh', 'throttle', 'vdi', 'vhdx',
-+            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
-             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
-@@ -4434,6 +4434,28 @@
-             '*key-secret': 'str',
-             '*server': ['InetSocketAddressBase'] } }
- 
-+##
-+# @BlockdevOptionsVitastor:
-+#
-+# Driver specific block device options for vitastor
-+#
-+# @image:       Image name
-+# @inode:       Inode number
-+# @pool:        Pool ID
-+# @size:        Desired image size in bytes
-+# @config-path: Path to Vitastor configuration
-+# @etcd-host:   etcd connection address(es)
-+# @etcd-prefix: etcd key/value prefix
-+##
-+{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { '*inode': 'uint64',
-+            '*pool': 'uint64',
-+            '*size': 'uint64',
-+            '*image': 'str',
-+            '*config-path': 'str',
-+            '*etcd-host': 'str',
-+            '*etcd-prefix': 'str' } }
-+
- ##
- # @ReplicationMode:
- #
-@@ -4902,6 +4924,7 @@
-       'throttle':   'BlockdevOptionsThrottle',
-       'vdi':        'BlockdevOptionsGenericFormat',
-       'vhdx':       'BlockdevOptionsGenericFormat',
-+      'vitastor':   'BlockdevOptionsVitastor',
-       'virtio-blk-vfio-pci':
-                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
-                       'if': 'CONFIG_BLKIO' },
-@@ -5376,6 +5399,20 @@
-             '*cluster-size' :   'size',
-             '*encrypt' :        'RbdEncryptionCreateOptions' } }
- 
-+##
-+# @BlockdevCreateOptionsVitastor:
-+#
-+# Driver specific image creation options for Vitastor.
-+#
-+# @location: Where to store the new image file.  This location cannot
-+#     point to a snapshot.
-+#
-+# @size: Size of the virtual disk in bytes
-+##
-+{ 'struct': 'BlockdevCreateOptionsVitastor',
-+  'data': { 'location':         'BlockdevOptionsVitastor',
-+            'size':             'size' } }
-+
- ##
- # @BlockdevVmdkSubformat:
- #
-@@ -5598,6 +5635,7 @@
-       'ssh':            'BlockdevCreateOptionsSsh',
-       'vdi':            'BlockdevCreateOptionsVdi',
-       'vhdx':           'BlockdevCreateOptionsVhdx',
-+      'vitastor':       'BlockdevCreateOptionsVitastor',
-       'vmdk':           'BlockdevCreateOptionsVmdk',
-       'vpc':            'BlockdevCreateOptionsVpc'
-   } }
-diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
-index 3d0d132344..65ee8c855e 100644
---- a/scripts/meson-buildoptions.sh
-+++ b/scripts/meson-buildoptions.sh
-@@ -177,6 +177,7 @@ meson_options_help() {
-   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
-   printf "%s\n" '  qpl             Query Processing Library support'
-   printf "%s\n" '  rbd             Ceph block device driver'
-+  printf "%s\n" '  vitastor        Vitastor block device driver'
-   printf "%s\n" '  rdma            Enable RDMA-based migration'
-   printf "%s\n" '  replication     replication support'
-   printf "%s\n" '  rust            Rust support'
-@@ -464,6 +465,8 @@ _meson_option_parse() {
-     --disable-qpl) printf "%s" -Dqpl=disabled ;;
-     --enable-rbd) printf "%s" -Drbd=enabled ;;
-     --disable-rbd) printf "%s" -Drbd=disabled ;;
-+    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
-+    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
-     --enable-rdma) printf "%s" -Drdma=enabled ;;
-     --disable-rdma) printf "%s" -Drdma=disabled ;;
-     --enable-relocatable) printf "%s" -Drelocatable=true ;;

rpm/vitastor-el10.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        3.0.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el10.tar.gz
+Source0:        vitastor-3.0.3.el10.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  gcc-c++

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        3.0.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el7.tar.gz
+Source0:        vitastor-3.0.3.el7.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  devtoolset-9-gcc-c++

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        3.0.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el8.tar.gz
+Source0:        vitastor-3.0.3.el8.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  gcc-toolset-9-gcc-c++

rpm/vitastor-el9.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        3.0.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el9.tar.gz
+Source0:        vitastor-3.0.3.el9.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  gcc-c++

src/CMakeLists.txt

@@ -1,8 +1,9 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
 include(GNUInstallDirs)
+include(CTest)
 include(CheckIncludeFile)
 
 find_package(PkgConfig)
@@ -20,7 +21,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 endif()
 set(ENABLE_COVERAGE false CACHE BOOL "Enable code coverage")
 
-add_definitions(-DVITASTOR_VERSION="3.0.12")
+add_definitions(-DVITASTOR_VERSION="3.0.3")
 add_definitions(-D_GNU_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fdiagnostics-color=always -fno-omit-frame-pointer -fvisibility=hidden -I ${CMAKE_SOURCE_DIR}/src)
 add_link_options(-fno-omit-frame-pointer)
 if (${WITH_ASAN})
@@ -74,6 +75,11 @@ if (RDMACM_LIBRARIES)
 	add_definitions(-DWITH_RDMACM)
 endif (RDMACM_LIBRARIES)
 
+find_package(OpenSSL)
+if (OPENSSL_FOUND)
+	add_definitions(-DWITH_OPENSSL)
+endif (OPENSSL_FOUND)
+
 if (${WITH_SYSTEM_LIBURING})
 	pkg_check_modules(LIBURING REQUIRED liburing>=2.10)
 	include_directories(${LIBURING_INCLUDE_DIRS})

src/blockstore/CMakeLists.txt

@@ -1,10 +1,10 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
 # libvitastor_blk.a
 add_library(vitastor_blk STATIC
-	../util/allocator.cpp ../util/crc32c.c ../util/ringloop.cpp
+	../util/allocator.cpp ../util/crc32c.c ../util/xxhash.c ../util/ringloop.cpp
 	multilist.cpp blockstore_heap.cpp blockstore_disk.cpp
 	blockstore.cpp blockstore_impl.cpp blockstore_init.cpp blockstore_open.cpp
 	blockstore_flush.cpp blockstore_read.cpp blockstore_stable.cpp blockstore_sync.cpp blockstore_write.cpp

src/blockstore/blockstore.h

@@ -228,9 +228,4 @@ public:
     virtual uint64_t get_journal_size() = 0;
 
     virtual uint32_t get_bitmap_granularity() = 0;
-
-    virtual uint64_t get_live_entries() = 0;
-    virtual uint64_t get_live_memory() = 0;
-    virtual uint64_t get_garbage_entries() = 0;
-    virtual uint64_t get_garbage_memory() = 0;
 };

src/blockstore/blockstore_disk.cpp

@@ -83,20 +83,21 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     {
         data_csum_type = BLOCKSTORE_CSUM_CRC32C;
     }
+    else if (config["data_csum_type"] == "xxh3_32")
+    {
+        data_csum_type = BLOCKSTORE_CSUM_XXH3_32;
+    }
     else if (config["data_csum_type"] == "" || config["data_csum_type"] == "none")
     {
         data_csum_type = BLOCKSTORE_CSUM_NONE;
     }
     else
     {
-        throw std::runtime_error("data_csum_type="+config["data_csum_type"]+" is unsupported, only \"crc32c\" and \"none\" are supported");
+        throw std::runtime_error("data_csum_type="+config["data_csum_type"]+" is unsupported, only \"crc32c\", \"xxh3_32\" and \"none\" are supported");
     }
     csum_block_size = parse_size(config["csum_block_size"]);
     discard_on_start = config.find("discard_on_start") != config.end() &&
         (config["discard_on_start"] == "true" || config["discard_on_start"] == "1" || config["discard_on_start"] == "yes");
-    gc_on_start = config.find("gc_on_start") == config.end() ||
-        (config["gc_on_start"] == "true" || config["gc_on_start"] == "1" || config["gc_on_start"] == "yes");
-    skip_double_claim = (config["skip_double_claim"] == "true" || config["skip_double_claim"] == "1" || config["skip_double_claim"] == "yes");
     min_discard_size = parse_size(config["min_discard_size"]);
     if (!min_discard_size)
         min_discard_size = 1024*1024;
@@ -176,7 +177,9 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     }
     if (data_block_size / bitmap_granularity < 8)
     {
-        throw std::runtime_error("Data block size must be at least bitmap_granularity*8");
+        fprintf(stderr, "Warning: block_size (%u) / bitmap_granularity (%u) = %u bits. "
+            "Consider using larger block_size or bitmap_granularity for better performance.\n",
+            data_block_size, bitmap_granularity, data_block_size / bitmap_granularity);
     }
     if (!data_csum_type)
     {

src/blockstore/blockstore_disk.h

@@ -16,6 +16,7 @@
 #define BLOCKSTORE_CSUM_NONE 0
 // Lower byte of checksum type is its length
 #define BLOCKSTORE_CSUM_CRC32C 0x104
+#define BLOCKSTORE_CSUM_XXH3_32 0x204
 
 #define MOCK_DATA_FD 1000
 #define MOCK_META_FD 1001
@@ -26,14 +27,14 @@ class allocator_t;
 struct blockstore_disk_t
 {
     std::string data_device, meta_device, journal_device;
-    uint64_t data_block_size;
+    uint32_t data_block_size;
     uint64_t cfg_journal_size, cfg_data_size;
     // Required write alignment and journal/metadata/data areas' location alignment
     uint32_t disk_alignment = 4096;
     // Journal block size - minimum_io_size of the journal device is the best choice
-    uint64_t journal_block_size = 4096;
+    uint32_t journal_block_size = 4096;
     // Metadata block size - minimum_io_size of the metadata device is the best choice
-    uint64_t meta_block_size = 4096;
+    uint32_t meta_block_size = 4096;
     // Atomic write size of the data block device
     uint32_t atomic_write_size = 4096;
     // Whether we should set RWF_ATOMIC on atomic writes
@@ -57,10 +58,6 @@ struct blockstore_disk_t
     bool inmemory_journal = true;
     // Data discard granularity and minimum size (for the sake of performance)
     bool discard_on_start = false;
-    // GC on start (new store)
-    bool gc_on_start = true;
-    // Skip double claim conflicts on start (new store, temporary until the bug is found)
-    bool skip_double_claim = false;
     uint64_t min_discard_size = 1024*1024;
     uint64_t discard_granularity = 0;
 

src/blockstore/blockstore_flush.cpp

@@ -174,18 +174,14 @@ bool journal_flusher_co::loop()
     else if (wait_state == 19) goto resume_19;
     else if (wait_state == 20) goto resume_20;
     else if (wait_state == 21) goto resume_21;
-    else if (wait_state == 22) goto resume_22;
-    else if (wait_state == 23) goto resume_23;
-    else if (wait_state == 24) goto resume_24;
-    else if (wait_state == 25) goto resume_25;
 resume_0:
     wait_state = 0;
     wait_count = 0;
     cur_oid = {};
     res = bs->heap->get_next_compact(cur_oid);
+    // Advance fsynced_lsn every <journal_trim_interval> intent writes
     if ((bs->intent_write_counter >= bs->journal_trim_interval) && co_id == 0)
     {
-        // Advance fsynced_lsn every <journal_trim_interval> intent writes
         bs->intent_write_counter = 0;
 resume_17:
 resume_18:
@@ -200,7 +196,6 @@ resume_21:
     if (res == ENOENT && flusher->force_start > 0 && co_id == 0 &&
         (!bs->dsk.disable_journal_fsync || !bs->dsk.disable_meta_fsync || !bs->dsk.disable_data_fsync))
     {
-        // When under pressure, do an additional fsync to force entries to be marked compactable
         flusher->active_flushers++;
 resume_14:
 resume_15:
@@ -264,9 +259,11 @@ resume_1:
         if (wr->type() == BS_HEAP_SMALL_WRITE ||
             wr->type() == BS_HEAP_INTENT_WRITE && bs->dsk.csum_block_size > bs->dsk.bitmap_granularity)
         {
-            bs->prepare_read(read_vec, cur_obj, wr, 0, bs->dsk.data_block_size,
+            auto res = bs->prepare_read(read_vec, cur_obj, wr, 0, bs->dsk.data_block_size,
                 wr->type() == BS_HEAP_INTENT_WRITE && bs->dsk.csum_block_size > bs->dsk.bitmap_granularity && !bs->perfect_csum_update
                 ? COPY_BUF_SKIP_CSUM : 0);
+            if (res > 0)
+                copy_count++;
         }
     });
     if (!compact_info.compact_lsn)
@@ -276,25 +273,6 @@ resume_1:
         bs->heap->unlock_entry(cur_oid);
         goto resume_0;
     }
-    flusher->active_flushers++;
-    for (i = 0; i < read_vec.size(); i++)
-    {
-        if ((read_vec[i].copy_flags & COPY_BUF_JOURNAL) &&
-            !(read_vec[i].copy_flags & COPY_BUF_COALESCED))
-        {
-            copy_count++;
-        }
-    }
-    if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-    {
-        init_fsync_data();
-    }
-    if (bs->log_level > 10)
-    {
-        printf("Compacting %jx:%jx v%ju..v%ju / l%ju..l%ju (%d writes)\n", cur_oid.inode, cur_oid.stripe,
-            compact_info.clean_wr->version, compact_info.compact_version,
-            compact_info.clean_wr->lsn, compact_info.compact_lsn, copy_count);
-    }
     mem_or(new_bmp, compact_info.clean_wr->get_int_bitmap(bs->heap), bs->dsk.clean_entry_bitmap_size);
     if (!bitmap_copied)
     {
@@ -313,6 +291,13 @@ resume_1:
         csum_copy.clear();
     }
     clean_loc = compact_info.clean_wr->big_location(bs->heap);
+    flusher->active_flushers++;
+    if (bs->log_level > 10)
+    {
+        printf("Compacting %jx:%jx v%ju..v%ju / l%ju..l%ju (%d writes)\n", cur_oid.inode, cur_oid.stripe,
+            compact_info.clean_wr->version, compact_info.compact_version,
+            compact_info.clean_wr->lsn, compact_info.compact_lsn, copy_count);
+    }
     overwrite_start = overwrite_end = 0;
     if (read_vec.size() > 0)
     {
@@ -351,13 +336,6 @@ resume_3:
     if (res == ENOENT || res == EDOM)
     {
         // Abort compaction
-abort_compact:
-        if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-        {
-            cur_sync->member_count--;
-            if (cur_sync->member_count > 0)
-                bs->ringloop->wakeup();
-        }
         flusher->flushing.erase(cur_oid);
         bs->heap->unlock_entry(cur_oid);
         flusher->active_flushers--;
@@ -371,7 +349,10 @@ resume_4:
         if (res == ENOENT)
         {
             // Abort compaction
-            goto abort_compact;
+            flusher->flushing.erase(cur_oid);
+            bs->heap->unlock_entry(cur_oid);
+            flusher->active_flushers--;
+            goto resume_0;
         }
         if (res == EAGAIN)
         {
@@ -400,14 +381,14 @@ resume_9:
     for (i = 0; i < read_vec.size(); i++)
     {
         if ((read_vec[i].copy_flags & COPY_BUF_JOURNAL) &&
-            !(read_vec[i].copy_flags & COPY_BUF_COALESCED))
+            !(read_vec[i].copy_flags & COPY_BUF_COALESCED) ||
+            (read_vec[i].copy_flags & COPY_BUF_PADDED)) // FIXME Shit, simplify these flags
         {
             assert(read_vec[i].buf);
             await_sqe(10);
             data->iov = (struct iovec){ read_vec[i].buf + (read_vec[i].copy_flags & COPY_BUF_PADDED
                 ? read_vec[i].offset - read_vec[i].disk_offset : 0), (size_t)read_vec[i].len };
             data->callback = simple_callback_w;
-            assert(clean_loc + read_vec[i].offset + data->iov.iov_len <= bs->dsk.block_count*bs->dsk.data_block_size);
             io_uring_prep_writev(sqe, bs->dsk.data_fd, &data->iov, 1, bs->dsk.data_offset + clean_loc + read_vec[i].offset);
             wait_count++;
         }
@@ -418,17 +399,6 @@ resume_11:
         wait_state = 11;
         return false;
     }
-    if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-    {
-resume_22:
-resume_23:
-resume_24:
-resume_25:
-        if (!fsync_data(22))
-        {
-            return false;
-        }
-    }
     // Lock is only needed to prevent freeing the big_write because we overwrite it...
     bs->heap->unlock_entry(cur_oid);
     // Mark the object compacted, but don't free and remove small_writes
@@ -438,14 +408,12 @@ resume_25:
     if (!cur_obj)
     {
         // Abort compaction
-        flusher->active_flushers--;
         flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
     if (!calc_block_checksums())
     {
         // Abort compaction
-        flusher->active_flushers--;
         flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
@@ -454,7 +422,6 @@ resume_25:
     if (res == EBUSY)
     {
         // Abort compaction, object is already overwritten by something else
-        flusher->active_flushers--;
         flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
@@ -619,7 +586,7 @@ int journal_flusher_co::check_and_punch_checksums()
             bs->heap->calc_block_checksums((uint32_t*)(new_csums+csum_off), vec.buf, punch_bmp, vec.offset, vec.offset+vec.len, true, NULL);
         }
     }
-    // Modified, we should punch_holes and then write the block to disk
+    // Modified, we should add_punch_holes and then write the block to disk
     return EBUSY;
 }
 
@@ -732,67 +699,6 @@ resume_1:
     return true;
 }
 
-void journal_flusher_co::init_fsync_data()
-{
-    cur_sync = flusher->data_syncs.begin();
-    if (cur_sync == flusher->data_syncs.end() || cur_sync->ready_count > 0)
-    {
-        cur_sync = flusher->data_syncs.emplace(cur_sync);
-    }
-    cur_sync->member_count++;
-}
-
-bool journal_flusher_co::fsync_data(int wait_base)
-{
-    if (wait_state == wait_base)
-        goto resume_0;
-    else if (wait_state == wait_base+1)
-        goto resume_1;
-    else if (wait_state == wait_base+2)
-        goto resume_2;
-    else if (wait_state == wait_base+3)
-        goto resume_3;
-    cur_sync->ready_count++;
-resume_0:
-    if (cur_sync->ready_count < cur_sync->member_count)
-    {
-        wait_state = wait_base;
-        return false;
-    }
-    if (!cur_sync->sent)
-    {
-        // Sync batch is ready. Do it.
-        await_sqe(1);
-        data->iov = { 0 };
-        data->callback = simple_callback_w;
-        io_uring_prep_fsync(sqe, bs->dsk.data_fd, IORING_FSYNC_DATASYNC);
-        cur_sync->sent = true;
-        wait_count++;
-resume_2:
-        if (wait_count > 0)
-        {
-            wait_state = wait_base+2;
-            return false;
-        }
-        cur_sync->done = true;
-        // Wake up other flushers
-        bs->ringloop->wakeup();
-    }
-resume_3:
-    if (!cur_sync->done)
-    {
-        wait_state = wait_base+3;
-        return false;
-    }
-    cur_sync->done_count++;
-    if (cur_sync->done_count >= cur_sync->member_count)
-    {
-        flusher->data_syncs.erase(cur_sync);
-        cur_sync = flusher->data_syncs.end();
-    }
-    return true;
-}
-
 bool journal_flusher_co::fsync_meta(int wait_base)
 {
     if (wait_state == wait_base)        goto resume_0;

src/blockstore/blockstore_flush.h

@@ -25,15 +25,6 @@ struct flusher_meta_write_t
     std::map<uint64_t, meta_sector_t>::iterator it;
 };
 
-struct flusher_data_sync_t
-{
-    int member_count = 0;
-    int ready_count = 0;
-    int done_count = 0;
-    bool sent = false;
-    bool done = false;
-};
-
 class journal_flusher_t;
 
 // Journal flusher coroutine
@@ -67,7 +58,6 @@ class journal_flusher_co
     int i, res;
     bool read_to_fill_incomplete;
     int copy_count;
-    std::list<flusher_data_sync_t>::iterator cur_sync;
 
     friend class journal_flusher_t;
 
@@ -78,8 +68,6 @@ class journal_flusher_co
     bool calc_block_checksums();
     bool write_meta_block(int wait_base);
     bool read_buffered(int wait_base);
-    void init_fsync_data();
-    bool fsync_data(int wait_base);
     bool fsync_meta(int wait_base);
     bool fsync_buffer(int wait_base);
     bool trim_lsn(int wait_base);
@@ -100,7 +88,6 @@ class journal_flusher_t
 
     robin_hood::unordered_flat_set<object_id> flushing;
     int active_flushers = 0;
-    std::list<flusher_data_sync_t> data_syncs;
     int wanting_meta_fsync = 0;
     bool fsyncing_meta = false;
     int syncing_buffer = 0;

src/blockstore/blockstore_heap.h

@@ -43,7 +43,7 @@ struct __attribute__((__packed__)) heap_entry_t
 {
     uint16_t size;
     uint16_t entry_type;
-    uint32_t crc32c;
+    uint32_t checksum;
     uint64_t lsn;
     uint64_t inode;
     uint64_t stripe;
@@ -57,11 +57,11 @@ struct __attribute__((__packed__)) heap_entry_t
     inline heap_small_write_t& small() { return *(heap_small_write_t*)this; }
     inline heap_big_write_t& big() { return *(heap_big_write_t*)this; }
     inline heap_big_intent_t& big_intent() { return *(heap_big_intent_t*)this; }
-    bool is_garbage() const;
+    bool is_garbage();
     void set_garbage();
-    bool is_overwrite() const;
+    bool is_overwrite();
-    bool is_compactable() const;
+    bool is_compactable();
-    bool is_before(const heap_entry_t *other) const;
+    bool is_before(heap_entry_t *other);
     uint32_t get_size(blockstore_heap_t *heap);
     uint8_t *get_ext_bitmap(blockstore_heap_t *heap);
     uint8_t *get_int_bitmap(blockstore_heap_t *heap);
@@ -69,7 +69,8 @@ struct __attribute__((__packed__)) heap_entry_t
     uint32_t *get_checksum(blockstore_heap_t *heap);
     uint64_t big_location(blockstore_heap_t *heap);
     void set_big_location(blockstore_heap_t *heap, uint64_t location);
-    uint32_t calc_crc32c();
+    uint32_t calc_checksum(blockstore_heap_t *heap);
+    uint32_t calc_checksum(blockstore_disk_t *dsk);
 };
 
 struct __attribute__((__packed__)) heap_small_write_t
@@ -80,7 +81,7 @@ struct __attribute__((__packed__)) heap_small_write_t
     uint32_t offset;
     uint32_t len;
 
-    // Also includes 1 bitmap and 1 crc32c after the bitmap if checksums are disabled
+    // Also includes 1 bitmap and 1 checksum after the bitmap if block checksums are disabled
 };
 
 struct __attribute__((__packed__)) heap_big_write_t
@@ -98,7 +99,7 @@ struct __attribute__((__packed__)) heap_big_intent_t
     uint32_t offset;
     uint32_t len;
 
-    // Also includes 2 bitmaps and 1 crc32c if checksums are disabled
+    // Also includes 2 bitmaps and 1 checksums if block checksums are disabled
 };
 
 struct __attribute__((__packed__)) heap_list_item_t
@@ -117,13 +118,10 @@ struct heap_object_mvcc_t
 
 struct heap_block_info_t
 {
-    struct __attribute__((__packed__))
+    uint32_t used_space = 0;
-    {
-        uint32_t used_space = 0;
-        uint32_t garbage_space = 0;
-    };
     uint64_t mod_lsn = 0, mod_lsn_to = 0; // only 1 block write of LSN sequence is allowed at a moment
-    bool is_writing = false;
+    bool is_writing: 1;
+    bool has_garbage: 1;
     std::vector<heap_list_item_t*> entries;
 };
 
@@ -158,16 +156,6 @@ struct heap_li_equal
     }
 };
 
-struct heap_recheck_state_t
-{
-    heap_entry_t *obj = NULL;
-    heap_entry_t *next_wr = NULL;
-    size_t total_reads = 0;
-    size_t sent_reads = 0;
-    size_t checked_reads = 0;
-    heap_entry_t *bad_wr = NULL;
-};
-
 using i64hash_t = robin_hood::hash<uint64_t>;
 using heap_inode_map_t = robin_hood::unordered_flat_set<heap_list_item_t*, heap_li_hash, heap_li_equal, 88>;
 using heap_block_index_t = robin_hood::unordered_flat_map<uint64_t,
@@ -197,11 +185,6 @@ class blockstore_heap_t
     uint64_t buffer_area_used_space = 0;
     uint64_t data_used_space = 0;
 
-    uint64_t live_entries = 0;
-    uint64_t live_memory = 0;
-    uint64_t garbage_entries = 0;
-    uint64_t garbage_memory = 0;
-
     uint64_t next_lsn = 0;
     uint32_t last_allocated_block = UINT32_MAX;
     heap_mvcc_map_t object_mvcc;
@@ -218,11 +201,9 @@ class blockstore_heap_t
 
     bool marked_used_blocks = false;
     bool recheck_queue_filled = false;
-    std::vector<heap_list_item_t*> postponed_items;
+    std::vector<heap_list_item_t*> loaded_list_items;
     std::set<uint32_t> recheck_modified_blocks;
     std::deque<heap_entry_t*> recheck_queue;
-    std::map<heap_entry_t*, heap_recheck_state_t> recheck_states;
-    size_t recheck_pending_reads = 0;
     int recheck_in_progress = 0;
     bool in_recheck = false;
     std::function<void(bool is_data, uint64_t offset, uint64_t len, uint8_t* buf, std::function<void()>)> recheck_cb;
@@ -231,22 +212,14 @@ class blockstore_heap_t
     uint64_t get_pg_id(inode_t inode, uint64_t stripe);
     bool validate_object(heap_entry_t *obj);
     void fill_recheck_queue();
-    void recheck_drop_entries(heap_entry_t *obj, heap_entry_t *bad_wr);
-    void recheck_start_reads(heap_recheck_state_t *st);
     int mark_used_blocks();
-    void init_free_bad_entry(heap_entry_t *wr);
-    void init_erase_bad_entry(heap_list_item_t *li);
-    bool init_erase_double_claim(heap_list_item_t *prev_li, heap_list_item_t *cur_li);
-    void recheck_full_gc();
     void recheck_buffer(heap_entry_t *cwr, uint8_t *buf);
     void defragment_block(uint32_t block_num);
     void reshard_add(heap_reshard_state_t *st, heap_list_item_t *li);
 
     void gc_block(heap_block_info_t & inf);
     int allocate_entry(uint32_t entry_size, uint32_t *block_num, bool allow_last_free);
-    void insert_list_items(heap_list_item_t** v, size_t count, bool postpone);
+    void insert_list_item(heap_list_item_t *li);
-    void remove_list_item(heap_list_item_t *li);
-    void unlink_list_item(heap_list_item_t *li);
     int add_entry(uint32_t wr_size, uint32_t *modified_block, bool allow_last_free,
         bool explicit_complete, std::function<void(heap_entry_t *wr)> fill_entry);
     int add_simple(heap_entry_t *obj, uint64_t version, uint32_t *modified_block, uint32_t entry_type);
@@ -373,10 +346,6 @@ public:
     uint32_t get_compact_queue_size();
     uint32_t get_to_compact_count();
     uint64_t get_compacted_count();
-    uint64_t get_live_entries();
-    uint64_t get_live_memory();
-    uint64_t get_garbage_entries();
-    uint64_t get_garbage_memory();
 
     uint64_t entry_pos(uint32_t block_num, uint32_t offset);
     heap_entry_t *entry_from_pos(uint64_t entry_pos, bool allow_unallocated = false);

src/blockstore/blockstore_impl.cpp

@@ -101,7 +101,6 @@ void blockstore_impl_t::loop()
         unsigned initial_ring_space = ringloop->space_left();
         int op_idx = 0, new_idx = 0;
         bool has_unfinished_writes = false;
-        bool has_unfinished_sync = false;
         for (; op_idx < submit_queue.size(); op_idx++, new_idx++)
         {
             auto op = submit_queue[op_idx];
@@ -139,13 +138,7 @@ void blockstore_impl_t::loop()
             else if (op->opcode == BS_OP_SYNC)
             {
                 // syncs only completed writes, so doesn't have to be blocked by anything
-                if (!has_unfinished_sync)
+                wr_st = continue_sync(op);
-                {
-                    wr_st = continue_sync(op);
-                    has_unfinished_sync = (wr_st != 2);
-                }
-                else
-                    wr_st = 0;
             }
             else if (op->opcode == BS_OP_STABLE || op->opcode == BS_OP_ROLLBACK)
             {
@@ -161,7 +154,9 @@ void blockstore_impl_t::loop()
                     wr_st = 2;
                 }
                 else
+                {
                     wr_st = 0;
+                }
             }
             if (wr_st == 2)
             {
@@ -198,12 +193,12 @@ void blockstore_impl_t::loop()
             heap->start_block_write(block_num);
             mb.sent = true;
         }
-        pending_modified_blocks.clear();
         int ret = ringloop->submit();
         if (ret < 0)
         {
             throw std::runtime_error(std::string("io_uring_submit: ") + strerror(-ret));
         }
+        pending_modified_blocks.clear();
         if ((initial_ring_space - ringloop->space_left()) > 0)
         {
             live = true;

src/blockstore/blockstore_impl.h

@@ -78,7 +78,6 @@ public:
     // Suitable only for server SSDs with capacitors, requires disabled data and journal fsyncs
     int immediate_commit = IMMEDIATE_NONE;
     bool inmemory_meta = false;
-    bool skip_corrupted_meta_entries = false;
     uint32_t meta_write_recheck_parallelism = 0;
     // Maximum and minimum flusher count
     unsigned max_flusher_count = 0, min_flusher_count = 0;
@@ -229,9 +228,4 @@ public:
     uint64_t get_free_block_count();
     inline uint32_t get_bitmap_granularity() { return dsk.bitmap_granularity; }
     inline uint64_t get_journal_size() { return dsk.journal_len; }
-
-    inline uint64_t get_live_entries() { return heap->get_live_entries(); }
-    inline uint64_t get_live_memory() { return heap->get_live_memory(); }
-    inline uint64_t get_garbage_entries() { return heap->get_garbage_entries(); }
-    inline uint64_t get_garbage_memory() { return heap->get_garbage_memory(); }
 };

src/blockstore/blockstore_init.cpp

@@ -145,7 +145,7 @@ resume_1:
             printf(
                 "Configuration stored in metadata superblock"
                 " (meta_block_size=%u, data_block_size=%u, bitmap_granularity=%u, data_csum_type=%u, csum_block_size=%u, meta_area_size=%ju)"
-                " differs from OSD configuration (%ju/%ju/%u, %u/%u, %ju).\n",
+                " differs from OSD configuration (%u/%u/%u, %u/%u, %ju).\n",
                 hdr->meta_block_size, hdr->data_block_size, hdr->bitmap_granularity,
                 hdr->data_csum_type, hdr->csum_block_size, hdr->meta_area_size,
                 bs->dsk.meta_block_size, bs->dsk.data_block_size, bs->dsk.bitmap_granularity,
@@ -153,14 +153,6 @@ resume_1:
             );
             exit(1);
         }
-        uint32_t csum = hdr->header_csum;
-        hdr->header_csum = 0;
-        if (crc32c(0, hdr, sizeof(*hdr)) != csum)
-        {
-            printf("Metadata header is corrupt (checksum mismatch).\n");
-            exit(1);
-        }
-        hdr->header_csum = csum;
     }
     bs->heap->start_load(((blockstore_meta_header_v3_t *)bs->meta_superblock)->completed_lsn);
     if (bs->dsk.inmemory_journal)
@@ -233,7 +225,7 @@ resume_4:
         {
             // Handle result
             uint64_t loaded = 0;
-            int r = bs->heap->load_blocks(bufs[i].offset-bs->dsk.meta_block_size, bufs[i].size, bufs[i].buf, bs->skip_corrupted_meta_entries, loaded);
+            int r = bs->heap->load_blocks(bufs[i].offset-bs->dsk.meta_block_size, bufs[i].size, bufs[i].buf, false, loaded);
             if (r != 0)
                 exit(1);
             entries_loaded += loaded;
@@ -248,7 +240,23 @@ resume_4:
     }
     // metadata read finished
     bs->heap->finish_load();
-    printf("Metadata entries loaded: %ju, rechecking unfinished writes and garbage entries\n", entries_loaded);
+    printf("Metadata entries loaded: %ju, used blocks: %ju / %ju\n", entries_loaded, bs->heap->get_data_used_space() / bs->dsk.data_block_size, bs->dsk.block_count);
+    if (zero_on_init && !bs->dsk.disable_meta_fsync)
+    {
+        GET_SQE();
+        io_uring_prep_fsync(sqe, bs->dsk.meta_fd, IORING_FSYNC_DATASYNC);
+        last_read_offset = 0;
+        data->iov = { 0 };
+        data->callback = [this](ring_data_t *data) { handle_event(data, -1); };
+        submitted++;
+        bs->ringloop->submit();
+    resume_5:
+        if (submitted > 0)
+        {
+            wait_state = 5;
+            return 1;
+        }
+    }
     // asynchronous recheck
 resume_6:
     wait_state = 6;
@@ -285,11 +293,6 @@ resume_7:
     if (bs->readonly)
     {
         recheck_mod.clear();
-        printf("Actual metadata entries: %ju\n", bs->heap->get_live_entries());
-    }
-    else
-    {
-        printf("Actual metadata entries: %ju, clearing garbage in %zu metadata blocks\n", bs->heap->get_live_entries(), recheck_mod.size());
     }
     for (i = 0; i < recheck_mod.size(); i++)
     {
@@ -303,7 +306,7 @@ resume_8:
         uint32_t block_num = recheck_mod[i];
         uint64_t block_offset = bs->dsk.meta_offset + (uint64_t)(block_num+1) * bs->dsk.meta_block_size;
         data = ((ring_data_t*)sqe->user_data);
-        uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, bs->dsk.meta_block_size);
+        uint8_t *buf = (uint8_t*)malloc_or_die(bs->dsk.meta_block_size);
         bs->heap->get_meta_block(block_num, buf);
         data->iov = { buf, bs->dsk.meta_block_size };
         data->callback = [this, buf, block_offset](ring_data_t *data)
@@ -329,25 +332,5 @@ resume_9:
     }
     free(metadata_buffer);
     metadata_buffer = NULL;
-    if (!bs->dsk.disable_meta_fsync && !bs->readonly)
-    {
-        GET_SQE();
-        io_uring_prep_fsync(sqe, bs->dsk.meta_fd, IORING_FSYNC_DATASYNC);
-        last_read_offset = 0;
-        data->iov = { 0 };
-        data->callback = [this](ring_data_t *data) { handle_event(data, -1); };
-        submitted++;
-        bs->ringloop->submit();
-    resume_5:
-        if (submitted > 0)
-        {
-            wait_state = 5;
-            return 1;
-        }
-    }
-    printf("Loading finished. Data used: %ju / %ju bytes (%s / %s)\n",
-        bs->heap->get_data_used_space(), bs->dsk.block_count * bs->dsk.data_block_size,
-        format_size(bs->heap->get_data_used_space()).c_str(),
-        format_size(bs->dsk.block_count * bs->dsk.data_block_size).c_str());
     return 0;
 }

src/blockstore/blockstore_open.cpp

@@ -28,7 +28,6 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
     throttle_target_parallelism = strtoull(config["throttle_target_parallelism"].c_str(), NULL, 10);
     throttle_threshold_us = strtoull(config["throttle_threshold_us"].c_str(), NULL, 10);
     perfect_csum_update = config["perfect_csum_update"] == "true" || config["perfect_csum_update"] == "1" || config["perfect_csum_update"] == "yes";
-    skip_corrupted_meta_entries = config["skip_corrupted_meta_entries"] == "true" || config["skip_corrupted_meta_entries"] == "1" || config["skip_corrupted_meta_entries"] == "yes";
     if (config["autosync_writes"] != "")
     {
         autosync_writes = strtoull(config["autosync_writes"].c_str(), NULL, 10);

src/blockstore/blockstore_read.cpp

@@ -462,10 +462,6 @@ int blockstore_impl_t::read_bitmap(object_id oid, uint64_t target_version, void
         {
             if (target_version >= wr->version)
             {
-                if (wr->type() == BS_HEAP_DELETE)
-                {
-                    return false;
-                }
                 found = true;
                 if (result_version)
                 {

src/blockstore/blockstore_stable.cpp

@@ -16,7 +16,6 @@ int blockstore_impl_t::dequeue_stable(blockstore_op_t *op)
     else if (priv->op_state == 5) goto resume_5;
     assert(!priv->op_state);
     op->retval = 0;
-    PRIV(op)->lsn = 0;
     priv->modified_block = priv->modified_block2 = UINT32_MAX;
     for (priv->stab_pos = 0; priv->stab_pos < op->len; priv->stab_pos++)
     {
@@ -37,12 +36,6 @@ int blockstore_impl_t::dequeue_stable(blockstore_op_t *op)
                 FINISH_OP(op);
                 return 2;
             }
-            if (res == ENOENT)
-            {
-                op->retval = -ENOENT;
-                FINISH_OP(op);
-                return 2;
-            }
             if (res == ENOSPC)
             {
                 if (!heap->get_to_compact_count())

src/blockstore/blockstore_sync.cpp

@@ -9,7 +9,6 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
     if (!PRIV(op)->op_state)
     {
         op->retval = 0;
-        PRIV(op)->lsn = 0;
     }
     int res = do_sync(op, 0);
     if (res == 2)
@@ -105,8 +104,7 @@ int blockstore_impl_t::do_sync(blockstore_op_t *op, int base_state)
         unsynced_data_write_count = unsynced_buffer_write_count = unsynced_meta_write_count = 0;
         return 2;
     }
-    assert(!PRIV(op)->lsn);
+    PRIV(op)->modified_block = heap->get_completed_lsn();
-    PRIV(op)->lsn = heap->get_completed_lsn();
     if (!submit_fsyncs(PRIV(op)->pending_ops))
     {
         PRIV(op)->wait_detail = 1;
@@ -120,6 +118,6 @@ resume_1:
         return 1;
     }
 resume_2:
-    heap->mark_lsn_fsynced(PRIV(op)->lsn);
+    heap->mark_lsn_fsynced(PRIV(op)->modified_block);
     return 2;
 }

src/blockstore/blockstore_write.cpp

@@ -22,24 +22,21 @@ void blockstore_impl_t::prepare_meta_block_write(uint32_t modified_block)
     ring_data_t *data = ((ring_data_t*)sqe->user_data);
     uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.meta_block_size);
     data->iov = (struct iovec){ buf, (size_t)dsk.meta_block_size };
-    data->callback = [this, modified_block](ring_data_t *data)
+    data->callback = [this, modified_block, buf](ring_data_t *data)
     {
+        free(buf);
         live = true;
         if (data->res != data->iov.iov_len)
         {
             // FIXME: our state becomes corrupted after a write error. maybe do something better than just die
             disk_error_abort("data write", data->res, data->iov.iov_len);
         }
-        auto it = modified_blocks.find(modified_block);
+        modified_blocks.erase(modified_block);
-        assert(it != modified_blocks.end());
-        free(it->second.buf);
-        modified_blocks.erase(it);
         heap->complete_block_write(modified_block);
         ringloop->wakeup();
     };
-    assert(((uint64_t)modified_block+2)*dsk.meta_block_size <= dsk.meta_area_size);
     io_uring_prep_writev(
-        sqe, dsk.meta_fd, &data->iov, 1, dsk.meta_offset + ((uint64_t)modified_block+1)*dsk.meta_block_size
+        sqe, dsk.meta_fd, &data->iov, 1, dsk.meta_offset + (modified_block+1)*dsk.meta_block_size
     );
     unsynced_meta_write_count++;
     pending_modified_blocks.push_back(modified_block);
@@ -178,7 +175,6 @@ enospc:
         ring_data_t *data = ((ring_data_t*)sqe->user_data);
         data->iov = (struct iovec){ op->buf, op->len };
         data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-        assert(loc+op->offset+op->len <= dsk.block_count*dsk.data_block_size);
         io_uring_prep_writev(sqe, dsk.data_fd, &data->iov, 1, dsk.data_offset + loc + op->offset);
         PRIV(op)->pending_ops++;
         write_iodepth++;
@@ -253,12 +249,13 @@ enospc:
             goto enospc;
         assert(res == 0);
         PRIV(op)->lsn = obj->lsn;
+        if (op->len)
+            heap->use_buffer_area(op->oid.inode, loc, op->len);
         prepare_meta_block_write(PRIV(op)->modified_block);
         PRIV(op)->pending_ops++;
         if (op->len > 0)
         {
             // Prepare buffered data write
-            heap->use_buffer_area(op->oid.inode, loc, op->len);
             if (dsk.inmemory_journal)
             {
                 memcpy((uint8_t*)buffer_area + loc, op->buf, op->len);
@@ -266,7 +263,6 @@ enospc:
             BS_SUBMIT_GET_SQE(sqe2, data2);
             data2->iov = (struct iovec){ op->buf, op->len };
             data2->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-            assert(loc+op->len <= dsk.journal_len);
             io_uring_prep_writev(sqe2, dsk.journal_fd, &data2->iov, 1, dsk.journal_offset + loc);
             PRIV(op)->pending_ops++;
         }
@@ -352,7 +348,6 @@ resume_12:
     }
 resume_4:
     {
-        BS_SUBMIT_CHECK_SQES(1);
         auto obj = heap->read_entry(op->oid);
         int res = 0;
         if (PRIV(op)->write_type == _REDIRECT_INTENT)
@@ -409,12 +404,11 @@ resume_6:
         if (ref_us > exec_us + throttle_threshold_us)
         {
             // Pause reply
-            PRIV(op)->pending_ops++;
             PRIV(op)->op_state = 7;
             // Remember that the timer can in theory be called right here
             tfd->set_timer_us(ref_us-exec_us, false, [this, op](int timer_id)
             {
-                PRIV(op)->pending_ops--;
+                PRIV(op)->op_state = 8;
                 ringloop->wakeup();
             });
             return 1;
@@ -456,7 +450,6 @@ resume_10:
     BS_SUBMIT_GET_SQE(sqe, data);
     data->iov = (struct iovec){ op->buf, op->len };
     data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-    assert(PRIV(op)->location + op->offset <= dsk.block_count*dsk.data_block_size);
     io_uring_prep_writev(sqe, dsk.data_fd, &data->iov, 1, dsk.data_offset + PRIV(op)->location + op->offset);
     if (dsk.use_atomic_flag)
         sqe->rw_flags = RWF_ATOMIC;

src/blockstore/ondisk_formats.h

@@ -141,7 +141,7 @@ struct __attribute__((__packed__)) journal_entry
 inline uint32_t je_crc32(journal_entry *je)
 {
     // 0x48674bc7 = crc32(4 zero bytes)
-    return je->size < 4 ? 0 : crc32c(0x48674bc7, ((uint8_t*)je)+4, je->size-4);
+    return crc32c(0x48674bc7, ((uint8_t*)je)+4, je->size-4);
 }
 
 // "VITAstor"

src/blockstore/v1/flush.cpp

@@ -520,7 +520,6 @@ resume_2:
                 await_sqe(15);
                 data->iov = (struct iovec){ it->buf, (size_t)it->len };
                 data->callback = simple_callback_w;
-                assert(clean_loc+it->offset+it->len <= bs->dsk.block_count*bs->dsk.data_block_size);
                 io_uring_prep_writev(
                     sqe, bs->dsk.data_fd, &data->iov, 1, bs->dsk.data_offset + clean_loc + it->offset
                 );
@@ -750,7 +749,6 @@ bool journal_flusher_co::write_meta_block(flusher_meta_write_t & meta_block, int
     await_sqe(0);
     data->iov = (struct iovec){ meta_block.buf, (size_t)bs->dsk.meta_block_size };
     data->callback = simple_callback_w;
-    assert(bs->dsk.meta_block_size + meta_block.sector + bs->dsk.meta_block_size <= bs->dsk.meta_area_size);
     io_uring_prep_writev(
         sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset + bs->dsk.meta_block_size + meta_block.sector
     );

src/blockstore/v1/impl.cpp

@@ -855,29 +855,4 @@ std::string blockstore_impl_t::get_op_diag(blockstore_op_t *op)
     return std::string(buf);
 }
 
-uint64_t blockstore_impl_t::get_live_entries()
-{
-    return used_blocks;
-}
-
-uint64_t blockstore_impl_t::get_live_memory()
-{
-    uint64_t used = 0;
-    for (auto & kv: clean_db_shards)
-    {
-        used += kv.second.size() * sizeof(blockstore_clean_db_t::value_type);
-    }
-    return used;
-}
-
-uint64_t blockstore_impl_t::get_garbage_entries()
-{
-    return dirty_db.size();
-}
-
-uint64_t blockstore_impl_t::get_garbage_memory()
-{
-    return (sizeof(obj_ver_id) + sizeof(dirty_entry) + 32) * dirty_db.size();
-}
-
 } // namespace v1

src/blockstore/v1/impl.h

@@ -332,10 +332,6 @@ public:
     inline uint64_t get_free_block_count() { return dsk.block_count - used_blocks; }
     inline uint32_t get_bitmap_granularity() { return dsk.disk_alignment; }
     inline uint64_t get_journal_size() { return dsk.journal_len; }
-    uint64_t get_live_entries();
-    uint64_t get_live_memory();
-    uint64_t get_garbage_entries();
-    uint64_t get_garbage_memory();
 };
 
 } // namespace v1

src/blockstore/v1/init.cpp

@@ -189,7 +189,7 @@ resume_1:
             printf(
                 "Configuration stored in metadata superblock"
                 " (meta_block_size=%u, data_block_size=%u, bitmap_granularity=%u, data_csum_type=%u, csum_block_size=%u)"
-                " differs from OSD configuration (%ju/%ju/%u, %u/%u).\n",
+                " differs from OSD configuration (%u/%u/%u, %u/%u).\n",
                 hdr->meta_block_size, hdr->data_block_size, hdr->bitmap_granularity,
                 hdr->data_csum_type, hdr->csum_block_size,
                 bs->dsk.meta_block_size, bs->dsk.data_block_size, bs->dsk.bitmap_granularity,

src/blockstore/v1/journal.cpp

@@ -193,7 +193,6 @@ void blockstore_impl_t::prepare_journal_sector_write(int cur_sector, blockstore_
             (size_t)journal.block_size
         };
         data->callback = [this, flush_id = journal.submit_id](ring_data_t *data) { handle_journal_write(data, flush_id); };
-        assert(journal.sector_info[cur_sector].offset+journal.block_size <= dsk.journal_len);
         io_uring_prep_writev(
             sqe, dsk.journal_fd, &data->iov, 1, journal.offset + journal.sector_info[cur_sector].offset
         );

src/blockstore/v1/read.cpp

@@ -620,7 +620,8 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
     else if (from_journal)
     {
         // Don't scan bitmap - journal writes don't have holes (internal bitmap)!
-        uint8_t *csum = !dsk.csum_block_size ? 0 : (clean_entry_bitmap + dsk.clean_entry_bitmap_size);
+        uint8_t *csum = !dsk.csum_block_size ? 0 : (clean_entry_bitmap + dsk.clean_entry_bitmap_size +
+            item_start/dsk.csum_block_size*(dsk.data_csum_type & 0xFF));
         if (!fulfill_read(read_op, fulfilled, item_start, item_end,
             (BS_ST_BIG_WRITE | BS_ST_STABLE), 0, clean_loc + item_start, 0, csum, dyn_data))
         {

src/blockstore/v1/write.cpp

@@ -368,9 +368,9 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
         }
         data->iov.iov_len = op->len + stripe_offset + stripe_end; // to check it in the callback
         data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-        const uint64_t write_offset = (loc * dsk.data_block_size) + op->offset - stripe_offset;
+        io_uring_prep_writev(
-        assert(write_offset+op->len+stripe_offset+stripe_end <= dsk.block_count*dsk.data_block_size);
+            sqe, dsk.data_fd, PRIV(op)->iov_zerofill, vcnt, dsk.data_offset + (loc * dsk.data_block_size) + op->offset - stripe_offset
-        io_uring_prep_writev(sqe, dsk.data_fd, PRIV(op)->iov_zerofill, vcnt, dsk.data_offset + write_offset);
+        );
         PRIV(op)->pending_ops = 1;
         if (!(dirty_it->second.state & BS_ST_INSTANT))
         {
@@ -495,8 +495,9 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
                 .op = op,
             });
             data2->callback = [this, flush_id = journal.submit_id](ring_data_t *data) { handle_journal_write(data, flush_id); };
-            assert(journal.next_free+op->len <= dsk.journal_len);
+            io_uring_prep_writev(
-            io_uring_prep_writev(sqe2, dsk.journal_fd, &data2->iov, 1, journal.offset + journal.next_free);
+                sqe2, dsk.journal_fd, &data2->iov, 1, journal.offset + journal.next_free
+            );
             PRIV(op)->pending_ops++;
         }
         else

src/client/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
@@ -12,11 +12,11 @@ if (RDMACM_LIBRARIES)
 	set(MSGR_RDMACM "msgr_rdmacm.cpp")
 endif (RDMACM_LIBRARIES)
 add_library(vitastor_common STATIC
-	../util/epoll_manager.cpp etcd_state_client.cpp messenger.cpp msgr_iothread.cpp ../util/addr_util.cpp
+	../util/epoll_manager.cpp etcd_state_client.cpp messenger.cpp ../util/addr_util.cpp
-	msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ../util/ringloop.cpp ../../json11/json11.cpp
+	msgr_encrypt.cpp msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ../util/ringloop.cpp ../../json11/json11.cpp
 	http_client.cpp osd_ops.cpp pg_states.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ../util/json_util.cpp ${MSGR_RDMA} ${MSGR_RDMACM}
 )
-target_link_libraries(vitastor_common pthread)
+target_link_libraries(vitastor_common pthread ${OPENSSL_LIBRARIES})
 target_compile_options(vitastor_common PUBLIC -fPIC)
 
 # libvitastor_client.so
@@ -33,6 +33,7 @@ target_link_libraries(vitastor_client
 	${LIBURING_LIBRARIES}
 	${IBVERBS_LIBRARIES}
 	${RDMACM_LIBRARIES}
+	${OPENSSL_LIBRARIES}
 )
 set_target_properties(vitastor_client PROPERTIES VERSION ${VITASTOR_VERSION} SOVERSION 0)
 configure_file(vitastor.pc.in vitastor.pc @ONLY)
@@ -52,6 +53,9 @@ if (${WITH_FIO})
 		../util/rw_blocking.cpp
 		../util/addr_util.cpp
 	)
+	target_link_libraries(fio_vitastor_sec
+		tcmalloc_minimal
+	)
 endif (${WITH_FIO})
 
 # vitastor-nbd
@@ -95,10 +99,10 @@ endif (${WITH_QEMU})
 add_executable(test_cluster_client
 	EXCLUDE_FROM_ALL
 	../test/test_cluster_client.cpp
-	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp cluster_client_wb.cpp msgr_op.cpp ../test/mock/messenger.cpp msgr_stop.cpp
+	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp cluster_client_wb.cpp msgr_op.cpp ../test/mock/messenger.cpp msgr_stop.cpp msgr_encrypt.cpp
 	etcd_state_client.cpp ../util/timerfd_manager.cpp ../util/addr_util.cpp ../util/str_util.cpp ../util/json_util.cpp ../../json11/json11.cpp
 )
-target_link_libraries(test_cluster_client ${LIBURING_LIBRARIES})
+target_link_libraries(test_cluster_client ${OPENSSL_LIBRARIES})
 target_compile_definitions(test_cluster_client PUBLIC -D__MOCK__)
 target_include_directories(test_cluster_client BEFORE PUBLIC ${CMAKE_SOURCE_DIR}/src/test/mock)
 add_dependencies(build_tests test_cluster_client)

src/client/cluster_client.cpp

@@ -27,7 +27,7 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
     msgr.ringloop = ringloop;
     msgr.repeer_pgs = [this](osd_num_t peer_osd)
     {
-        if (msgr.osd_peers.find(peer_osd) != msgr.osd_peers.end())
+        if (msgr.osd_peer_fds.find(peer_osd) != msgr.osd_peer_fds.end())
         {
             // peer_osd just connected
             continue_ops();
@@ -47,8 +47,8 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
     msgr.exec_op = [this](osd_op_t *op)
     {
         // Garbage in
-        fprintf(stderr, "Can't handle incoming operation from client %lu\n", op->client_id);
+        fprintf(stderr, "Incoming garbage from peer %d\n", op->peer_fd);
-        msgr.stop_client(op->client_id);
+        msgr.stop_client(op->peer_fd);
         delete op;
     };
     msgr.parse_config(config);
@@ -156,7 +156,7 @@ void cluster_client_t::continue_raw_ops(osd_num_t peer_osd)
     {
         auto op = it->second;
         op->op_type = OSD_OP_OUT;
-        op->client_id = msgr.osd_peers.at(peer_osd)->client_id;
+        op->peer_fd = msgr.osd_peer_fds.at(peer_osd);
         msgr.outbox_push(op);
         raw_ops.erase(it++);
     }
@@ -590,7 +590,7 @@ void cluster_client_t::on_change_pool_config_hook()
         {
             if (log_level > 2 && pg_counts[pool_item.first])
             {
-                fprintf(stderr, "Pool %u (%s) PG count changed from %lu to %lu\n", pool_item.first, pool_item.second.name.c_str(),
+                printf("Pool %u (%s) PG count changed from %lu to %lu\n", pool_item.first, pool_item.second.name.c_str(),
                     pg_counts[pool_item.first], pool_item.second.real_pg_count);
             }
             // At this point, all pool operations should have been suspended
@@ -871,13 +871,13 @@ void cluster_client_t::execute_cas(cluster_op_t *op)
         if (op->retval != expected && op->retval >= 0)
             op->retval = -EIO;
         op->retval = op->retval == -EPIPE ? -EINTR : op->retval;
-        auto peer_it = msgr.osd_peers.find(op->parts[0].osd_num);
+        auto peer_it = msgr.osd_peer_fds.find(op->parts[0].osd_num);
         if (op->retval != 0 || (op->flags & OP_IMMEDIATE_COMMIT))
         {
             auto cb = std::move(op->callback);
             cb(op);
         }
-        else if (peer_it == msgr.osd_peers.end())
+        else if (peer_it == msgr.osd_peer_fds.end())
         {
             // Care must be taken to make sure that the client doesn't reconnect to the OSD
             // before executing the previously completed operation callback (!)
@@ -888,10 +888,10 @@ void cluster_client_t::execute_cas(cluster_op_t *op)
         else
         {
             // CAS writes have a built-in sync
-            osd_client_t *cl = peer_it->second;
+            auto peer_fd = peer_it->second;
             *part = (osd_op_t){
                 .op_type = OSD_OP_OUT,
-                .client_id = cl->client_id,
+                .peer_fd = peer_fd,
                 .req = {
                     .hdr = {
                         .magic = SECONDARY_OSD_OP_MAGIC,
@@ -958,9 +958,22 @@ bool cluster_client_t::check_rw(cluster_op_t *op)
     {
         op->flags |= OP_IMMEDIATE_COMMIT;
     }
+    auto ino_it = st_cli.inode_config.find(op->inode);
+    if (ino_it != st_cli.inode_config.end() && ino_it->second.enc)
+    {
+        // FIXME: Rework client API by adding open/close and cache inode information in the "FD"
+        op->enc = ino_it->second.enc;
+        if (!op->enc->bitmap_granularity)
+        {
+            op->enc->bitmap_granularity = pool_it->second.bitmap_granularity;
+        }
+    }
+    else
+    {
+        op->enc.reset();
+    }
     if ((op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_DELETE) && !(op->flags & OSD_OP_IGNORE_READONLY))
     {
-        auto ino_it = st_cli.inode_config.find(op->inode);
         if (ino_it != st_cli.inode_config.end() && ino_it->second.readonly)
         {
             op->retval = -EROFS;
@@ -972,7 +985,6 @@ bool cluster_client_t::check_rw(cluster_op_t *op)
     op->deoptimise_snapshot = false;
     if (enable_writeback && (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP || op->opcode == OSD_OP_READ_CHAIN_BITMAP))
     {
-        auto ino_it = st_cli.inode_config.find(op->inode);
         if (ino_it != st_cli.inode_config.end())
         {
             int chain_size = 0;
@@ -1004,11 +1016,11 @@ bool cluster_client_t::check_rw(cluster_op_t *op)
 
 void cluster_client_t::execute_raw(osd_num_t osd_num, osd_op_t *op)
 {
-    auto peer_it = msgr.osd_peers.find(osd_num);
+    auto fd_it = msgr.osd_peer_fds.find(osd_num);
-    if (peer_it != msgr.osd_peers.end())
+    if (fd_it != msgr.osd_peer_fds.end())
     {
         op->op_type = OSD_OP_OUT;
-        op->client_id = peer_it->second->client_id;
+        op->peer_fd = fd_it->second;
         msgr.outbox_push(op);
     }
     else
@@ -1119,13 +1131,6 @@ resume_2:
         // Finished successfully
         // Even if the PG count has changed in meanwhile we treat it as success
         // because if some operations were invalid for the new PG count we'd get errors
-        if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP || op->opcode == OSD_OP_READ_CHAIN_BITMAP)
-        {
-            // Copy part bitmaps only after finishing all part reads
-            for (auto & part: op->parts)
-                if ((part.flags & (PART_SENT|PART_DONE|PART_VALID)) == (PART_SENT|PART_DONE|PART_VALID))
-                    copy_part_bitmap(op, &part);
-        }
         if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_CHAIN_BITMAP)
         {
             // Check parent inode
@@ -1171,7 +1176,7 @@ resume_2:
         erase_op(op);
         return 1;
     }
-    else if (op->retval != 0 && op->opcode != OSD_OP_SYNC && !(op->flags & OP_FLUSH_BUFFER) &&
+    else if (op->retval != 0 && !(op->flags & OP_FLUSH_BUFFER) &&
         op->retval != -EPIPE && (op->retval != -EIO || !client_eio_retry_interval) && (op->retval != -ENOSPC || !client_retry_enospc))
     {
         // Fatal error (neither -EPIPE, -EIO nor -ENOSPC)
@@ -1408,10 +1413,10 @@ int cluster_client_t::try_send(cluster_op_t *op, int i, std::function<void(osd_o
                 primary_osd = nearest_osd;
         }
         part->osd_num = primary_osd;
-        auto peer_it = msgr.osd_peers.find(primary_osd);
+        auto peer_it = msgr.osd_peer_fds.find(primary_osd);
-        if (peer_it != msgr.osd_peers.end())
+        if (peer_it != msgr.osd_peer_fds.end())
         {
-            osd_client_t *cl = peer_it->second;
+            int peer_fd = peer_it->second;
             part->flags |= PART_SENT|PART_VALID;
             op->inflight_count++;
             uint64_t pg_bitmap_size = (pool_cfg.data_block_size / pool_cfg.bitmap_granularity / 8) * (
@@ -1426,7 +1431,7 @@ int cluster_client_t::try_send(cluster_op_t *op, int i, std::function<void(osd_o
             }
             part->op = (osd_op_t){
                 .op_type = OSD_OP_OUT,
-                .client_id = cl->client_id,
+                .peer_fd = peer_fd,
                 .req = { .rw = {
                     .header = {
                         .magic = SECONDARY_OSD_OP_MAGIC,
@@ -1442,6 +1447,7 @@ int cluster_client_t::try_send(cluster_op_t *op, int i, std::function<void(osd_o
                     ? (uint8_t*)op->part_bitmaps + pg_bitmap_size*i : NULL),
                 .bitmap_len = (unsigned)(op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP || op->opcode == OSD_OP_READ_CHAIN_BITMAP
                     ? pg_bitmap_size : 0),
+                .enc = op->enc,
                 .callback = cb ? cb : [this, part](osd_op_t *op_part)
                 {
                     handle_op_part(part);
@@ -1475,8 +1481,8 @@ int cluster_client_t::continue_sync(cluster_op_t *op)
     for (auto do_it = dirty_osds.begin(); do_it != dirty_osds.end(); )
     {
         osd_num_t sync_osd = *do_it;
-        auto peer_it = msgr.osd_peers.find(sync_osd);
+        auto peer_it = msgr.osd_peer_fds.find(sync_osd);
-        if (peer_it == msgr.osd_peers.end())
+        if (peer_it == msgr.osd_peer_fds.end())
             dirty_osds.erase(do_it++);
         else
             do_it++;
@@ -1529,12 +1535,12 @@ resume_1:
 
 void cluster_client_t::send_sync(cluster_op_t *op, cluster_op_part_t *part)
 {
-    osd_client_t *cl = msgr.osd_peers.at(part->osd_num);
+    auto peer_fd = msgr.osd_peer_fds.at(part->osd_num);
     part->flags |= PART_SENT;
     op->inflight_count++;
     part->op = (osd_op_t){
         .op_type = OSD_OP_OUT,
-        .client_id = cl->client_id,
+        .peer_fd = peer_fd,
         .req = {
             .hdr = {
                 .magic = SECONDARY_OSD_OP_MAGIC,
@@ -1574,10 +1580,10 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
             // Error priority: EIO > ENOSPC > ETIMEDOUT > EPIPE
             op->retval = part->op.reply.hdr.retval;
         }
-        uint64_t stop_client_id = 0;
+        int stop_fd = -1;
         if (op->retval != -EINTR && op->retval != -EIO && op->retval != -ENOSPC)
         {
-            stop_client_id = part->op.client_id;
+            stop_fd = part->op.peer_fd;
             if (op->retval != -EPIPE || log_level > 0)
             {
                 fprintf(
@@ -1604,9 +1610,9 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
             op->retry_after = op->retval != -EPIPE ? client_eio_retry_interval : client_retry_interval;
         }
         reset_retry_timer(op->retry_after);
-        if (stop_client_id)
+        if (stop_fd >= 0)
         {
-            msgr.stop_client(stop_client_id);
+            msgr.stop_client(stop_fd);
         }
         op->inflight_count--;
         if (op->inflight_count == 0 && !op->retry_after)
@@ -1637,6 +1643,13 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
         }
         if (op->inflight_count == 0 && !op->retry_after)
         {
+            // Copy part bitmaps only after finishing all part reads
+            if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP || op->opcode == OSD_OP_READ_CHAIN_BITMAP)
+            {
+                for (auto & part: op->parts)
+                    if (part.flags == (PART_SENT|PART_VALID|PART_DONE))
+                        copy_part_bitmap(op, &part);
+            }
             if (op->opcode == OSD_OP_SYNC)
                 continue_sync(op);
             else

src/client/cluster_client.h

@@ -71,6 +71,7 @@ protected:
     cluster_op_t *prev = NULL, *next = NULL;
     int prev_wait = 0;
     uint64_t flush_id = 0;
+    std::shared_ptr<inode_enc_t> enc;
     friend class cluster_client_t;
     friend class writeback_cache_t;
 };
@@ -83,6 +84,9 @@ class writeback_cache_t;
 // FIXME: Split into public and private interfaces
 class __attribute__((visibility("default"))) cluster_client_t
 {
+#ifdef __MOCK__
+public:
+#endif
     timerfd_manager_t *tfd = NULL;
     ring_loop_t *ringloop = NULL;
 
@@ -152,9 +156,15 @@ public:
     void list_inode(inode_t inode, uint64_t min_offset, uint64_t max_offset, int max_parallel_pgs, std::function<void(
         int status, int pgs_left, pg_num_t pg_num, std::set<object_id>&& objects)> pg_callback);
 
+    //inline uint32_t get_bs_bitmap_granularity() { return st_cli.global_bitmap_granularity; }
+    //inline uint64_t get_bs_block_size() { return st_cli.global_block_size; }
+
+#ifndef __MOCK__
 protected:
+#endif
     void continue_ops(int time_passed = 0);
 
+protected:
     bool affects_osd(uint64_t inode, uint64_t offset, uint64_t len, osd_num_t osd);
     bool affects_pg(uint64_t inode, uint64_t offset, uint64_t len, pool_id_t pool_id, pg_num_t pg_num);
 
@@ -195,5 +205,4 @@ protected:
     osd_num_t select_nearest_osd(const std::vector<osd_num_t> & osds);
 
     friend class writeback_cache_t;
-    friend class cluster_client_test_t;
 };

src/client/cluster_client_list.cpp

@@ -295,7 +295,7 @@ int cluster_client_t::start_pg_listing(inode_list_pg_t *pg)
     bool conn = true;
     for (osd_num_t peer_osd: all_peers)
     {
-        if (msgr.osd_peers.find(peer_osd) == msgr.osd_peers.end())
+        if (msgr.osd_peer_fds.find(peer_osd) == msgr.osd_peer_fds.end())
         {
             // Initiate connection
             if (st_cli.peer_states[peer_osd].is_null())
@@ -340,7 +340,7 @@ void cluster_client_t::send_list(inode_list_osd_t *cur_list)
     osd_op_t *op = new osd_op_t();
     op->op_type = OSD_OP_OUT;
     // Already checked that it exists above, but anyway
-    op->client_id = msgr.osd_peers.at(cur_list->osd_num)->client_id;
+    op->peer_fd = msgr.osd_peer_fds.at(cur_list->osd_num);
     op->req = (osd_any_op_t){
         .sec_list = {
             .header = {

src/client/cluster_client_wb.cpp

@@ -88,11 +88,6 @@ void writeback_cache_t::copy_write(cluster_op_t *op, int state, uint64_t new_flu
     // ...or just save it for writeback if write buffering is enabled
     if (op->len == 0)
     {
-        // FIXME: OSD_OP_DELETEs are currently only sent by vitastor-cli rm/rm-data and
-        // actually have len=0, because delete is actually a delete of the full object
-        // containing the requested offset, not a "punch hole" operation. But here, writeback
-        // cache assumes it IS a "punch hole" operation. I should select one of these
-        // approaches and fix everything accordingly when I decide to implement TRIM.
         return;
     }
     auto dirty_it = find_dirty(op->inode, op->offset);
@@ -249,13 +244,12 @@ void writeback_cache_t::copy_write(cluster_op_t *op, int state, uint64_t new_flu
             writeback_queue_size--;
         }
     }
-    if (!is_del && op->len > 0)
+    if (!is_del)
     {
         uint64_t pos = 0, len = op->len, iov_idx = 0;
-        while (iov_idx < op->iov.count)
+        while (len > 0 && iov_idx < op->iov.count)
         {
             auto & iov = op->iov.buf[iov_idx];
-            assert(pos + iov.iov_len <= len);
             memcpy(buf + pos, iov.iov_base, iov.iov_len);
             pos += iov.iov_len;
             iov_idx++;
@@ -449,7 +443,7 @@ void writeback_cache_t::start_writebacks(cluster_client_t *cli, int count)
         started++;
         assert(writeback_queue_size > 0);
         writeback_queue_size--;
-        writeback_bytes -= (is_del ? 0 : off - from_it->first.stripe);
+        writeback_bytes -= off - from_it->first.stripe;
         assert(writeback_queue_size > 0 || !writeback_bytes);
         flush_buffers(cli, from_it, to_it);
     }

src/client/etcd_state_client.cpp

@@ -22,14 +22,19 @@ etcd_state_client_t::~etcd_state_client_t()
     stop_ws_keepalive();
     if (etcd_watch_ws)
     {
-        http_close(etcd_watch_ws);
+        http_destroy(etcd_watch_ws);
         etcd_watch_ws = NULL;
     }
     if (keepalive_client)
     {
-        http_close(keepalive_client);
+        http_destroy(keepalive_client);
         keepalive_client = NULL;
     }
+    if (http_ctx)
+    {
+        http_context_destroy(http_ctx);
+        http_ctx = NULL;
+    }
 #endif
     if (load_pgs_timer_id >= 0)
     {
@@ -72,10 +77,27 @@ std::vector<std::string> etcd_state_client_t::get_addresses()
     return addrs;
 }
 
+http_context_t *etcd_state_client_t::get_http_ctx()
+{
+    if (!http_ctx)
+    {
+        std::string error;
+        http_ctx = http_context_init(etcd_client_cert, etcd_client_key, etcd_ca, true, error);
+        if (!http_ctx)
+        {
+            fprintf(stderr, "Failed to initialize HTTP context: %s\n", error.c_str());
+            exit(1);
+        }
+    }
+    return http_ctx;
+}
+
 void etcd_state_client_t::etcd_call_oneshot(std::string etcd_address, std::string api, json11::Json payload,
     int timeout, std::function<void(std::string, json11::Json)> callback)
 {
     std::string etcd_api_path;
+    bool ssl = etcd_address.substr(0, 8) == "https://";
+    etcd_address = etcd_address.substr(ssl ? 8 : 7);
     int pos = etcd_address.find('/');
     if (pos >= 0)
     {
@@ -89,16 +111,16 @@ void etcd_state_client_t::etcd_call_oneshot(std::string etcd_address, std::strin
         "Content-Length: "+std::to_string(req.size())+"\r\n"
         "Connection: close\r\n"
         "\r\n"+req;
-    auto http_cli = http_init(tfd);
+    auto http_cli = http_init(tfd, get_http_ctx());
-    auto cb = [http_cli, callback](const http_response_t *response)
+    auto cb = [http_cli, callback](http_message_t *response)
     {
         std::string err;
         json11::Json data;
         response->parse_json_response(err, data);
         callback(err, data);
-        http_close(http_cli);
+        http_destroy(http_cli);
     };
-    http_request(http_cli, etcd_address, req, { .timeout = timeout }, cb);
+    http_request(http_cli, etcd_address, req, { .timeout = timeout, .ssl = ssl }, cb);
 }
 
 void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int timeout,
@@ -112,6 +134,8 @@ void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int t
     pick_next_etcd();
     std::string etcd_address = selected_etcd_address;
     std::string etcd_api_path;
+    bool ssl = etcd_address.substr(0, 8) == "https://";
+    etcd_address = etcd_address.substr(ssl ? 8 : 7);
     int pos = etcd_address.find('/');
     if (pos >= 0)
     {
@@ -128,7 +152,7 @@ void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int t
         "\r\n"+req;
     retries--;
     auto cb = [this, api, payload, timeout, retries, interval, callback,
-        cur_addr = selected_etcd_address](const http_response_t *response)
+        cur_addr = selected_etcd_address](http_message_t *response)
     {
         std::string err;
         json11::Json data;
@@ -164,22 +188,21 @@ void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int t
             callback(err, data);
     };
     if (!keepalive_client)
-    {
+        keepalive_client = http_init(tfd, get_http_ctx());
-        keepalive_client = http_init(tfd);
+    http_request(keepalive_client, etcd_address, req, { .timeout = timeout, .keepalive = true, .ssl = ssl }, cb);
-    }
-    http_request(keepalive_client, etcd_address, req, { .timeout = timeout, .keepalive = true }, cb);
 }
 
 void etcd_state_client_t::add_etcd_url(std::string addr)
 {
     if (addr.length() > 0)
     {
+        bool ssl = false;
         if (strtolower(addr.substr(0, 7)) == "http://")
             addr = addr.substr(7);
         else if (strtolower(addr.substr(0, 8)) == "https://")
         {
-            fprintf(stderr, "HTTPS is unsupported for etcd. Either use plain HTTP or setup a local proxy for etcd interaction\n");
+            addr = addr.substr(8);
-            exit(1);
+            ssl = true;
         }
         if (!local_ips.size())
             local_ips = getifaddr_list(std::vector<addr_mask_t>(), true);
@@ -194,6 +217,7 @@ void etcd_state_client_t::add_etcd_url(std::string addr)
             check_addr = addr;
         if (pos == std::string::npos)
             addr += "/v3";
+        addr = (ssl ? "https://" : "http://") + addr;
         bool local = false;
         int i;
         for (i = 0; i < local_ips.size(); i++)
@@ -239,6 +263,9 @@ void etcd_state_client_t::parse_config(const json11::Json & config)
             add_etcd_url(ea.string_value());
         }
     }
+    this->etcd_client_cert = config["etcd_client_cert"].string_value();
+    this->etcd_client_key = config["etcd_client_key"].string_value();
+    this->etcd_ca = config["etcd_ca"].string_value();
     this->etcd_prefix = config["etcd_prefix"].string_value();
     if (this->etcd_prefix == "")
     {
@@ -331,6 +358,8 @@ void etcd_state_client_t::start_etcd_watcher()
     pick_next_etcd();
     std::string etcd_address = selected_etcd_address;
     std::string etcd_api_path;
+    bool ssl = etcd_address.substr(0, 8) == "https://";
+    etcd_address = etcd_address.substr(ssl ? 8 : 7);
     int pos = etcd_address.find('/');
     if (pos >= 0)
     {
@@ -339,18 +368,17 @@ void etcd_state_client_t::start_etcd_watcher()
     }
     etcd_watches_initialised = 0;
     ws_alive = 1;
-    if (etcd_watch_ws)
-    {
-        http_close(etcd_watch_ws);
-        etcd_watch_ws = NULL;
-    }
     if (this->log_level > 1)
     {
         fprintf(stderr, "Trying to connect to etcd websocket at %s, watch from revision %ju/%ju/%ju\n", etcd_address.c_str(),
             etcd_watch_revision_config, etcd_watch_revision_osd, etcd_watch_revision_pg);
     }
-    etcd_watch_ws = open_websocket(tfd, etcd_address, etcd_api_path+"/watch", etcd_slow_timeout,
+    if (!etcd_watch_ws)
-        [this, cur_addr = selected_etcd_address](const http_response_t *msg)
+        etcd_watch_ws = http_init(tfd, get_http_ctx());
+    else
+        http_close(etcd_watch_ws);
+    open_websocket(etcd_watch_ws, etcd_address, etcd_api_path+"/watch", { .timeout = etcd_slow_timeout, .ssl = ssl },
+        [this, cur_addr = selected_etcd_address](http_message_t *msg)
     {
         if (msg->body.length())
         {
@@ -393,7 +421,6 @@ void etcd_state_client_t::start_etcd_watcher()
                                 fprintf(stderr, "Revisions before %ju were compacted by etcd, reloading state\n",
                                     data["result"]["compact_revision"].uint64_value());
                                 http_close(etcd_watch_ws);
-                                etcd_watch_ws = NULL;
                                 etcd_watch_revision_config = etcd_watch_revision_osd = etcd_watch_revision_pg = 0;
                                 on_reload_hook();
                             }
@@ -414,10 +441,7 @@ void etcd_state_client_t::start_etcd_watcher()
                 }
                 // Save revision only if it's present in the message - because sometimes etcd sends something without a header, like:
                 // {"error": {"grpc_code": 14, "http_code": 503, "http_status": "Service Unavailable", "message": "error reading from server: EOF"}}
-                // Also don't save revision from the initial created: true messages because they always contain the latest revision
+                if (etcd_watches_initialised == ETCD_TOTAL_WATCHES && !data["result"]["header"]["revision"].is_null())
-                if (etcd_watches_initialised == ETCD_TOTAL_WATCHES &&
-                    !data["result"]["header"]["revision"].is_null() &&
-                    !data["result"]["created"].bool_value())
                 {
                     // Restart watchers from the same revision number as in the last received message,
                     // not from the next one to protect against revision being split into multiple messages,
@@ -470,11 +494,6 @@ void etcd_state_client_t::start_etcd_watcher()
             fprintf(stderr, "Disconnected from etcd %s\n", cur_addr.c_str());
             if (cur_addr == selected_etcd_address)
                 selected_etcd_address = "";
-            if (etcd_watch_ws)
-            {
-                http_close(etcd_watch_ws);
-                etcd_watch_ws = NULL;
-            }
             if (etcd_watches_initialised == 0)
             {
                 // Connection not established, retry in <etcd_quick_timeout>
@@ -551,11 +570,6 @@ void etcd_state_client_t::start_ws_keepalive()
                 {
                     fprintf(stderr, "Websocket ping failed, disconnecting from etcd %s\n", selected_etcd_address.c_str());
                 }
-                if (etcd_watch_ws)
-                {
-                    http_close(etcd_watch_ws);
-                    etcd_watch_ws = NULL;
-                }
                 start_etcd_watcher();
             }
             else
@@ -1185,6 +1199,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 if (i >= pg_state_bit_count)
                 {
                     fprintf(stderr, "Unexpected pool %u PG %u state keyword in etcd: %s\n", pool_id, pg_num, e.dump().c_str());
+                    return;
                 }
             }
             if (!cur_primary || !value["state"].is_array() || !state ||
@@ -1193,6 +1208,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 (state & PG_INCOMPLETE) && state != PG_INCOMPLETE && state != (PG_INCOMPLETE|PG_HAS_INVALID))
             {
                 fprintf(stderr, "Unexpected pool %u PG %u state in etcd: primary=%ju, state=%s\n", pool_id, pg_num, cur_primary, value["state"].dump().c_str());
+                return;
             }
             pg_cfg.cur_primary = cur_primary;
             pg_cfg.cur_state = state;
@@ -1280,6 +1296,16 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                     else
                         parent_inode_num |= parent_pool_id << (64-POOL_ID_BITS);
                 }
+                std::shared_ptr<inode_enc_t> enc;
+                if (!value["enc_key"].string_value().empty())
+                {
+                    std::vector<uint8_t> k = hexdecode(value["enc_key"].string_value());
+                    if (k.size() == 512/8)
+                    {
+                        enc = std::make_shared<inode_enc_t>();
+                        enc->key = std::move(k);
+                    }
+                }
                 insert_inode_config((inode_config_t){
                     .num = inode_num,
                     .name = value["name"].string_value(),
@@ -1287,6 +1313,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                     .parent_id = parent_inode_num,
                     .readonly = value["readonly"].bool_value(),
                     .deleted = value["deleted"].bool_value(),
+                    .enc = enc,
                     .meta = value["meta"],
                     .mod_revision = kv.mod_revision,
                 });

src/client/etcd_state_client.h

@@ -4,9 +4,10 @@
 #pragma once
 
 #include <set>
+#include <memory>
 
 #include "json11/json11.hpp"
-#include "object_id.h"
+#include "osd_id.h"
 #include "timerfd_manager.h"
 
 #define ETCD_CONFIG_WATCH_ID 1
@@ -75,6 +76,14 @@ struct pool_config_t
     void *reshard_state = NULL;
 };
 
+struct inode_enc_t
+{
+    int refs = 0;
+    std::vector<uint8_t> key;
+    // FIXME It may also contain snapshot chain and key information
+    uint32_t bitmap_granularity = 0;
+};
+
 struct inode_config_t
 {
     uint64_t num = 0;
@@ -83,6 +92,7 @@ struct inode_config_t
     inode_t parent_id = 0;
     bool readonly = false;
     bool deleted = false;
+    std::shared_ptr<inode_enc_t> enc;
     // Arbitrary metadata
     json11::Json meta;
     // Change revision of the metadata in etcd
@@ -96,6 +106,7 @@ struct inode_watch_t
 };
 
 struct http_co_t;
+struct http_context_t;
 
 struct __attribute__((visibility("default"))) etcd_state_client_t
 {
@@ -125,9 +136,13 @@ public:
     uint32_t global_immediate_commit = IMMEDIATE_NONE;
 
     std::string etcd_prefix;
+    std::string etcd_client_cert;
+    std::string etcd_client_key;
+    std::string etcd_ca;
     int log_level = 0;
     timerfd_manager_t *tfd = NULL;
 
+    http_context_t *http_ctx = NULL;
     http_co_t *etcd_watch_ws = NULL, *keepalive_client = NULL;
     int etcd_watches_initialised = 0;
     uint64_t etcd_watch_revision_config = 0;
@@ -160,6 +175,7 @@ public:
     json11::Json::object serialize_inode_cfg(inode_config_t *cfg);
     etcd_kv_t parse_etcd_kv(const json11::Json & kv_json);
     std::vector<std::string> get_addresses();
+    http_context_t *get_http_ctx();
     void etcd_call_oneshot(std::string etcd_address, std::string api, json11::Json payload, int timeout, std::function<void(std::string, json11::Json)> callback);
     void etcd_call(std::string api, json11::Json payload, int timeout, int retries, int interval, std::function<void(std::string, json11::Json)> callback);
     void etcd_txn(json11::Json txn, int timeout, int retries, int interval, std::function<void(std::string, json11::Json)> callback);

src/client/http_client.cpp

@@ -10,9 +10,17 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include <string.h>
+#include <assert.h>
 
 #include <stdexcept>
 
+#ifdef WITH_OPENSSL
+#include <openssl/bio.h>
+#include <openssl/err.h>
+#include <openssl/pem.h>
+#include <openssl/ssl.h>
+#endif
+
 #include "addr_util.h"
 #include "str_util.h"
 #include "json_util.h"
@@ -24,14 +32,51 @@
 
 static std::string ws_format_frame(int type, uint64_t size);
 static bool ws_parse_frame(std::string & buf, uint8_t & type, std::string & res);
-static void parse_http_headers(std::string & res, http_response_t *parsed);
+static void parse_http_headers(std::string & res, http_message_t *parsed, bool is_request);
+
+struct http_context_t
+{
+    std::string ssl_cert;
+    std::string ssl_key;
+    std::string ssl_ca;
+
+#ifdef WITH_OPENSSL
+    SSL_CTX *ssl_ctx = NULL;
+#endif
+
+    ~http_context_t()
+    {
+#ifdef WITH_OPENSSL
+        if (ssl_ctx)
+        {
+            SSL_CTX_free(ssl_ctx);
+            ssl_ctx = NULL;
+        }
+#endif
+    }
+};
+
+struct http_call_t
+{
+    std::string host;
+    std::string request;
+    http_options_t options;
+    std::function<void(http_message_t *)> cb;
+};
 
 struct http_co_t
 {
+    http_context_t *ctx = NULL;
+#ifdef WITH_OPENSSL
+    SSL *ssl_cli = NULL;
+    BIO *ssl_bio = NULL;
+#endif
+
     timerfd_manager_t *tfd;
-    std::function<void(const http_response_t*)> response_callback;
+    std::function<void(http_message_t*)> response_callback;
 
     int request_timeout = 0;
+    bool ssl = false;
     std::string host;
     std::string request;
     std::string ws_outbox;
@@ -39,7 +84,7 @@ struct http_co_t
     bool want_streaming;
     bool keepalive;
 
-    std::vector<std::function<void()>> keepalive_queue;
+    std::vector<http_call_t> keepalive_queue;
 
     int state = 0;
     std::string connected_host;
@@ -47,10 +92,10 @@ struct http_co_t
     int timeout_id = -1;
     int epoll_events = 0;
     int sent = 0;
-    std::vector<char> rbuf;
+    std::vector<uint8_t> rbuf;
     iovec read_iov, send_iov;
     msghdr read_msg = { 0 }, send_msg = { 0 };
-    http_response_t parsed;
+    http_message_t parsed;
     uint64_t target_response_size = 0;
 
     int onstack = 0;
@@ -70,9 +115,14 @@ struct http_co_t
     void submit_read(bool check_timeout);
     void submit_send();
     bool handle_read();
+#ifdef WITH_OPENSSL
+    bool do_ssl_handshake(bool init_send);
+    void on_ssl_error(int res);
+#endif
     void post_message(uint8_t type, const std::string & msg);
+    void reply(const std::string & msg);
     void send_request(const std::string & host, const std::string & request,
-        const http_options_t & options, std::function<void(const http_response_t *response)> response_callback);
+        const http_options_t & options, std::function<void(http_message_t *response)> response_callback);
 };
 
 #define HTTP_CO_CLOSED 0
@@ -83,20 +133,64 @@ struct http_co_t
 #define HTTP_CO_WEBSOCKET 5
 #define HTTP_CO_CHUNKED 6
 #define HTTP_CO_KEEPALIVE 7
+#define HTTP_CO_SERVER 8
+#define HTTP_CO_REQ_HDR_RECEIVED 9
+#define HTTP_CO_REQUEST_RECEIVED 10
 
 #define DEFAULT_TIMEOUT 5000
 
-http_co_t *http_init(timerfd_manager_t *tfd)
+http_context_t* http_context_init(const std::string & ssl_cert, const std::string & ssl_key,
+    const std::string & ssl_ca, bool verify_peer, std::string & error)
+{
+    http_context_t *ctx = new http_context_t;
+#ifdef WITH_OPENSSL
+    SSL_CTX *ssl_ctx = SSL_CTX_new(TLS_method());
+    ctx->ssl_cert = ssl_cert;
+    ctx->ssl_key = ssl_key;
+    ctx->ssl_ca = ssl_ca;
+    ctx->ssl_ctx = ssl_ctx;
+    if (!ssl_ctx)
+        goto init_err;
+    SSL_CTX_set_verify(ssl_ctx, verify_peer ? SSL_VERIFY_PEER : SSL_VERIFY_NONE, NULL);
+    if (!SSL_CTX_set_min_proto_version(ssl_ctx, TLS1_2_VERSION))
+        goto init_err;
+    if ((ssl_ca != "")
+        ? !SSL_CTX_load_verify_locations(ssl_ctx, ssl_ca.c_str(), NULL)
+        : !SSL_CTX_set_default_verify_paths(ssl_ctx))
+        goto init_err;
+    if (ssl_cert != "" && ssl_key != "" &&
+        (!SSL_CTX_use_certificate_file(ssl_ctx, ssl_cert.c_str(), SSL_FILETYPE_PEM) ||
+        !SSL_CTX_use_PrivateKey_file(ssl_ctx, ssl_key.c_str(), SSL_FILETYPE_PEM)))
+        goto init_err;
+#endif
+    return ctx;
+init_err:
+    error = std::string("openssl initialization failed: ")+ERR_error_string(ERR_get_error(), NULL);
+    delete ctx;
+    return NULL;
+}
+
+void http_context_destroy(http_context_t *ctx)
+{
+    delete ctx;
+}
+
+http_co_t *http_init(timerfd_manager_t *tfd, http_context_t *ctx)
 {
     http_co_t *handler = new http_co_t();
     handler->tfd = tfd;
     handler->state = HTTP_CO_CLOSED;
+    handler->ctx = ctx;
     return handler;
 }
 
-http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
+void open_websocket(http_co_t *handler, const std::string & host, const std::string & path,
-    int timeout, std::function<void(const http_response_t *msg)> response_callback)
+    const http_options_t & options, std::function<void(http_message_t *msg)> response_callback)
 {
+    if (handler->state == HTTP_CO_KEEPALIVE && (handler->connected_host != host || handler->ssl != options.ssl))
+        handler->close_connection();
+    if (handler->state != HTTP_CO_KEEPALIVE && handler->state != HTTP_CO_CLOSED)
+        throw std::runtime_error("Attempt to open websocket on a keepalive stream");
     std::string request = "GET "+path+" HTTP/1.1\r\n"
         "Host: "+host+"\r\n"
         "Upgrade: websocket\r\n"
@@ -104,29 +198,54 @@ http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, cons
         "Sec-WebSocket-Key: x3JJHMbDL1EzLkh9GBhXDw==\r\n"
         "Sec-WebSocket-Version: 13\r\n"
         "\r\n";
-    http_co_t *handler = new http_co_t();
-    handler->tfd = tfd;
-    handler->state = HTTP_CO_CLOSED;
     handler->host = host;
-    handler->request_timeout = timeout < 0 ? -1 : (timeout == 0 ? DEFAULT_TIMEOUT : timeout);
+    handler->request_timeout = options.timeout < 0 ? -1 : (options.timeout == 0 ? DEFAULT_TIMEOUT : options.timeout);
     handler->want_streaming = false;
     handler->keepalive = false;
+    handler->ssl = options.ssl;
     handler->request = request;
     handler->response_callback = response_callback;
+    handler->ws_outbox = "";
+    handler->response = "";
+    handler->sent = 0;
+    handler->parsed = {};
     handler->start_ws_connection();
-    return handler;
 }
 
 void http_request(http_co_t *handler, const std::string & host, const std::string & request,
-    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback)
+    const http_options_t & options, std::function<void(http_message_t *response)> response_callback)
 {
     handler->send_request(host, request, options, response_callback);
 }
 
+void http_serve(http_co_t *handler, int peer_fd, const http_options_t & options, std::function<void(http_message_t *msg)> request_callback)
+{
+    if (handler->state != HTTP_CO_SERVER || handler->peer_fd != peer_fd)
+        handler->close_connection();
+    handler->host = "";
+    handler->request_timeout = options.timeout < 0 ? -1 : (options.timeout == 0 ? DEFAULT_TIMEOUT : options.timeout);
+    handler->want_streaming = false;
+    handler->keepalive = false;
+    handler->ssl = options.ssl;
+    handler->request = "";
+    handler->response_callback = request_callback;
+    handler->ws_outbox = "";
+    handler->response = "";
+    handler->sent = 0;
+    handler->parsed = {};
+    handler->peer_fd = peer_fd;
+    handler->state = HTTP_CO_SERVER;
+    handler->tfd->set_fd_handler(peer_fd, false, [handler](int peer_fd, int epoll_events)
+    {
+        handler->epoll_events |= epoll_events;
+        handler->handle_events();
+    });
+}
+
 void http_co_t::run_cb_and_clear()
 {
     parsed.eof = true;
-    std::function<void(const http_response_t*)> cb;
+    std::function<void(http_message_t*)> cb;
     cb.swap(response_callback);
     // Call callback after clearing it because otherwise we may hit reenterability problems
     if (cb != NULL)
@@ -135,7 +254,7 @@ void http_co_t::run_cb_and_clear()
 }
 
 void http_co_t::send_request(const std::string & host, const std::string & request,
-    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback)
+    const http_options_t & options, std::function<void(http_message_t *response)> response_callback)
 {
     stackin();
     if (state == HTTP_CO_WEBSOCKET)
@@ -145,20 +264,18 @@ void http_co_t::send_request(const std::string & host, const std::string & reque
     }
     else if (state != HTTP_CO_KEEPALIVE && state != HTTP_CO_CLOSED)
     {
-        keepalive_queue.push_back([this, host, request, options, response_callback]()
+        keepalive_queue.emplace_back((http_call_t){ host, request, options, std::move(response_callback) });
-        {
-            this->send_request(host, request, options, response_callback);
-        });
         stackout();
         return;
     }
-    if (state == HTTP_CO_KEEPALIVE && connected_host != host)
+    if (state == HTTP_CO_KEEPALIVE && (connected_host != host || ssl != options.ssl))
     {
         close_connection();
     }
     this->request_timeout = options.timeout < 0 ? 0 : (options.timeout == 0 ? DEFAULT_TIMEOUT : options.timeout);
     this->want_streaming = options.want_streaming;
     this->keepalive = options.keepalive;
+    this->ssl = options.ssl;
     this->host = host;
     this->request = request;
     this->response = "";
@@ -190,7 +307,7 @@ void http_co_t::send_request(const std::string & host, const std::string & reque
             else
             {
                 close_connection();
-                parsed = { .error = "HTTP request timed out" };
+                parsed = { .error = "HTTP request timed out", .status_code = ETIMEDOUT };
                 run_cb_and_clear();
             }
             stackout();
@@ -204,6 +321,11 @@ void http_post_message(http_co_t *handler, uint8_t type, const std::string & msg
     handler->post_message(type, msg);
 }
 
+void http_reply(http_co_t *handler, const std::string & reply)
+{
+    handler->reply(reply);
+}
+
 void http_co_t::post_message(uint8_t type, const std::string & msg)
 {
     stackin();
@@ -213,7 +335,8 @@ void http_co_t::post_message(uint8_t type, const std::string & msg)
         request += msg;
         submit_send();
     }
-    else if (state == HTTP_CO_KEEPALIVE || state == HTTP_CO_CHUNKED)
+    else if (state == HTTP_CO_KEEPALIVE || state == HTTP_CO_CHUNKED ||
+        state == HTTP_CO_SERVER || state == HTTP_CO_REQ_HDR_RECEIVED || state == HTTP_CO_REQUEST_RECEIVED)
     {
         throw std::runtime_error("Attempt to send websocket message on a regular HTTP connection");
     }
@@ -225,12 +348,29 @@ void http_co_t::post_message(uint8_t type, const std::string & msg)
     stackout();
 }
 
-void http_close(http_co_t *handler)
+void http_co_t::reply(const std::string & reply)
+{
+    stackin();
+    if (state != HTTP_CO_REQUEST_RECEIVED)
+    {
+        throw std::runtime_error("Attempt to send HTTP response in invalid connection state");
+    }
+    request += reply;
+    submit_send();
+    stackout();
+}
+
+void http_destroy(http_co_t *handler)
 {
     handler->end();
 }
 
-void http_response_t::parse_json_response(std::string & error, json11::Json & r) const
+void http_close(http_co_t *handler)
+{
+    handler->close_connection();
+}
+
+void http_message_t::parse_json_response(std::string & error, json11::Json & r) const
 {
     if (this->error != "")
     {
@@ -277,6 +417,15 @@ void http_co_t::close_connection()
         close(peer_fd);
         peer_fd = -1;
     }
+#ifdef WITH_OPENSSL
+    if (ssl_cli)
+    {
+        // Frees client and bios at once
+        SSL_free(ssl_cli);
+        ssl_cli = NULL;
+    }
+    ssl_bio = NULL;
+#endif
     state = HTTP_CO_CLOSED;
     connected_host = "";
     response = "";
@@ -295,7 +444,7 @@ void http_co_t::start_ws_connection()
             if (state != HTTP_CO_WEBSOCKET)
             {
                 close_connection();
-                parsed = { .error = "Websocket connection timed out" };
+                parsed = { .error = "Websocket connection timed out", .status_code = ETIMEDOUT };
                 run_cb_and_clear();
             }
             stackout();
@@ -311,7 +460,7 @@ void http_co_t::start_connection()
     if (!string_to_addr(host.c_str(), 1, 80, &addr))
     {
         close_connection();
-        parsed = { .error = "Invalid address: "+host };
+        parsed = { .error = "Invalid address: "+host, .status_code = EINVAL };
         run_cb_and_clear();
         stackout();
         return;
@@ -320,19 +469,56 @@ void http_co_t::start_connection()
     if (peer_fd < 0)
     {
         close_connection();
-        parsed = { .error = std::string("socket: ")+strerror(errno) };
+        parsed = { .error = std::string("socket: ")+strerror(errno), .status_code = errno };
         run_cb_and_clear();
         stackout();
         return;
     }
     fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
     epoll_events = 0;
+#ifdef WITH_OPENSSL
+    // https://wiki.openssl.org/index.php/Hostname_validation
+    if (ssl)
+    {
+        if (!ctx)
+            goto init_err;
+        ssl_bio = BIO_new(BIO_s_socket());
+        if (!ssl_bio)
+            goto init_err;
+        if (!BIO_set_fd(ssl_bio, peer_fd, BIO_NOCLOSE))
+            goto init_err;
+        ssl_cli = SSL_new(ctx->ssl_ctx);
+        if (!ssl_cli)
+            goto init_err;
+        SSL_set_bio(ssl_cli, ssl_bio, ssl_bio);
+        if (!SSL_set_tlsext_host_name(ssl_cli, host.c_str()))
+        {
+init_err:
+            if (ssl_cli)
+            {
+                SSL_free(ssl_cli);
+                ssl_cli = NULL;
+            }
+            else if (ssl_bio)
+            {
+                BIO_free(ssl_bio);
+                ssl_bio = NULL;
+            }
+            parsed = { .error = std::string("openssl initialization failed: ")+ERR_error_string(ERR_get_error(), NULL) };
+            response_callback(&parsed);
+            response_callback = NULL;
+            stackout();
+            return;
+        }
+        SSL_set_connect_state(ssl_cli);
+    }
+#endif
     // Finally call connect
     int r = ::connect(peer_fd, (sockaddr*)&addr, sizeof(addr));
     if (r < 0 && errno != EINPROGRESS)
     {
         close_connection();
-        parsed = { .error = std::string("connect: ")+strerror(errno) };
+        parsed = { .error = std::string("connect: ")+strerror(errno), .status_code = errno };
         run_cb_and_clear();
         stackout();
         return;
@@ -367,6 +553,8 @@ void http_co_t::handle_events()
             {
                 if (state == HTTP_CO_HEADERS_RECEIVED)
                     std::swap(parsed.body, response);
+                else if (state == HTTP_CO_SERVER || state == HTTP_CO_REQ_HDR_RECEIVED || state == HTTP_CO_REQUEST_RECEIVED)
+                    parsed = { .error = "client has disconnected normally" };
                 close_connection();
                 run_cb_and_clear();
                 break;
@@ -388,7 +576,7 @@ void http_co_t::handle_connect_result()
     if (result != 0)
     {
         close_connection();
-        parsed = { .error = std::string("connect: ")+strerror(result) };
+        parsed = { .error = std::string("connect: ")+strerror(result), .status_code = result };
         run_cb_and_clear();
         stackout();
         return;
@@ -408,18 +596,44 @@ void http_co_t::handle_connect_result()
 void http_co_t::submit_send()
 {
     stackin();
-    int res;
+    ssize_t res = 0;
 again:
     if (sent < request.size())
     {
-        send_iov = (iovec){ .iov_base = (void*)(request.c_str()+sent), .iov_len = request.size()-sent };
+        send_iov = (iovec){ .iov_base = (void*)(request.data()+sent), .iov_len = request.size()-sent };
-        send_msg.msg_iov = &send_iov;
+#ifdef WITH_OPENSSL
-        send_msg.msg_iovlen = 1;
+        if (!ssl)
-        res = sendmsg(peer_fd, &send_msg, MSG_NOSIGNAL);
+#endif
-        if (res < 0)
         {
-            res = -errno;
+            send_msg.msg_iov = &send_iov;
+            send_msg.msg_iovlen = 1;
+            res = sendmsg(peer_fd, &send_msg, MSG_NOSIGNAL);
+            if (res < 0)
+                res = -errno;
         }
+#ifdef WITH_OPENSSL
+        else
+        {
+            if (!do_ssl_handshake(false))
+                goto out;
+            int ok = SSL_write_ex(ssl_cli, send_iov.iov_base, send_iov.iov_len, (size_t*)&res);
+            if (!ok)
+            {
+                res = SSL_get_error(ssl_cli, ok);
+                if (res == SSL_ERROR_WANT_WRITE || res == 0)
+                    res = 0;
+                else if (res == SSL_ERROR_WANT_READ)
+                    goto out;
+                else if (res == SSL_ERROR_SYSCALL)
+                    res = -errno;
+                else
+                {
+                    on_ssl_error(res);
+                    goto out;
+                }
+            }
+        }
+#endif
         if (res == -EAGAIN || res == -EINTR)
         {
             res = 0;
@@ -427,13 +641,33 @@ again:
         else if (res < 0)
         {
             close_connection();
-            parsed = { .error = std::string("sendmsg: ")+strerror(errno) };
+            parsed = { .error = std::string("sendmsg: ")+strerror(errno), .status_code = errno };
             run_cb_and_clear();
             stackout();
             return;
         }
         sent += res;
-        if (state == HTTP_CO_SENDING_REQUEST)
+        if (state == HTTP_CO_REQUEST_RECEIVED)
+        {
+            if (sent >= request.size())
+            {
+                if (!keepalive)
+                {
+                    close_connection();
+                    parsed = { .error = "connection is not keep-alive" };
+                    run_cb_and_clear();
+                    stackout();
+                    return;
+                }
+                state = HTTP_CO_SERVER;
+                request = "";
+                sent = 0;
+            }
+            else
+                goto again;
+            handle_read();
+        }
+        else if (state == HTTP_CO_SENDING_REQUEST)
         {
             if (sent >= request.size())
                 state = HTTP_CO_REQUEST_SENT;
@@ -447,26 +681,53 @@ again:
             goto again;
         }
     }
+out:
     stackout();
 }
 
 void http_co_t::submit_read(bool check_timeout)
 {
     stackin();
-    int res;
+    ssize_t res = 0;
 again:
     if (rbuf.size() != READ_BUFFER_SIZE)
     {
         rbuf.resize(READ_BUFFER_SIZE);
     }
     read_iov = { .iov_base = rbuf.data(), .iov_len = READ_BUFFER_SIZE };
-    read_msg.msg_iov = &read_iov;
+#ifdef WITH_OPENSSL
-    read_msg.msg_iovlen = 1;
+    if (!ssl)
-    res = recvmsg(peer_fd, &read_msg, 0);
+#endif
-    if (res < 0)
     {
-        res = -errno;
+        read_msg.msg_iov = &read_iov;
+        read_msg.msg_iovlen = 1;
+        res = recvmsg(peer_fd, &read_msg, 0);
+        if (res < 0)
+            res = -errno;
     }
+#ifdef WITH_OPENSSL
+    else
+    {
+        if (!do_ssl_handshake(true))
+            goto out;
+        int ok = SSL_read_ex(ssl_cli, read_iov.iov_base, read_iov.iov_len, (size_t*)&res);
+        if (!ok)
+        {
+            res = SSL_get_error(ssl_cli, ok);
+            if (res == SSL_ERROR_WANT_READ)
+                res = -EAGAIN;
+            else if (res == SSL_ERROR_SYSCALL)
+                res = -errno;
+            else if (res == SSL_ERROR_ZERO_RETURN)
+                res = 0;
+            else
+            {
+                on_ssl_error(res);
+                goto out;
+            }
+        }
+    }
+#endif
     if (res == -EAGAIN || res == -EINTR)
     {
         if (check_timeout)
@@ -477,7 +738,7 @@ again:
             {
                 // Timeout happened and there is no data to read
                 close_connection();
-                parsed = { .error = "HTTP request timed out" };
+                parsed = { .error = "HTTP request timed out", .status_code = ETIMEDOUT };
                 run_cb_and_clear();
             }
         }
@@ -492,23 +753,76 @@ again:
         epoll_events = epoll_events & ~EPOLLIN;
         if (state == HTTP_CO_HEADERS_RECEIVED)
             std::swap(parsed.body, response);
-        close_connection();
         if (res < 0)
-            parsed = { .error = std::string("recvmsg: ")+strerror(-res) };
+            parsed = { .error = std::string("recvmsg: ")+strerror(-res), .status_code = (int)-res };
+        else if (state == HTTP_CO_SERVER || state == HTTP_CO_REQ_HDR_RECEIVED || state == HTTP_CO_REQUEST_RECEIVED)
+            parsed = { .error = "client has disconnected normally" };
+        close_connection();
         run_cb_and_clear();
     }
     else
     {
-        response += std::string(rbuf.data(), res);
+        response += std::string((char*)rbuf.data(), res);
         handle_read();
     }
+out:
     stackout();
 }
 
+#ifdef WITH_OPENSSL
+void http_co_t::on_ssl_error(int res)
+{
+    close_connection();
+    if (res == SSL_ERROR_ZERO_RETURN)
+    {
+        // Client closed the connection
+        parsed = { .error = "peer closed the SSL connection" };
+    }
+    else
+        parsed = { .error = std::string("SSL error: ")+ERR_error_string(ERR_get_error(), NULL), .status_code = EIO };
+    run_cb_and_clear();
+}
+
+bool http_co_t::do_ssl_handshake(bool init_send)
+{
+    if (SSL_is_init_finished(ssl_cli))
+        return true;
+    int r;
+    while (1)
+    {
+        r = SSL_do_handshake(ssl_cli);
+        if (r > 0)
+        {
+            // OK
+            if (init_send)
+                submit_send();
+            return true;
+        }
+        r = SSL_get_error(ssl_cli, r);
+        if (r == SSL_ERROR_WANT_READ)
+        {
+            break;
+        }
+        else
+        {
+            int errcode = ERR_get_error();
+            parsed = { .error = ERR_error_string(errcode, NULL), .status_code = EIO };
+            close_connection();
+            run_cb_and_clear();
+            return false;
+        }
+    }
+    return false;
+}
+#endif
+
 bool http_co_t::handle_read()
 {
     stackin();
-    if (state == HTTP_CO_REQUEST_SENT)
+    if (state == HTTP_CO_REQUEST_RECEIVED)
+    {
+    }
+    else if (state == HTTP_CO_REQUEST_SENT)
     {
         int pos = response.find("\r\n\r\n");
         if (pos >= 0)
@@ -520,7 +834,7 @@ bool http_co_t::handle_read()
                 timeout_id = -1;
             }
             state = HTTP_CO_HEADERS_RECEIVED;
-            parse_http_headers(response, &parsed);
+            parse_http_headers(response, &parsed, false);
             if (parsed.status_code == 101 &&
                 parsed.headers.find("sec-websocket-accept") != parsed.headers.end() &&
                 parsed.headers["upgrade"] == "websocket" &&
@@ -544,7 +858,7 @@ bool http_co_t::handle_read()
                 {
                     // Sorry, unsupported response
                     close_connection();
-                    parsed = { .error = "Response has neither Connection: close, nor Transfer-Encoding: chunked nor Content-Length headers" };
+                    parsed = { .error = "Response has neither Connection: close, nor Transfer-Encoding: chunked nor Content-Length headers", .status_code = EINVAL };
                     run_cb_and_clear();
                     stackout();
                     return false;
@@ -556,14 +870,57 @@ bool http_co_t::handle_read()
             }
         }
     }
-    if (state == HTTP_CO_HEADERS_RECEIVED && target_response_size > 0 && response.size() >= target_response_size)
+    else if (state == HTTP_CO_SERVER)
+    {
+        int pos = response.find("\r\n\r\n");
+        if (pos >= 0)
+        {
+            if (timeout_id >= 0)
+            {
+                // Timeout is cleared when headers are received
+                tfd->clear_timer(timeout_id);
+                timeout_id = -1;
+            }
+            state = HTTP_CO_REQ_HDR_RECEIVED;
+            parse_http_headers(response, &parsed, true);
+            auto conn_it = parsed.headers.find("connection");
+            keepalive = (conn_it != parsed.headers.end() && conn_it->second == "keep-alive");
+            auto enc_it = parsed.headers.find("transfer-encoding");
+            if (enc_it != parsed.headers.end())
+            {
+                // Sorry, unsupported request
+                close_connection();
+                parsed = { .error = "Chunked requests are not supported", .status_code = EINVAL };
+                run_cb_and_clear();
+                stackout();
+                return false;
+            }
+            auto len_it = parsed.headers.find("content-length");
+            target_response_size = stoull_full(len_it != parsed.headers.end() ? len_it->second : "");
+            if (!target_response_size)
+            {
+                state = HTTP_CO_REQUEST_RECEIVED;
+                response_callback(&parsed);
+            }
+        }
+    }
+    if ((state == HTTP_CO_HEADERS_RECEIVED || state == HTTP_CO_REQ_HDR_RECEIVED) &&
+        target_response_size > 0 && response.size() >= target_response_size)
     {
         std::swap(parsed.body, response);
-        if (!keepalive)
+        if (state == HTTP_CO_REQ_HDR_RECEIVED)
-            close_connection();
+        {
+            state = HTTP_CO_REQUEST_RECEIVED;
+            response_callback(&parsed);
+        }
         else
-            state = HTTP_CO_KEEPALIVE;
+        {
-        run_cb_and_clear();
+            if (!keepalive)
+                close_connection();
+            else
+                state = HTTP_CO_KEEPALIVE;
+            run_cb_and_clear();
+        }
     }
     else if (state == HTTP_CO_CHUNKED && response.size() > 0)
     {
@@ -626,26 +983,34 @@ void http_co_t::next_request()
 {
     if (keepalive_queue.size() > 0)
     {
-        auto next = keepalive_queue[0];
+        auto next = std::move(keepalive_queue[0]);
-        keepalive_queue.erase(keepalive_queue.begin(), keepalive_queue.begin()+1);
+        keepalive_queue.erase(keepalive_queue.begin());
-        next();
+        send_request(next.host, next.request, next.options, next.cb);
     }
 }
 
-static void parse_http_headers(std::string & res, http_response_t *parsed)
+static void parse_http_headers(std::string & res, http_message_t *parsed, bool is_request)
 {
     int pos = res.find("\r\n");
     pos = pos < 0 ? res.length() : pos+2;
     std::string status_line = res.substr(0, pos);
     int http_version;
     char *status_text = NULL;
-    sscanf(status_line.c_str(), "HTTP/1.%d %d %ms", &http_version, &parsed->status_code, &status_text);
+    if (!is_request)
-    if (status_text)
     {
-        parsed->status_line = status_text;
+        sscanf(status_line.c_str(), "HTTP/1.%d %d %ms", &http_version, &parsed->status_code, &status_text);
-        // %ms = allocate a buffer
+        if (status_text)
-        free(status_text);
+        {
-        status_text = NULL;
+            parsed->status_line = status_text;
+            // %ms = allocate a buffer
+            free(status_text);
+            status_text = NULL;
+        }
+    }
+    else
+    {
+        // Should be GET/POST / HTTP/1.1
+        parsed->status_line = status_line;
     }
     int prev = pos;
     while ((pos = res.find("\r\n", prev)) >= prev)

src/client/http_client.h

@@ -17,14 +17,19 @@
 
 class timerfd_manager_t;
 
+#pragma GCC visibility push(default)
+
 struct http_options_t
 {
     int timeout;
     bool want_streaming;
     bool keepalive;
+    bool ssl;
 };
 
-struct http_response_t
+struct http_context_t;
+
+struct http_message_t
 {
     std::string error;
 
@@ -41,10 +46,19 @@ struct http_response_t
 // Opened websocket or keepalive HTTP connection
 struct http_co_t;
 
-http_co_t* http_init(timerfd_manager_t *tfd);
+http_context_t* http_context_init(const std::string & ssl_cert, const std::string & ssl_key,
-http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
+    const std::string & ssl_ca, bool verify_peer, std::string & error);
-    int timeout, std::function<void(const http_response_t *msg)> on_message);
+void http_context_destroy(http_context_t *ctx);
+http_co_t* http_init(timerfd_manager_t *tfd, http_context_t *ctx = NULL);
+void open_websocket(http_co_t *handler, const std::string & host, const std::string & path,
+    const http_options_t & options, std::function<void(http_message_t *msg)> on_message);
 void http_request(http_co_t *handler, const std::string & host, const std::string & request,
-    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback);
+    const http_options_t & options, std::function<void(http_message_t *response)> response_callback);
 void http_post_message(http_co_t *handler, uint8_t type, const std::string & msg);
+void http_serve(http_co_t *handler, int peer_fd, const http_options_t & options,
+    std::function<void(http_message_t *msg)> request_callback);
+void http_reply(http_co_t *handler, const std::string & reply);
 void http_close(http_co_t *co);
+void http_destroy(http_co_t *co);
+
+#pragma GCC visibility pop

src/client/messenger.cpp

@@ -15,6 +15,106 @@
 #include "msgr_rdma.h"
 #endif
 
+#include <sys/poll.h>
+
+msgr_iothread_t::msgr_iothread_t():
+    ring(RINGLOOP_DEFAULT_SIZE, true),
+    thread(&msgr_iothread_t::run, this)
+{
+    eventfd = ring.register_eventfd();
+    if (eventfd < 0)
+    {
+        throw std::runtime_error(std::string("failed to register eventfd: ") + strerror(-eventfd));
+    }
+}
+
+msgr_iothread_t::~msgr_iothread_t()
+{
+    stop();
+}
+
+void msgr_iothread_t::add_sqe(io_uring_sqe & sqe)
+{
+    mu.lock();
+    queue.push_back((iothread_sqe_t){ .sqe = sqe, .data = std::move(*(ring_data_t*)sqe.user_data) });
+    if (queue.size() == 1)
+    {
+        cond.notify_all();
+    }
+    mu.unlock();
+}
+
+void msgr_iothread_t::stop()
+{
+    mu.lock();
+    if (stopped)
+    {
+        mu.unlock();
+        return;
+    }
+    stopped = true;
+    if (outer_loop_data)
+    {
+        outer_loop_data->callback = [](ring_data_t*){};
+    }
+    cond.notify_all();
+    close(eventfd);
+    mu.unlock();
+    thread.join();
+}
+
+void msgr_iothread_t::add_to_ringloop(ring_loop_t *outer_loop)
+{
+    assert(!this->outer_loop || this->outer_loop == outer_loop);
+    io_uring_sqe *sqe = outer_loop->get_sqe();
+    assert(sqe != NULL);
+    this->outer_loop = outer_loop;
+    this->outer_loop_data = ((ring_data_t*)sqe->user_data);
+    io_uring_prep_poll_add(sqe, eventfd, POLLIN);
+    outer_loop_data->callback = [this](ring_data_t *data)
+    {
+        if (data->res < 0)
+        {
+            throw std::runtime_error(std::string("eventfd poll failed: ") + strerror(-data->res));
+        }
+        outer_loop_data = NULL;
+        if (stopped)
+        {
+            return;
+        }
+        add_to_ringloop(this->outer_loop);
+        ring.loop();
+    };
+}
+
+void msgr_iothread_t::run()
+{
+    while (true)
+    {
+        {
+            std::unique_lock<std::mutex> lk(mu);
+            while (!stopped && !queue.size())
+                cond.wait(lk);
+            if (stopped)
+                return;
+            int i = 0;
+            for (; i < queue.size(); i++)
+            {
+                io_uring_sqe *sqe = ring.get_sqe();
+                if (!sqe)
+                    break;
+                ring_data_t *data = ((ring_data_t*)sqe->user_data);
+                *data = std::move(queue[i].data);
+                *sqe = queue[i].sqe;
+                sqe->user_data = (uint64_t)data;
+            }
+            queue.erase(queue.begin(), queue.begin()+i);
+        }
+        // We only want to offload sendmsg/recvmsg. Callbacks will be called in main thread
+        ring.submit();
+    }
+}
+
 void osd_messenger_t::init()
 {
 #ifdef WITH_RDMACM
@@ -73,17 +173,21 @@ void osd_messenger_t::init()
     }
     if (ringloop && iothread_count > 0)
     {
-        init_iothreads();
+        for (int i = 0; i < iothread_count; i++)
+        {
+            auto iot = new msgr_iothread_t();
+            iothreads.push_back(iot);
+            iot->add_to_ringloop(ringloop);
+        }
     }
     keepalive_timer_id = tfd->set_timer(1000, true, [this](int)
     {
-        std::vector<uint64_t> clients_to_stop;
-        std::vector<osd_op_t*> ops_to_send;
         auto cl_it = clients.begin();
         while (cl_it != clients.end())
         {
             auto cl = cl_it->second;
             cl_it++;
+            auto peer_fd = cl->peer_fd;
             if (!cl->osd_num && !cl->in_osd_num || cl->peer_state != PEER_CONNECTED && cl->peer_state != PEER_RDMA)
             {
                 // Do not run keepalive on regular clients
@@ -95,9 +199,10 @@ void osd_messenger_t::init()
                 if (!cl->ping_time_remaining)
                 {
                     // Ping timed out, stop the client
-                    fprintf(stderr, "Ping timed out for OSD %ju (client %ju), disconnecting peer\n",
+                    fprintf(stderr, "Ping timed out for OSD %ju (client %d), disconnecting peer\n", cl->in_osd_num ? cl->in_osd_num : cl->osd_num, cl->peer_fd);
-                        cl->in_osd_num ? cl->in_osd_num : cl->osd_num, cl->client_id);
+                    stop_client(peer_fd, true);
-                    clients_to_stop.push_back(cl->client_id);
+                    // Restart iterator because it may be invalidated
+                    cl_it = clients.upper_bound(peer_fd);
                 }
             }
             else if (cl->idle_time_remaining > 0)
@@ -108,36 +213,37 @@ void osd_messenger_t::init()
                     // Connection is idle for <osd_idle_time>, send ping
                     osd_op_t *op = new osd_op_t();
                     op->op_type = OSD_OP_OUT;
-                    op->client_id = cl->client_id;
+                    op->peer_fd = cl->peer_fd;
                     op->req = (osd_any_op_t){
                         .hdr = {
                             .magic = SECONDARY_OSD_OP_MAGIC,
                             .opcode = OSD_OP_PING,
                         },
                     };
-                    op->callback = [this](osd_op_t *op)
+                    op->callback = [this, cl](osd_op_t *op)
                     {
-                        auto cl_it = clients.find(op->client_id);
+                        auto cl_it = clients.find(op->peer_fd);
-                        if (cl_it == clients.end())
+                        if (cl_it == clients.end() || cl_it->second != cl)
                         {
                             // client is already dropped
                             delete op;
                             return;
                         }
-                        auto cl = cl_it->second;
+                        int fail_fd = (op->reply.hdr.retval != 0 ? op->peer_fd : -1);
-                        uint64_t fail_client_id = (op->reply.hdr.retval != 0 ? op->client_id : 0);
                         auto fail_osd_num = cl->in_osd_num ? cl->in_osd_num : cl->osd_num;
                         cl->ping_time_remaining = 0;
                         delete op;
-                        if (fail_client_id)
+                        if (fail_fd >= 0)
                         {
-                            fprintf(stderr, "Ping failed for OSD %ju (client %ju), disconnecting peer\n", fail_osd_num, fail_client_id);
+                            fprintf(stderr, "Ping failed for OSD %ju (client %d), disconnecting peer\n", fail_osd_num, fail_fd);
-                            stop_client(fail_client_id);
+                            stop_client(fail_fd, true);
                         }
                     };
                     cl->ping_time_remaining = osd_ping_timeout;
                     cl->idle_time_remaining = osd_idle_timeout;
-                    ops_to_send.push_back(op);
+                    outbox_push(op);
+                    // Restart iterator because it may be invalidated
+                    cl_it = clients.upper_bound(peer_fd);
                 }
             }
             else
@@ -145,14 +251,6 @@ void osd_messenger_t::init()
                 cl->idle_time_remaining = osd_idle_timeout;
             }
         }
-        for (uint64_t client_id: clients_to_stop)
-        {
-            stop_client(client_id);
-        }
-        for (osd_op_t *op: ops_to_send)
-        {
-            outbox_push(op);
-        }
     });
 }
 
@@ -165,9 +263,16 @@ osd_messenger_t::~osd_messenger_t()
     }
     while (clients.size() > 0)
     {
-        stop_client(clients.begin()->first, true);
+        stop_client(clients.begin()->first, true, true);
+    }
+    if (iothreads.size())
+    {
+        for (auto iot: iothreads)
+        {
+            delete iot;
+        }
+        iothreads.clear();
     }
-    destroy_iothreads();
 #ifdef WITH_RDMA
     for (auto rdma_context: rdma_contexts)
     {
@@ -184,6 +289,16 @@ osd_messenger_t::~osd_messenger_t()
         rdmacm_evch = NULL;
     }
 #endif
+#ifdef WITH_OPENSSL
+    for (auto encrypt_ctx: encrypt_ctx_pool)
+    {
+        destroy_aes_xts_encrypt(encrypt_ctx);
+    }
+    for (auto decrypt_ctx: decrypt_ctx_pool)
+    {
+        destroy_aes_xts_decrypt(decrypt_ctx);
+    }
+#endif
 }
 
 void osd_messenger_t::parse_config(const json11::Json & config)
@@ -218,6 +333,9 @@ void osd_messenger_t::parse_config(const json11::Json & config)
     if (!this->rdma_max_msg || this->rdma_max_msg > 128*1024*1024)
         this->rdma_max_msg = 129*1024;
 #endif
+    this->max_aes_xts_pool_size = config["max_aes_xts_pool_size"].uint64_value();
+    if (!this->max_aes_xts_pool_size)
+        this->max_aes_xts_pool_size = 256;
     if (!osd_num)
         this->iothread_count = (uint32_t)config["client_iothread_count"].uint64_value();
     else
@@ -335,7 +453,7 @@ void osd_messenger_t::try_connect_peer(uint64_t peer_osd)
     {
         return;
     }
-    if (osd_peers.find(peer_osd) != osd_peers.end())
+    if (osd_peer_fds.find(peer_osd) != osd_peer_fds.end())
     {
         wanted_peers.erase(peer_osd);
         return;
@@ -362,20 +480,20 @@ void osd_messenger_t::try_connect_peer_tcp(osd_num_t peer_osd, const char *peer_
 #ifdef WITH_RDMACM
     if (disable_tcp)
     {
-        on_connect_peer(peer_osd, -EINVAL, 0);
+        on_connect_peer(peer_osd, -EINVAL);
         return;
     }
 #endif
     struct sockaddr_storage addr;
     if (!string_to_addr(peer_host, 0, peer_port, &addr))
     {
-        on_connect_peer(peer_osd, -EINVAL, 0);
+        on_connect_peer(peer_osd, -EINVAL);
         return;
     }
     int peer_fd = socket(addr.ss_family, SOCK_STREAM, 0);
     if (peer_fd < 0)
     {
-        on_connect_peer(peer_osd, -errno, 0);
+        on_connect_peer(peer_osd, -errno);
         return;
     }
     fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
@@ -383,25 +501,21 @@ void osd_messenger_t::try_connect_peer_tcp(osd_num_t peer_osd, const char *peer_
     if (r < 0 && errno != EINPROGRESS)
     {
         close(peer_fd);
-        on_connect_peer(peer_osd, -errno, 0);
+        on_connect_peer(peer_osd, -errno);
         return;
     }
-    const uint64_t client_id = next_client_id++;
+    clients[peer_fd] = new osd_client_t();
-    osd_client_t *cl = new osd_client_t();
     if (log_level > 0)
     {
-        fprintf(stderr, "Connecting to OSD %ju at %s:%d (client %ju, FD %d)\n", peer_osd, peer_host, peer_port, client_id, peer_fd);
+        fprintf(stderr, "Connecting to OSD %ju at %s:%d (client %d)\n", peer_osd, peer_host, peer_port, peer_fd);
     }
-    cl->client_id = client_id;
+    clients[peer_fd]->peer_addr = addr;
-    cl->peer_addr = addr;
+    clients[peer_fd]->peer_port = peer_port;
-    cl->peer_port = peer_port;
+    clients[peer_fd]->peer_fd = peer_fd;
-    cl->peer_fd = peer_fd;
+    clients[peer_fd]->peer_state = PEER_CONNECTING;
-    cl->peer_state = PEER_CONNECTING;
+    clients[peer_fd]->connect_timeout_id = -1;
-    cl->connect_timeout_id = -1;
+    clients[peer_fd]->osd_num = peer_osd;
-    cl->osd_num = peer_osd;
+    clients[peer_fd]->in_buf = (uint8_t*)malloc_or_die(receive_buffer_size);
-    cl->in_buf = malloc_or_die(receive_buffer_size);
-    clients[client_id] = cl;
-    clients_by_fd[peer_fd] = cl;
     tfd->set_fd_handler(peer_fd, true, [this](int peer_fd, int epoll_events)
     {
         // Either OUT (connected) or HUP
@@ -409,11 +523,11 @@ void osd_messenger_t::try_connect_peer_tcp(osd_num_t peer_osd, const char *peer_
     });
     if (peer_connect_timeout > 0)
     {
-        cl->connect_timeout_id = tfd->set_timer(1000*peer_connect_timeout, false, [this, client_id](int timer_id)
+        clients[peer_fd]->connect_timeout_id = tfd->set_timer(1000*peer_connect_timeout, false, [this, peer_fd](int timer_id)
         {
-            osd_num_t peer_osd = clients.at(client_id)->osd_num;
+            osd_num_t peer_osd = clients.at(peer_fd)->osd_num;
-            stop_client(client_id);
+            stop_client(peer_fd, true);
-            on_connect_peer(peer_osd, -EPIPE, 0);
+            on_connect_peer(peer_osd, -EPIPE);
             return;
         });
     }
@@ -421,7 +535,7 @@ void osd_messenger_t::try_connect_peer_tcp(osd_num_t peer_osd, const char *peer_
 
 void osd_messenger_t::handle_connect_epoll(int peer_fd)
 {
-    auto cl = clients_by_fd.at(peer_fd);
+    auto cl = clients[peer_fd];
     if (cl->connect_timeout_id >= 0)
     {
         tfd->clear_timer(cl->connect_timeout_id);
@@ -436,8 +550,8 @@ void osd_messenger_t::handle_connect_epoll(int peer_fd)
     }
     if (result != 0)
     {
-        stop_client(cl->client_id);
+        stop_client(peer_fd, true);
-        on_connect_peer(peer_osd, -result, 0);
+        on_connect_peer(peer_osd, -result);
         return;
     }
     int one = 1;
@@ -454,23 +568,23 @@ void osd_messenger_t::handle_connect_epoll(int peer_fd)
 void osd_messenger_t::handle_peer_epoll(int peer_fd, int epoll_events)
 {
     // Mark client as ready (i.e. some data is available)
-    auto cl = clients_by_fd.at(peer_fd);
     if (epoll_events & EPOLLRDHUP)
     {
         // Stop client
         if (log_level > 0)
         {
-            fprintf(stderr, "[OSD %ju] client %ju disconnected\n", this->osd_num, cl->client_id);
+            fprintf(stderr, "[OSD %ju] client %d disconnected\n", this->osd_num, peer_fd);
         }
-        stop_client(cl->client_id);
+        stop_client(peer_fd, true);
     }
     else if (epoll_events & EPOLLIN)
     {
         // Mark client as ready (i.e. some data is available)
+        auto cl = clients[peer_fd];
         cl->read_ready++;
         if (cl->read_ready == 1)
         {
-            read_ready_clients.push_back(cl->client_id);
+            read_ready_clients.push_back(cl->peer_fd);
             if (ringloop)
                 ringloop->wakeup();
             else
@@ -479,13 +593,13 @@ void osd_messenger_t::handle_peer_epoll(int peer_fd, int epoll_events)
     }
 }
 
-void osd_messenger_t::on_connect_peer(osd_num_t peer_osd, int errcode, uint64_t client_id)
+void osd_messenger_t::on_connect_peer(osd_num_t peer_osd, int peer_fd)
 {
     auto & wp = wanted_peers.at(peer_osd);
     wp.connecting = false;
-    if (errcode < 0)
+    if (peer_fd < 0)
     {
-        fprintf(stderr, "Failed to connect to peer OSD %ju address %s port %d: %s\n", peer_osd, wp.cur_addr.c_str(), wp.cur_port, strerror(-errcode));
+        fprintf(stderr, "Failed to connect to peer OSD %ju address %s port %d: %s\n", peer_osd, wp.cur_addr.c_str(), wp.cur_port, strerror(-peer_fd));
         if (wp.address_changed)
         {
             wp.address_changed = false;
@@ -512,7 +626,7 @@ void osd_messenger_t::on_connect_peer(osd_num_t peer_osd, int errcode, uint64_t
     }
     if (log_level > 0)
     {
-        fprintf(stderr, "[OSD %ju] Connected with peer OSD %ju (client %ju)\n", osd_num, peer_osd, client_id);
+        fprintf(stderr, "[OSD %ju] Connected with peer OSD %ju (client %d)\n", osd_num, peer_osd, peer_fd);
     }
     wanted_peers.erase(peer_osd);
     repeer_pgs(peer_osd);
@@ -522,7 +636,7 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
 {
     osd_op_t *op = new osd_op_t();
     op->op_type = OSD_OP_OUT;
-    op->client_id = cl->client_id;
+    op->peer_fd = cl->peer_fd;
     op->req = (osd_any_op_t){
         .show_conf = {
             .header = {
@@ -546,7 +660,7 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
         if (!selected_ctx)
         {
             if (log_level > 0)
-                fprintf(stderr, "No RDMA context for OSD %ju connection (client %ju), using only TCP\n", cl->osd_num, cl->client_id);
+                fprintf(stderr, "No RDMA context for OSD %ju connection (peer %d), using only TCP\n", cl->osd_num, cl->peer_fd);
         }
         else
         {
@@ -607,8 +721,8 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
         if (err)
         {
             osd_num_t peer_osd = cl->osd_num;
-            stop_client(op->client_id);
+            stop_client(op->peer_fd);
-            on_connect_peer(peer_osd, -EINVAL, 0);
+            on_connect_peer(peer_osd, -EINVAL);
             delete op;
             return;
         }
@@ -638,13 +752,21 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
                     fprintf(stderr, "Connected to OSD %ju using RDMA\n", cl->osd_num);
                 }
                 cl->peer_state = PEER_RDMA;
+                tfd->set_fd_handler(cl->peer_fd, false, [this](int peer_fd, int epoll_events)
+                {
+                    // Do not miss the disconnection!
+                    if (epoll_events & EPOLLRDHUP)
+                    {
+                        handle_peer_epoll(peer_fd, epoll_events);
+                    }
+                });
                 // Add the initial receive request
                 init_recv_rdma(cl);
             }
         }
 #endif
-        osd_peers[cl->osd_num] = cl;
+        osd_peer_fds[cl->osd_num] = cl->peer_fd;
-        on_connect_peer(cl->osd_num, 0, cl->client_id);
+        on_connect_peer(cl->osd_num, cl->peer_fd);
         delete op;
     };
     outbox_push(op);
@@ -659,23 +781,20 @@ void osd_messenger_t::accept_connections(int listen_fd)
     while ((peer_fd = accept(listen_fd, (sockaddr*)&addr, &peer_addr_size)) >= 0)
     {
         assert(peer_fd != 0);
-        const uint64_t client_id = next_client_id++;
+        fprintf(stderr, "[OSD %ju] new client %d: connection from %s\n", this->osd_num, peer_fd,
-        fprintf(stderr, "[OSD %ju] new client %ju (FD %d): connection from %s\n", this->osd_num, client_id, peer_fd,
             addr_to_string(addr).c_str());
         fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
         int one = 1;
         setsockopt(peer_fd, SOL_TCP, TCP_NODELAY, &one, sizeof(one));
         auto cl = new osd_client_t();
-        cl->client_id = client_id;
+        clients[peer_fd] = cl;
-        clients[cl->client_id] = cl;
-        clients_by_fd[peer_fd] = cl;
         cl->is_incoming = true;
         cl->peer_addr = addr;
         cl->peer_addr = addr;
         cl->peer_port = ntohs(((sockaddr_in*)&addr)->sin_port);
         cl->peer_fd = peer_fd;
         cl->peer_state = PEER_CONNECTED;
-        cl->in_buf = malloc_or_die(receive_buffer_size);
+        cl->in_buf = (uint8_t*)malloc_or_die(receive_buffer_size);
         // Add FD to epoll
         tfd->set_fd_handler(peer_fd, false, [this](int peer_fd, int epoll_events)
         {

src/client/messenger.h

@@ -12,7 +12,6 @@
 #include <deque>
 #include <vector>
 
-#include "../util/robin_hood.h"
 #include "malloc_or_die.h"
 #include "json11/json11.hpp"
 #include "msgr_op.h"
@@ -35,9 +34,6 @@
 
 #define DEFAULT_MIN_ZEROCOPY_SEND_SIZE 32*1024
 
-#define MSGR_SENDP_HDR 1
-#define MSGR_SENDP_FREE 2
-
 struct msgr_sendp_t
 {
     osd_op_t *op;
@@ -49,9 +45,15 @@ struct msgr_rdma_connection_t;
 struct msgr_rdma_context_t;
 #endif
 
+#ifdef WITH_OPENSSL
+struct op_aes_xts_encrypt_t;
+struct op_aes_xts_decrypt_t;
+void destroy_aes_xts_encrypt(op_aes_xts_encrypt_t *encrypt_ctx);
+void destroy_aes_xts_decrypt(op_aes_xts_decrypt_t *decrypt_ctx);
+#endif
+
 struct osd_client_t
 {
-    uint64_t client_id = 0;
     int refs = 0;
 
     sockaddr_storage peer_addr = {};
@@ -65,20 +67,23 @@ struct osd_client_t
     osd_num_t in_osd_num = 0;
     bool is_incoming = false;
 
-    void *in_buf = NULL;
+    uint8_t *in_buf = NULL;
 
 #ifdef WITH_RDMA
     msgr_rdma_connection_t *rdma_conn = NULL;
 #endif
 
     // Read state
+    op_aes_xts_decrypt_t *decrypt_ctx = NULL;
     int read_ready = 0;
     osd_op_t *read_op = NULL;
+    size_t read_op_size = 0;
+    size_t read_op_pos = 0;
+    size_t read_op_inline_decrypt_pos = 0;
     iovec read_iov = { 0 };
     msghdr read_msg = { 0 };
-    int read_remaining = 0;
+    std::vector<iovec> recv_list;
-    int read_state = 0;
+    size_t recv_list_size = 0;
-    osd_op_buf_list_t recv_list;
     uint64_t read_op_id = 1;
     bool check_sequencing = false;
     bool enable_pg_locks = false;
@@ -87,17 +92,22 @@ struct osd_client_t
     std::vector<osd_op_t*> received_ops;
 
     // Outbound operations
-    robin_hood::unordered_flat_map<uint64_t, osd_op_t*> sent_ops;
+    std::map<uint64_t, osd_op_t*> sent_ops;
     uint64_t send_op_id = 0;
 
     // PGs dirtied by this client's primary-writes
     std::set<pool_pg_num_t> dirty_pgs;
 
     // Write state
+    op_aes_xts_encrypt_t *encrypt_ctx = NULL;
+    std::deque<osd_op_t *> write_ops;
+    osd_op_t *write_op = NULL;
+    size_t write_op_pos = 0;
     msghdr write_msg = { 0 };
     int write_state = 0;
-    std::vector<iovec> send_list, next_send_list;
+    std::vector<iovec> send_list;
-    std::vector<msgr_sendp_t> outbox, next_outbox;
+    size_t send_list_size = 0;
+    std::deque<osd_op_t*> send_free_ops;
     std::vector<osd_op_t*> zc_free_list;
 
     ~osd_client_t();
@@ -129,7 +139,43 @@ struct osd_op_stats_t
     uint64_t subop_stat_count[OSD_OP_MAX+1] = { 0 };
 };
 
+#include <mutex>
+#include <condition_variable>
+#include <thread>
+
+#ifdef __MOCK__
 class msgr_iothread_t;
+#else
+struct iothread_sqe_t
+{
+    io_uring_sqe sqe;
+    ring_data_t data;
+};
+
+class msgr_iothread_t
+{
+protected:
+    ring_loop_t ring;
+    ring_loop_t *outer_loop = NULL;
+    ring_data_t *outer_loop_data = NULL;
+    int eventfd = -1;
+    bool stopped = false;
+    std::mutex mu;
+    std::condition_variable cond;
+    std::vector<iothread_sqe_t> queue;
+    std::thread thread;
+
+    void run();
+public:
+
+    msgr_iothread_t();
+    ~msgr_iothread_t();
+
+    void add_sqe(io_uring_sqe & sqe);
+    void stop();
+    void add_to_ringloop(ring_loop_t *outer_loop);
+};
+#endif
 
 #ifdef WITH_RDMA
 struct rdma_event_channel;
@@ -154,6 +200,7 @@ protected:
     bool use_sync_send_recv = false;
     int min_zerocopy_send_size = DEFAULT_MIN_ZEROCOPY_SEND_SIZE;
     int iothread_count = 0;
+    int max_aes_xts_pool_size = 256;
 
 #ifdef WITH_RDMA
     bool use_rdma = true;
@@ -167,27 +214,30 @@ protected:
     uint64_t rdma_max_sge = 0, rdma_max_send = 0, rdma_max_recv = 0;
     uint64_t rdma_max_msg = 0;
     rdma_event_channel *rdmacm_evch = NULL;
-    robin_hood::unordered_flat_map<rdma_cm_id*, osd_client_t*> rdmacm_connections;
+    std::map<rdma_cm_id*, osd_client_t*> rdmacm_connections;
-    robin_hood::unordered_flat_map<rdma_cm_id*, rdmacm_connecting_t*> rdmacm_connecting;
+    std::map<rdma_cm_id*, rdmacm_connecting_t*> rdmacm_connecting;
 #endif
 
     std::vector<msgr_iothread_t*> iothreads;
-    std::vector<uint64_t> read_ready_clients;
+    std::vector<int> read_ready_clients;
-    std::vector<uint64_t> write_ready_clients;
+    std::vector<int> write_ready_clients;
     // We don't use ringloop->set_immediate here because we may have no ringloop in client :)
-    std::deque<osd_op_t*> set_immediate_ops;
+    std::vector<osd_op_t*> set_immediate_ops;
+
+#ifdef WITH_OPENSSL
+    std::vector<op_aes_xts_encrypt_t*> encrypt_ctx_pool;
+    std::vector<op_aes_xts_decrypt_t*> decrypt_ctx_pool;
+#endif
 
 public:
     timerfd_manager_t *tfd = NULL;
-    ring_loop_i *ringloop = NULL;
+    ring_loop_t *ringloop = NULL;
     bool has_sendmsg_zc = false;
     // osd_num_t is only for logging and asserts
-    uint64_t next_client_id = 1;
     osd_num_t osd_num;
-    robin_hood::unordered_flat_map<uint64_t, osd_client_t*> clients;
+    std::map<int, osd_client_t*> clients;
-    robin_hood::unordered_flat_map<uint64_t, osd_client_t*> osd_peers;
+    std::map<osd_num_t, osd_wanted_peer_t> wanted_peers;
-    robin_hood::unordered_flat_map<int, osd_client_t*> clients_by_fd;
+    std::map<uint64_t, int> osd_peer_fds;
-    robin_hood::unordered_flat_map<osd_num_t, osd_wanted_peer_t> wanted_peers;
     std::vector<std::string> osd_networks;
     std::vector<addr_mask_t> osd_network_masks;
     std::vector<std::string> osd_cluster_networks;
@@ -198,11 +248,9 @@ public:
     osd_op_stats_t stats, recovery_stats;
 
     void init();
-    void init_iothreads();
     void parse_config(const json11::Json & config);
     void connect_peer(uint64_t osd_num, json11::Json peer_state);
-    void stop_client(uint64_t client_id, bool force_delete = false);
+    void stop_client(int peer_fd, bool force = false, bool force_delete = false);
-    void destroy_client(osd_client_t *cl);
     void outbox_push(osd_op_t *cur_op);
     std::function<void(osd_op_t*)> exec_op;
     std::function<void(osd_num_t)> repeer_pgs;
@@ -211,7 +259,6 @@ public:
     void read_requests();
     void send_replies();
     void accept_connections(int listen_fd);
-    void destroy_iothreads();
     ~osd_messenger_t();
 
     static json11::Json::object read_config(const json11::Json & config);
@@ -222,7 +269,7 @@ public:
 
 #ifdef WITH_RDMA
     bool is_rdma_enabled();
-    bool connect_rdma(uint64_t client_id, std::string rdma_address, uint64_t client_max_msg);
+    bool connect_rdma(int peer_fd, std::string rdma_address, uint64_t client_max_msg);
 #endif
 #ifdef WITH_RDMACM
     bool is_use_rdmacm();
@@ -238,28 +285,43 @@ protected:
     void try_connect_peer_tcp(osd_num_t peer_osd, const char *peer_host, int peer_port);
     void handle_peer_epoll(int peer_fd, int epoll_events);
     void handle_connect_epoll(int peer_fd);
-    void on_connect_peer(osd_num_t peer_osd, int errcode, uint64_t client_id);
+    void on_connect_peer(osd_num_t peer_osd, int peer_fd);
     void check_peer_config(osd_client_t *cl);
     void cancel_osd_ops(osd_client_t *cl);
     void cancel_op(osd_op_t *op);
 
     bool try_send(osd_client_t *cl);
     void handle_send(int result, bool prev, bool more, osd_client_t *cl);
+    bool op_encrypted_copy_data_to(osd_client_t* cl, uint8_t *buf, size_t len, size_t from, size_t & done);
+    size_t op_copy_to(osd_client_t *cl, uint8_t *dst, size_t dst_len);
+    void op_get_write_buffers(osd_client_t *cl, std::vector<iovec> & lst);
+
+    void handle_read(int result, osd_client_t *cl);
+    bool handle_read_buffer(osd_client_t *cl, uint8_t *curbuf, size_t bufsize);
+    bool handle_hdr(osd_client_t *cl);
+    bool allocate_op_buffers(osd_client_t *cl);
+    bool allocate_reply_buffers(osd_client_t *cl, osd_op_t *op);
+    size_t op_copy_from(osd_client_t *cl, uint8_t *src, size_t src_len, size_t & done);
+    bool op_decrypted_copy_data_from(osd_client_t* cl, uint8_t *buf, size_t len, size_t from, size_t & done);
+    void op_decrypt_start(osd_client_t* cl);
+    void op_decrypt_inline(osd_client_t* cl);
+    void op_decrypt_free(osd_client_t* cl);
+    size_t op_get_read_buffers(osd_client_t *cl, std::vector<iovec> & lst);
+    void handle_finished_op(osd_client_t *cl);
 
-    bool handle_read(int result, osd_client_t *cl);
-    bool handle_read_buffer(osd_client_t *cl, void *curbuf, int remain);
     bool handle_finished_read(osd_client_t *cl);
     void handle_op_hdr(osd_client_t *cl);
     bool handle_reply_hdr(osd_client_t *cl);
     void handle_reply_ready(osd_op_t *op);
     void handle_immediate_ops();
+    void clear_immediate_ops(int peer_fd);
 
 #ifdef WITH_RDMA
     void try_send_rdma(osd_client_t *cl);
+    int try_send_rdma_copy(osd_client_t *cl, uint8_t *dst, int dst_len);
     bool init_recv_rdma(osd_client_t *cl);
     void handle_rdma_events(msgr_rdma_context_t *rdma_context);
     msgr_rdma_context_t* choose_rdma_context(osd_client_t *cl);
-    void destroy_rdma_conn(msgr_rdma_connection_t *rdma_conn);
 #endif
 #ifdef WITH_RDMACM
     void handle_rdmacm_events();

src/client/msgr_encrypt.cpp

@@ -0,0 +1,328 @@
+// Copyright (c) Vitaliy Filippov, 2026+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+#define _XOPEN_SOURCE
+#include <limits.h>
+#include <assert.h>
+
+#include "etcd_state_client.h"
+#include "messenger.h"
+#include "msgr_encrypt.h"
+
+// FIXME Fuck, no streaming...
+
+op_aes_xts_encrypt_t::op_aes_xts_encrypt_t()
+{
+    if (!(ctx = EVP_CIPHER_CTX_new()))
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+    EVP_CIPHER_CTX_set_padding(ctx, 0);
+    if (EVP_EncryptInit_ex(ctx, EVP_aes_256_xts(), NULL, NULL, NULL) != 1)
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+}
+
+op_aes_xts_encrypt_t::~op_aes_xts_encrypt_t()
+{
+    EVP_CIPHER_CTX_free(ctx);
+}
+
+void op_aes_xts_encrypt_t::start(const uint8_t *key, uint64_t start_offset, size_t block_size)
+{
+    this->start_offset = start_offset;
+    this->key = key;
+    this->block_size = block_size;
+    this->offset = 0;
+}
+
+void op_aes_xts_encrypt_t::update(uint8_t *in, size_t max_in, uint8_t *out, size_t max_out, size_t & done_in, size_t & done_out)
+{
+    if (max_in > block_size - offset%block_size)
+        max_in = block_size - offset%block_size;
+    size_t insize = max_in;
+    size_t outsize = ((offset+insize)/16 - offset/16) * 16;
+    if (outsize > max_out)
+    {
+        // encrypt is used to send data through temporary buffer(s),
+        // so we don't care to support fragmenting output into < 16 b parts
+        insize = (max_out < 16 ? 0 : (max_out & ~15) - offset%16);
+        outsize = ((offset+insize)/16 - offset/16) * 16;
+    }
+    assert(insize <= max_in);
+    assert(outsize <= max_out);
+    if (!(offset % block_size))
+    {
+        uint8_t iv[16] = { 0 };
+        *((uint64_t*)iv) = start_offset + offset;
+        if (EVP_EncryptInit_ex(ctx, NULL, NULL, key, iv) != 1)
+        {
+            ERR_print_errors_fp(stderr);
+            abort();
+        }
+    }
+    int actual_out = 0;
+    if (EVP_EncryptUpdate(ctx, out, &actual_out, in, insize) != 1)
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+    assert(actual_out == outsize);
+    done_in += insize;
+    done_out += outsize;
+    offset += insize;
+}
+
+void destroy_aes_xts_encrypt(op_aes_xts_encrypt_t *encrypt_ctx)
+{
+    delete encrypt_ctx;
+}
+
+op_aes_xts_decrypt_t::op_aes_xts_decrypt_t()
+{
+    if (!(ctx = EVP_CIPHER_CTX_new()))
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+    EVP_CIPHER_CTX_set_padding(ctx, 0);
+    if (EVP_DecryptInit_ex(ctx, EVP_aes_256_xts(), NULL, NULL, NULL) != 1)
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+}
+
+op_aes_xts_decrypt_t::~op_aes_xts_decrypt_t()
+{
+    EVP_CIPHER_CTX_free(ctx);
+}
+
+void op_aes_xts_decrypt_t::start(const uint8_t *key, uint64_t start_offset, size_t block_size)
+{
+    this->start_offset = start_offset;
+    this->key = key;
+    this->block_size = block_size;
+    this->in_offset = 0;
+    this->tmp_pos = 16;
+}
+
+void op_aes_xts_decrypt_t::update(uint8_t *in, size_t max_in, uint8_t *out, size_t max_out, size_t & done_in, size_t & done_out)
+{
+    if (max_in > block_size - in_offset%block_size)
+        max_in = block_size - in_offset%block_size;
+    int actual_out = 0;
+    // Write previously buffered block to support small output buffers
+    if (tmp_pos < 16)
+    {
+        size_t tmp_size = 16-tmp_pos;
+        if (tmp_size > max_out)
+            tmp_size = max_out;
+        memcpy(out, tmp_buf+tmp_pos, tmp_size);
+        tmp_pos += tmp_size;
+        done_out += tmp_size;
+        out += tmp_size;
+        max_out -= tmp_size;
+        if (!max_out)
+            return;
+        assert(tmp_pos == 16);
+    }
+    if (!(in_offset % block_size))
+    {
+        uint8_t iv[16] = { 0 };
+        *((uint64_t*)iv) = start_offset+in_offset;
+        if (EVP_DecryptInit_ex(ctx, NULL, NULL, key, iv) != 1)
+        {
+            ERR_print_errors_fp(stderr);
+            abort();
+        }
+    }
+    size_t insize = max_in;
+    size_t outsize = ((in_offset+insize)/16 - in_offset/16) * 16;
+    if (outsize > max_out)
+    {
+        if (max_out < 16)
+        {
+            // We can only decrypt a partial block
+            insize = 16 - in_offset%16;
+            if (EVP_DecryptUpdate(ctx, tmp_buf, &actual_out, in, insize) != 1)
+            {
+                ERR_print_errors_fp(stderr);
+                abort();
+            }
+            assert(actual_out == 16);
+            in_offset += insize;
+            in += insize;
+            max_in -= insize;
+            tmp_pos = 0;
+            return;
+        }
+        // Otherwise, we can decrypt at least some data into <out> directly
+        insize = (max_out & ~15) - (in_offset % 16);
+        outsize = (max_out & ~15);
+        assert(insize < max_in);
+    }
+    if (EVP_DecryptUpdate(ctx, out, &actual_out, in, insize) != 1)
+    {
+        ERR_print_errors_fp(stderr);
+        abort();
+    }
+    assert(actual_out == outsize);
+    in_offset += insize;
+    done_in += insize;
+    done_out += actual_out;
+}
+
+void destroy_aes_xts_decrypt(op_aes_xts_decrypt_t *decrypt_ctx)
+{
+    delete decrypt_ctx;
+}
+
+bool osd_messenger_t::op_encrypted_copy_data_to(osd_client_t* cl, uint8_t *enc_buf, size_t enc_len, size_t from, size_t & done)
+{
+    auto op = cl->write_op;
+    auto & op_pos = cl->write_op_pos;
+    assert(op->req.hdr.opcode == OSD_OP_WRITE);
+    if (!from)
+    {
+        if (!cl->encrypt_ctx)
+        {
+            if (encrypt_ctx_pool.size())
+            {
+                cl->encrypt_ctx = encrypt_ctx_pool.back();
+                encrypt_ctx_pool.pop_back();
+            }
+            else
+                cl->encrypt_ctx = new op_aes_xts_encrypt_t();
+        }
+        assert(op->enc->key.size() == 512/8);
+        cl->encrypt_ctx->start(op->enc->key.data(), op->req.rw.offset, op->enc->bitmap_granularity);
+    }
+    for (int i = 0; i < op->iov.count; i++)
+    {
+        uint8_t *plain = (uint8_t*)op->iov.buf[i].iov_base;
+        size_t plain_len = op->iov.buf[i].iov_len;
+        while (from < plain_len)
+        {
+            size_t done_in = 0;
+            size_t done_out = 0;
+            cl->encrypt_ctx->update(plain+from, plain_len-from, enc_buf+done, enc_len-done, done_in, done_out);
+            done += done_out;
+            op_pos += done_in;
+            from += done_in;
+            if (!done_in)
+                return false;
+        }
+        from -= plain_len;
+    }
+    if (cl->encrypt_ctx)
+    {
+        if (encrypt_ctx_pool.size() > max_aes_xts_pool_size)
+            delete cl->encrypt_ctx;
+        else
+            encrypt_ctx_pool.push_back(cl->encrypt_ctx);
+        cl->encrypt_ctx = NULL;
+    }
+    return true;
+}
+
+bool osd_messenger_t::op_decrypted_copy_data_from(osd_client_t* cl, uint8_t *enc_buf, size_t enc_len, size_t from, size_t & done)
+{
+    op_decrypt_start(cl);
+    auto op = cl->read_op;
+    auto & op_pos = cl->read_op_pos;
+    assert(op->req.hdr.opcode == OSD_OP_READ);
+    uint64_t offset = from;
+    for (int i = 0; i < op->iov.count; i++)
+    {
+        uint8_t *plain = (uint8_t*)op->iov.buf[i].iov_base;
+        size_t plain_len = op->iov.buf[i].iov_len;
+        while (from < plain_len)
+        {
+            size_t done_in = 0;
+            size_t done_out = 0;
+            cl->decrypt_ctx->update(enc_buf+done, enc_len-done, plain+from, plain_len-from, done_in, done_out);
+            done += done_in;
+            offset += done_in;
+            op_pos += done_out;
+            from += done_out;
+            if (!done_in)
+                return false;
+        }
+        from -= plain_len;
+    }
+    op_decrypt_free(cl);
+    return true;
+}
+
+void osd_messenger_t::op_decrypt_start(osd_client_t* cl)
+{
+    if (!cl->decrypt_ctx)
+    {
+        if (decrypt_ctx_pool.size())
+        {
+            cl->decrypt_ctx = decrypt_ctx_pool.back();
+            decrypt_ctx_pool.pop_back();
+        }
+        else
+            cl->decrypt_ctx = new op_aes_xts_decrypt_t();
+        assert(cl->read_op->enc->key.size() == 512/8);
+        cl->decrypt_ctx->start(cl->read_op->enc->key.data(), cl->read_op->req.rw.offset, cl->read_op->enc->bitmap_granularity);
+    }
+}
+
+void osd_messenger_t::op_decrypt_inline(osd_client_t* cl)
+{
+    op_decrypt_start(cl);
+    osd_op_t *op = cl->read_op;
+    size_t from_in = cl->read_op_inline_decrypt_pos - OSD_PACKET_SIZE - op->reply.rw.bitmap_len;
+    int i = 0;
+    while (i < op->iov.count && from_in >= op->iov.buf[i].iov_len)
+    {
+        from_in -= op->iov.buf[i].iov_len;
+        i++;
+    }
+    size_t from_out = from_in;
+    int j = i;
+    while (i < op->iov.count && j < op->iov.count)
+    {
+        uint8_t *in = (uint8_t*)op->iov.buf[i].iov_base + from_in;
+        size_t in_len = op->iov.buf[i].iov_len - from_in;
+        uint8_t *out = (uint8_t*)op->iov.buf[j].iov_base + from_out;
+        size_t out_len = op->iov.buf[j].iov_len - from_out;
+        size_t done_in = 0;
+        size_t done_out = 0;
+        cl->decrypt_ctx->update(in, in_len, out, out_len, done_in, done_out);
+        if (done_in >= in_len)
+        {
+            i++;
+            from_in = 0;
+        }
+        else
+            from_in += done_in;
+        if (done_out >= out_len)
+        {
+            j++;
+            from_out = 0;
+        }
+        else
+            from_out += done_out;
+    }
+    assert(j >= op->iov.count);
+    op_decrypt_free(cl);
+}
+
+void osd_messenger_t::op_decrypt_free(osd_client_t* cl)
+{
+    if (cl->decrypt_ctx)
+    {
+        if (decrypt_ctx_pool.size() > max_aes_xts_pool_size)
+            delete cl->decrypt_ctx;
+        else
+            decrypt_ctx_pool.push_back(cl->decrypt_ctx);
+        cl->decrypt_ctx = NULL;
+    }
+}

src/client/msgr_encrypt.h

@@ -0,0 +1,44 @@
+// Copyright (c) Vitaliy Filippov, 2026+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+#include <stdint.h>
+
+#include <openssl/conf.h>
+#include <openssl/evp.h>
+#include <openssl/err.h>
+
+struct op_aes_xts_encrypt_t
+{
+    EVP_CIPHER_CTX *ctx = NULL;
+    uint64_t start_offset = 0;
+    const uint8_t *key = NULL;
+    size_t offset = 0;
+    size_t block_size = 0;
+
+    op_aes_xts_encrypt_t();
+    ~op_aes_xts_encrypt_t();
+
+    void start(const uint8_t *key, uint64_t start_offset, size_t block_size);
+    void update(uint8_t *in, size_t max_in, uint8_t *out, size_t max_out, size_t & done_in, size_t & done_out);
+};
+
+void destroy_aes_xts_encrypt(op_aes_xts_encrypt_t *encrypt_ctx);
+
+struct op_aes_xts_decrypt_t
+{
+    EVP_CIPHER_CTX *ctx = NULL;
+    uint64_t start_offset = 0;
+    const uint8_t *key = NULL;
+    uint8_t tmp_buf[16];
+    size_t tmp_pos = 16;
+    size_t in_offset = 0;
+    size_t block_size = 0;
+
+    op_aes_xts_decrypt_t();
+    ~op_aes_xts_decrypt_t();
+
+    void start(const uint8_t *key, uint64_t start_offset, size_t block_size);
+    void update(uint8_t *in, size_t max_in, uint8_t *out, size_t max_out, size_t & done_in, size_t & done_out);
+};
+
+void destroy_aes_xts_decrypt(op_aes_xts_decrypt_t *decrypt_ctx);

src/client/msgr_iothread.cpp

@@ -1,129 +0,0 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
-
-#include <stdexcept>
-#include <sys/poll.h>
-#include <unistd.h>
-
-#include "messenger.h"
-#include "msgr_iothread.h"
-
-msgr_iothread_t::msgr_iothread_t():
-    ring(RINGLOOP_DEFAULT_SIZE, true),
-    thread(&msgr_iothread_t::run, this)
-{
-    eventfd = ring.register_eventfd();
-    if (eventfd < 0)
-    {
-        throw std::runtime_error(std::string("failed to register eventfd: ") + strerror(-eventfd));
-    }
-}
-
-msgr_iothread_t::~msgr_iothread_t()
-{
-    stop();
-}
-
-void msgr_iothread_t::add_sqe(io_uring_sqe & sqe)
-{
-    mu.lock();
-    queue.push_back((iothread_sqe_t){ .sqe = sqe, .data = std::move(*(ring_data_t*)sqe.user_data) });
-    if (queue.size() == 1)
-    {
-        cond.notify_all();
-    }
-    mu.unlock();
-}
-
-void msgr_iothread_t::stop()
-{
-    mu.lock();
-    if (stopped)
-    {
-        mu.unlock();
-        return;
-    }
-    stopped = true;
-    if (outer_loop_data)
-    {
-        outer_loop_data->callback = [](ring_data_t*){};
-    }
-    cond.notify_all();
-    close(eventfd);
-    mu.unlock();
-    thread.join();
-}
-
-void msgr_iothread_t::add_to_ringloop(ring_loop_i *outer_loop)
-{
-    assert(!this->outer_loop || this->outer_loop == outer_loop);
-    io_uring_sqe *sqe = outer_loop->get_sqe();
-    assert(sqe != NULL);
-    this->outer_loop = outer_loop;
-    this->outer_loop_data = ((ring_data_t*)sqe->user_data);
-    io_uring_prep_poll_add(sqe, eventfd, POLLIN);
-    outer_loop_data->callback = [this](ring_data_t *data)
-    {
-        if (data->res < 0)
-        {
-            throw std::runtime_error(std::string("eventfd poll failed: ") + strerror(-data->res));
-        }
-        outer_loop_data = NULL;
-        if (stopped)
-        {
-            return;
-        }
-        add_to_ringloop(this->outer_loop);
-        ring.loop();
-    };
-}
-
-void msgr_iothread_t::run()
-{
-    while (true)
-    {
-        {
-            std::unique_lock<std::mutex> lk(mu);
-            while (!stopped && !queue.size())
-                cond.wait(lk);
-            if (stopped)
-                return;
-            int i = 0;
-            for (; i < queue.size(); i++)
-            {
-                io_uring_sqe *sqe = ring.get_sqe();
-                if (!sqe)
-                    break;
-                ring_data_t *data = ((ring_data_t*)sqe->user_data);
-                *data = std::move(queue[i].data);
-                *sqe = queue[i].sqe;
-                sqe->user_data = (uint64_t)data;
-            }
-            queue.erase(queue.begin(), queue.begin()+i);
-        }
-        // We only want to offload sendmsg/recvmsg. Callbacks will be called in main thread
-        ring.submit();
-    }
-}
-
-void osd_messenger_t::init_iothreads()
-{
-    for (int i = 0; i < iothread_count; i++)
-    {
-        auto iot = new msgr_iothread_t();
-        iothreads.push_back(iot);
-        iot->add_to_ringloop(ringloop);
-    }
-}
-
-void osd_messenger_t::destroy_iothreads()
-{
-    if (iothreads.size())
-    {
-        for (auto iot: iothreads)
-        {
-            delete iot;
-        }
-        iothreads.clear();
-    }
-}

src/client/msgr_iothread.h

@@ -1,38 +0,0 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
-
-#include <mutex>
-#include <condition_variable>
-#include <thread>
-
-#include "ringloop.h"
-
-struct iothread_sqe_t
-{
-    io_uring_sqe sqe;
-    ring_data_t data;
-};
-
-class msgr_iothread_t
-{
-protected:
-    ring_loop_t ring;
-    ring_loop_i *outer_loop = NULL;
-    ring_data_t *outer_loop_data = NULL;
-    int eventfd = -1;
-    bool stopped = false;
-    std::mutex mu;
-    std::condition_variable cond;
-    std::vector<iothread_sqe_t> queue;
-    std::thread thread;
-
-    void run();
-public:
-
-    msgr_iothread_t();
-    ~msgr_iothread_t();
-
-    void add_sqe(io_uring_sqe & sqe);
-    void stop();
-    void add_to_ringloop(ring_loop_i *outer_loop);
-};

src/client/msgr_op.cpp

@@ -23,6 +23,10 @@ osd_op_t::~osd_op_t()
         // So we don't reuse it, but free it every time
         free(buf);
     }
+    if (enc_buf)
+    {
+        free(enc_buf);
+    }
 }
 
 bool osd_op_t::is_recovery_related()

src/client/msgr_op.h

@@ -3,6 +3,8 @@
 
 #pragma once
 
+#include <memory>
+
 #include <sys/uio.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -152,12 +154,13 @@ struct blockstore_op_t;
 
 struct osd_primary_op_data_t;
 
+struct inode_enc_t;
+
 struct __attribute__((visibility("default"))) osd_op_t
 {
     timespec tv_begin = { 0 }, tv_end = { 0 };
     uint64_t op_type = OSD_OP_IN;
-    uint64_t client_id = 0;
+    int peer_fd;
-    osd_num_t osd_num = 0;
     osd_any_op_t req;
     osd_any_reply_t reply;
     blockstore_op_t *bs_op = NULL;
@@ -168,6 +171,8 @@ struct __attribute__((visibility("default"))) osd_op_t
     unsigned bmp_data = 0;
     void *bitmap_buf = NULL;
     void *rmw_buf = NULL;
+    std::shared_ptr<inode_enc_t> enc;
+    uint8_t *enc_buf = NULL;
     osd_primary_op_data_t* op_data = NULL;
     std::function<void(osd_op_t*)> callback;
 

src/client/msgr_rdma.cpp

@@ -187,8 +187,6 @@ std::vector<msgr_rdma_context_t*> msgr_rdma_context_t::create_all(const std::vec
     ibv_device **raw_dev_list = NULL;
     ibv_device **dev_list = NULL;
     ibv_device *single_list[2] = {};
-    int up_ports = 0;
-    int single_port_num = 0;
 
     raw_dev_list = dev_list = ibv_get_device_list(NULL);
     if (!dev_list || !*dev_list)
@@ -223,7 +221,6 @@ std::vector<msgr_rdma_context_t*> msgr_rdma_context_t::create_all(const std::vec
         dev_list = single_list;
     }
 
-retry:
     for (int i = 0; dev_list[i]; ++i)
     {
         auto dev = dev_list[i];
@@ -261,9 +258,6 @@ retry:
                 fprintf(stderr, "RDMA device %s port %d GID %d does not exist\n", ibv_get_device_name(dev), port_num, sel_gid_index);
                 continue;
             }
-            up_ports++;
-            single_port_num = port_num;
-            single_list[0] = dev;
             uint32_t port_mtu = sel_mtu ? sel_mtu : ibv_mtu_to_bytes(portinfo.active_mtu);
 #ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
             if (sel_gid_index < 0)
@@ -304,14 +298,6 @@ cleanup_dev:
         ibv_close_device(context);
     }
 
-    if (!ret.size() && up_ports == 1 && dev_list != single_list)
-    {
-        // Auto-select the only available device/port if there is only one
-        dev_list = single_list;
-        sel_port_num = single_port_num;
-        goto retry;
-    }
-
 cleanup:
     if (raw_dev_list)
         ibv_free_device_list(raw_dev_list);
@@ -507,7 +493,7 @@ int msgr_rdma_connection_t::connect(msgr_rdma_address_t *dest)
     return 0;
 }
 
-bool osd_messenger_t::connect_rdma(uint64_t client_id, std::string rdma_address, uint64_t client_max_msg)
+bool osd_messenger_t::connect_rdma(int peer_fd, std::string rdma_address, uint64_t client_max_msg)
 {
     // Try to connect to the peer using RDMA
     msgr_rdma_address_t addr;
@@ -517,12 +503,12 @@ bool osd_messenger_t::connect_rdma(uint64_t client_id, std::string rdma_address,
         {
             client_max_msg = rdma_max_msg;
         }
-        auto cl = clients.at(client_id);
+        auto cl = clients.at(peer_fd);
         msgr_rdma_context_t *selected_ctx = choose_rdma_context(cl);
         if (!selected_ctx)
         {
             if (log_level > 0)
-                fprintf(stderr, "No RDMA context for peer %ju, using only TCP\n", client_id);
+                fprintf(stderr, "No RDMA context for peer %d, using only TCP\n", cl->peer_fd);
             return false;
         }
         msgr_rdma_connection_t *rdma_conn = msgr_rdma_connection_t::create(selected_ctx, rdma_max_send, rdma_max_recv, rdma_max_sge, client_max_msg);
@@ -533,13 +519,14 @@ bool osd_messenger_t::connect_rdma(uint64_t client_id, std::string rdma_address,
             {
                 delete rdma_conn;
                 fprintf(
-                    stderr, "Failed to connect RDMA queue pair to %s (client %ju)\n",
+                    stderr, "Failed to connect RDMA queue pair to %s (client %d)\n",
-                    addr.to_string().c_str(), client_id
+                    addr.to_string().c_str(), peer_fd
                 );
             }
             else
             {
                 // Remember connection, but switch to RDMA only after sending the configuration response
+                auto cl = clients.at(peer_fd);
                 cl->rdma_conn = rdma_conn;
                 cl->peer_state = PEER_RDMA_CONNECTING;
                 return true;
@@ -553,7 +540,7 @@ static void try_send_rdma_wr(osd_client_t *cl, ibv_sge *sge, int op_sge)
 {
     ibv_send_wr *bad_wr = NULL;
     ibv_send_wr wr = {
-        .wr_id = cl->client_id,
+        .wr_id = (uint64_t)(cl->peer_fd*2+1),
         .sg_list = sge,
         .num_sge = op_sge,
         .opcode = IBV_WR_SEND,
@@ -568,23 +555,28 @@ static void try_send_rdma_wr(osd_client_t *cl, ibv_sge *sge, int op_sge)
     cl->rdma_conn->cur_send++;
 }
 
-static int try_send_rdma_copy(osd_client_t *cl, uint8_t *dst, int dst_len)
+int osd_messenger_t::try_send_rdma_copy(osd_client_t *cl, uint8_t *dst, int dst_len)
 {
-    auto rc = cl->rdma_conn;
     int total_dst_len = dst_len;
-    while (dst_len > 0 && rc->send_pos < cl->send_list.size())
+    while (dst_len > 0 && cl->write_ops.size())
     {
-        iovec & iov = cl->send_list[rc->send_pos];
+        if (!cl->write_op)
-        uint32_t len = (uint32_t)(iov.iov_len-rc->send_buf_pos < dst_len
-            ? iov.iov_len-rc->send_buf_pos : dst_len);
-        memcpy(dst, (uint8_t*)iov.iov_base+rc->send_buf_pos, len);
-        dst += len;
-        dst_len -= len;
-        rc->send_buf_pos += len;
-        if (rc->send_buf_pos >= iov.iov_len)
         {
-            rc->send_pos++;
+            cl->write_op = cl->write_ops.front();
-            rc->send_buf_pos = 0;
+            cl->write_ops.pop_front();
+        }
+        osd_op_t *op = cl->write_op;
+        size_t copied = op_copy_to(cl, dst, dst_len);
+        if (!copied)
+        {
+            break;
+        }
+        dst += copied;
+        dst_len -= copied;
+        if (!cl->write_op && op->op_type == OSD_OP_IN)
+        {
+            // this is a reply, free the op after sending it
+            cl->send_free_ops.push_back(op);
         }
     }
     return total_dst_len-dst_len;
@@ -612,9 +604,7 @@ void osd_messenger_t::try_send_rdma(osd_client_t *cl)
     while (!rc->send_out_full && copied > 0 && rc->cur_send < rc->max_send)
     {
         dst = (uint8_t*)rc->send_out.buf + rc->send_out_pos;
-        dst_len = (rc->send_out_pos >= rc->send_done_pos
+        dst_len = (rc->send_out_pos < rc->send_out_size ? rc->send_out_size-rc->send_out_pos : rc->send_done_pos-rc->send_out_pos);
-            ? rc->send_out_size-rc->send_out_pos
-            : rc->send_done_pos-rc->send_out_pos);
         if (dst_len > rc->max_msg)
             dst_len = rc->max_msg;
         copied = try_send_rdma_copy(cl, dst, dst_len);
@@ -624,7 +614,7 @@ void osd_messenger_t::try_send_rdma(osd_client_t *cl)
             if (rc->send_out_pos == rc->send_out_size)
                 rc->send_out_pos = 0;
             assert(rc->send_out_pos < rc->send_out_size);
-            if (rc->send_out_pos == rc->send_done_pos)
+            if (rc->send_out_pos >= rc->send_done_pos)
                 rc->send_out_full = true;
             ibv_sge sge = {
                 .addr = (uintptr_t)dst,
@@ -632,7 +622,7 @@ void osd_messenger_t::try_send_rdma(osd_client_t *cl)
                 .lkey = rc->send_out.mr->lkey,
             };
             try_send_rdma_wr(cl, &sge, 1);
-            rc->send_sizes.push_back(copied);
+            cl->send_free_ops.push_back(NULL); // end marker
         }
     }
 }
@@ -646,7 +636,7 @@ static void try_recv_rdma_wr(osd_client_t *cl, void *buf)
     };
     ibv_recv_wr *bad_wr = NULL;
     ibv_recv_wr wr = {
-        .wr_id = cl->client_id,
+        .wr_id = (uint64_t)(cl->peer_fd*2),
         .sg_list = &sge,
         .num_sge = 1,
     };
@@ -703,28 +693,25 @@ void osd_messenger_t::handle_rdma_events(msgr_rdma_context_t *rdma_context)
         event_count = ibv_poll_cq(rdma_context->cq, RDMA_EVENTS_AT_ONCE, wc);
         for (int i = 0; i < event_count; i++)
         {
-            uint64_t client_id = wc[i].wr_id;
+            int client_id = wc[i].wr_id >> 1;
-            bool is_send = wc[i].opcode == IBV_WC_SEND;
+            bool is_send = wc[i].wr_id & 1;
             auto cl_it = clients.find(client_id);
             if (cl_it == clients.end())
             {
                 continue;
             }
             osd_client_t *cl = cl_it->second;
-            if (cl->peer_state == PEER_STOPPED)
-            {
-                continue;
-            }
             auto rc = cl->rdma_conn;
             if (wc[i].status != IBV_WC_SUCCESS)
             {
-                fprintf(stderr, "RDMA work request failed for client %ju", client_id);
+                fprintf(stderr, "RDMA work request failed for client %d", client_id);
                 if (cl->osd_num)
                 {
                     fprintf(stderr, " (OSD %ju)", cl->osd_num);
                 }
                 fprintf(stderr, " with status: %s, stopping client\n", ibv_wc_status_str(wc[i].status));
                 stop_client(client_id);
+                clear_immediate_ops(client_id);
                 continue;
             }
             if (!is_send)
@@ -735,6 +722,8 @@ void osd_messenger_t::handle_rdma_events(msgr_rdma_context_t *rdma_context)
                 rc->cur_recv--;
                 if (!handle_read_buffer(cl, rc->recv_buffers[rc->next_recv_buf], wc[i].byte_len))
                 {
+                    // handle_read_buffer may stop the client
+                    clear_immediate_ops(client_id);
                     continue;
                 }
                 try_recv_rdma_wr(cl, rc->recv_buffers[rc->next_recv_buf]);
@@ -743,67 +732,21 @@ void osd_messenger_t::handle_rdma_events(msgr_rdma_context_t *rdma_context)
             else
             {
                 rc->cur_send--;
-                uint64_t sent_size = rc->send_sizes.at(0);
+                uint64_t sent_size = wc[i].byte_len;
-                rc->send_sizes.erase(rc->send_sizes.begin(), rc->send_sizes.begin()+1);
                 rc->send_done_pos += sent_size;
                 rc->send_out_full = false;
                 if (rc->send_done_pos == rc->send_out_size)
                     rc->send_done_pos = 0;
                 assert(rc->send_done_pos < rc->send_out_size);
-                int send_pos = 0, send_buf_pos = 0;
+                while (cl->send_free_ops.front())
-                while (sent_size > 0)
                 {
-                    if (sent_size >= cl->send_list.at(send_pos).iov_len)
+                    delete cl->send_free_ops.front();
-                    {
+                    cl->send_free_ops.pop_front();
-                        sent_size -= cl->send_list[send_pos].iov_len;
-                        send_pos++;
-                    }
-                    else
-                    {
-                        send_buf_pos = sent_size;
-                        sent_size = 0;
-                    }
-                }
-                assert(rc->send_pos >= send_pos);
-                if (rc->send_pos == send_pos)
-                {
-                    rc->send_buf_pos -= send_buf_pos;
-                }
-                rc->send_pos -= send_pos;
-                for (int i = 0; i < send_pos; i++)
-                {
-                    if (cl->outbox[i].flags & MSGR_SENDP_FREE)
-                    {
-                        // Reply fully sent
-                        delete cl->outbox[i].op;
-                    }
-                }
-                if (send_pos > 0)
-                {
-                    cl->send_list.erase(cl->send_list.begin(), cl->send_list.begin()+send_pos);
-                    cl->outbox.erase(cl->outbox.begin(), cl->outbox.begin()+send_pos);
-                }
-                if (send_buf_pos > 0)
-                {
-                    cl->send_list[0].iov_base = (uint8_t*)cl->send_list[0].iov_base + send_buf_pos;
-                    cl->send_list[0].iov_len -= send_buf_pos;
                 }
+                cl->send_free_ops.pop_front();
                 try_send_rdma(cl);
             }
         }
     } while (event_count > 0);
     handle_immediate_ops();
 }
-
-void osd_messenger_t::destroy_rdma_conn(msgr_rdma_connection_t *rdma_conn)
-{
-    if (rdma_conn->cmid)
-    {
-        auto rdma_it = rdmacm_connections.find(rdma_conn->cmid);
-        if (rdma_it != rdmacm_connections.end() && rdma_it->second->rdma_conn == rdma_conn)
-        {
-            rdmacm_connections.erase(rdma_it);
-        }
-    }
-    delete rdma_conn;
-}

src/client/msgr_rdma.h

@@ -10,6 +10,8 @@
 #include <vector>
 #include "addr_util.h"
 
+struct osd_op_t;
+
 struct msgr_rdma_address_t
 {
     ibv_gid gid;
@@ -72,9 +74,8 @@ struct msgr_rdma_connection_t
     int cur_send = 0, cur_recv = 0;
     int send_pos = 0, send_buf_pos = 0;
     int next_recv_buf = 0;
-    std::vector<void*> recv_buffers;
+    std::vector<uint8_t*> recv_buffers;
     msgr_rdma_buf_t recv_buf;
-    std::vector<uint64_t> send_sizes;
     msgr_rdma_buf_t send_out;
     int send_out_pos = 0, send_done_pos = 0, send_out_size = 0;
     bool send_out_full = false;

src/client/msgr_rdmacm.cpp

@@ -11,7 +11,7 @@
 struct rdmacm_connecting_t
 {
     rdma_cm_id *cmid = NULL;
-    uint64_t client_id = 0;
+    int peer_fd = -1;
     osd_num_t peer_osd = 0;
     std::string addr;
     sockaddr_storage parsed_addr = {};
@@ -117,9 +117,9 @@ void osd_messenger_t::handle_rdmacm_events()
             auto cli_it = rdmacm_connections.find(ev->id);
             if (cli_it != rdmacm_connections.end())
             {
-                fprintf(stderr, "Received %s event for client %ju, closing connection\n",
+                fprintf(stderr, "Received %s event for peer %d, closing connection\n",
-                    event_type_name, cli_it->second->client_id);
+                    event_type_name, cli_it->second->peer_fd);
-                stop_client(cli_it->second->client_id);
+                stop_client(cli_it->second->peer_fd);
             }
             else if (rdmacm_connecting.find(ev->id) != rdmacm_connecting.end())
             {
@@ -265,6 +265,14 @@ msgr_rdma_context_t* osd_messenger_t::rdmacm_create_qp(rdma_cm_id *cmid)
 
 void osd_messenger_t::rdmacm_accept(rdma_cm_event *ev)
 {
+    // Make a fake FD (FIXME: do not use FDs for identifying clients!)
+    int fake_fd = socket(AF_INET, SOCK_STREAM, 0);
+    if (fake_fd < 0)
+    {
+        fprintf(stderr, "Failed to allocate a fake socket for RDMA-CM client: %s (code %d)\n", strerror(errno), errno);
+        rdma_destroy_id(ev->id);
+        return;
+    }
     auto rdma_context = rdmacm_create_qp(ev->id);
     if (!rdma_context)
     {
@@ -289,12 +297,12 @@ void osd_messenger_t::rdmacm_accept(rdma_cm_event *ev)
     // Wait for RDMA_CM_ESTABLISHED, and enable the connection only after it
     auto conn = new rdmacm_connecting_t;
     conn->cmid = ev->id;
-    conn->client_id = next_client_id++;
+    conn->peer_fd = fake_fd;
     conn->parsed_addr = *(sockaddr_storage*)rdma_get_peer_addr(ev->id);
     conn->rdma_context = rdma_context;
     rdmacm_set_conn_timeout(conn);
     rdmacm_connecting[ev->id] = conn;
-    fprintf(stderr, "[OSD %ju] new client %ju: connection from %s via RDMA-CM\n", this->osd_num, conn->client_id,
+    fprintf(stderr, "[OSD %ju] new client %d: connection from %s via RDMA-CM\n", this->osd_num, conn->peer_fd,
         addr_to_string(conn->parsed_addr).c_str());
 }
 
@@ -324,6 +332,8 @@ void osd_messenger_t::rdmacm_on_connect_peer_error(rdma_cm_id *cmid, int res)
     auto peer_osd = conn->peer_osd;
     if (conn->timeout_id >= 0)
         tfd->clear_timer(conn->timeout_id);
+    if (conn->peer_fd >= 0)
+        close(conn->peer_fd);
     if (conn->rdma_context)
         conn->rdma_context->reserve_cqe(-rdma_max_send-rdma_max_recv);
     if (conn->cmid)
@@ -344,7 +354,7 @@ void osd_messenger_t::rdmacm_on_connect_peer_error(rdma_cm_id *cmid, int res)
         else
         {
             // TCP is disabled
-            on_connect_peer(peer_osd, res == 0 ? -EINVAL : (res > 0 ? -res : res), 0);
+            on_connect_peer(peer_osd, res == 0 ? -EINVAL : (res > 0 ? -res : res));
         }
     }
 }
@@ -355,7 +365,7 @@ void osd_messenger_t::rdmacm_try_connect_peer(uint64_t peer_osd, const std::stri
     if (!string_to_addr(addr, false, rdmacm_port, &sa))
     {
         fprintf(stderr, "Address %s is invalid\n", addr.c_str());
-        on_connect_peer(peer_osd, -EINVAL, 0);
+        on_connect_peer(peer_osd, -EINVAL);
         return;
     }
     rdma_cm_id *cmid = NULL;
@@ -366,7 +376,17 @@ void osd_messenger_t::rdmacm_try_connect_peer(uint64_t peer_osd, const std::stri
         if (!disable_tcp)
             try_connect_peer_tcp(peer_osd, addr.c_str(), fallback_tcp_port);
         else
-            on_connect_peer(peer_osd, res, 0);
+            on_connect_peer(peer_osd, res);
+        return;
+    }
+    // Make a fake FD (FIXME: do not use FDs for identifying clients!)
+    int fake_fd = socket(AF_INET, SOCK_STREAM, 0);
+    if (fake_fd < 0)
+    {
+        int res = -errno;
+        rdma_destroy_id(cmid);
+        // Can't create socket, pointless to try TCP
+        on_connect_peer(peer_osd, res);
         return;
     }
     if (log_level > 0)
@@ -374,7 +394,7 @@ void osd_messenger_t::rdmacm_try_connect_peer(uint64_t peer_osd, const std::stri
     auto conn = new rdmacm_connecting_t;
     rdmacm_connecting[cmid] = conn;
     conn->cmid = cmid;
-    conn->client_id = next_client_id++;
+    conn->peer_fd = fake_fd;
     conn->peer_osd = peer_osd;
     conn->addr = addr;
     conn->parsed_addr = sa;
@@ -491,13 +511,13 @@ void osd_messenger_t::rdmacm_established(rdma_cm_event *ev)
     auto cl = new osd_client_t();
     cl->peer_addr = conn->parsed_addr;
     cl->peer_port = conn->rdmacm_port;
-    cl->client_id = conn->client_id;
+    cl->peer_fd = conn->peer_fd;
     cl->peer_state = PEER_RDMA;
     cl->connect_timeout_id = -1;
     cl->osd_num = peer_osd;
-    cl->in_buf = malloc_or_die(receive_buffer_size);
+    cl->in_buf = (uint8_t*)malloc_or_die(receive_buffer_size);
     cl->rdma_conn = rc;
-    clients[conn->client_id] = cl;
+    clients[conn->peer_fd] = cl;
     if (conn->timeout_id >= 0)
     {
         tfd->clear_timer(conn->timeout_id);

src/client/msgr_receive.cpp

@@ -1,22 +1,26 @@
 // Copyright (c) Vitaliy Filippov, 2019+
 // License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
 
+#define _XOPEN_SOURCE
+#include <limits.h>
 #include "messenger.h"
-#include "msgr_iothread.h"
 
 void osd_messenger_t::read_requests()
 {
     for (int i = 0; i < read_ready_clients.size(); i++)
     {
-        uint64_t client_id = read_ready_clients[i];
+        int peer_fd = read_ready_clients[i];
-        auto cl_it = clients.find(client_id);
+        auto cl_it = clients.find(peer_fd);
-        if (cl_it == clients.end() || !cl_it->second || cl_it->second->read_msg.msg_iovlen ||
+        if (cl_it == clients.end() || !cl_it->second || cl_it->second->read_msg.msg_iovlen)
-            cl_it->second->peer_state != PEER_CONNECTED)
         {
             continue;
         }
         auto cl = cl_it->second;
-        if (cl->read_remaining < receive_buffer_size)
+        if (cl->read_op && cl->read_op_size-(cl->read_op_pos-OSD_PACKET_SIZE) >= receive_buffer_size)
+        {
+            op_get_read_buffers(cl, cl->recv_list);
+        }
+        if (!cl->recv_list.size())
         {
             cl->read_iov.iov_base = cl->in_buf;
             cl->read_iov.iov_len = receive_buffer_size;
@@ -26,14 +30,15 @@ void osd_messenger_t::read_requests()
         else
         {
             cl->read_iov.iov_base = 0;
-            cl->read_iov.iov_len = cl->read_remaining;
+            cl->read_iov.iov_len = 0;
-            cl->read_msg.msg_iov = cl->recv_list.get_iovec();
+            cl->read_msg.msg_iov = cl->recv_list.data();
-            cl->read_msg.msg_iovlen = cl->recv_list.get_size();
+            cl->read_msg.msg_iovlen = cl->recv_list.size();
         }
+        assert(!cl->read_op || cl->read_op_pos < OSD_PACKET_SIZE || cl->read_op_size >= (cl->read_op_pos-OSD_PACKET_SIZE));
         cl->refs++;
         if (ringloop && !use_sync_send_recv)
         {
-            auto iothread = iothreads.size() ? iothreads[cl->peer_fd % iothreads.size()] : NULL;
+            auto iothread = iothreads.size() ? iothreads[peer_fd % iothreads.size()] : NULL;
             io_uring_sqe sqe_local;
             ring_data_t data_local;
             io_uring_sqe* sqe = (iothread ? &sqe_local : ringloop->get_sqe());
@@ -51,7 +56,7 @@ void osd_messenger_t::read_requests()
             }
             ring_data_t* data = ((ring_data_t*)sqe->user_data);
             data->callback = [this, cl](ring_data_t *data) { handle_read(data->res, cl); };
-            io_uring_prep_recvmsg(sqe, cl->peer_fd, &cl->read_msg, 0);
+            io_uring_prep_recvmsg(sqe, peer_fd, &cl->read_msg, cl->recv_list.size() ? MSG_WAITALL : 0);
             if (iothread)
             {
                 iothread->add_sqe(sqe_local);
@@ -59,7 +64,7 @@ void osd_messenger_t::read_requests()
         }
         else
         {
-            int result = recvmsg(cl->peer_fd, &cl->read_msg, 0);
+            int result = recvmsg(peer_fd, &cl->read_msg, 0);
             if (result < 0)
             {
                 result = -errno;
@@ -71,89 +76,110 @@ void osd_messenger_t::read_requests()
     read_ready_clients.clear();
 }
 
-bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
+void osd_messenger_t::handle_read(int result, osd_client_t *cl)
 {
-    bool ret = false;
+    int peer_fd = cl->peer_fd;
-    cl->read_msg.msg_iovlen = 0;
     cl->refs--;
-    if (cl->peer_state == PEER_RDMA)
-    {
-        return true;
-    }
     if (cl->peer_state == PEER_STOPPED)
     {
         if (cl->refs <= 0)
         {
-            destroy_client(cl);
+            delete cl;
         }
-        return false;
+        return;
     }
     if (result <= 0 && result != -EAGAIN && result != -EINTR)
     {
         // this is a client socket, so don't panic on error. just disconnect it
         if (result != 0)
         {
-            fprintf(stderr, "Client %ju socket read error: %d (%s). Disconnecting client\n", cl->client_id, -result, strerror(-result));
+            fprintf(stderr, "Client %d socket read error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
         }
-        stop_client(cl->client_id);
+        stop_client(cl->peer_fd);
-        return false;
+        return;
-    }
-    if (result == -EAGAIN || result == -EINTR || result < cl->read_iov.iov_len)
-    {
-        cl->read_ready--;
-        if (cl->read_ready > 0)
-            read_ready_clients.push_back(cl->client_id);
-    }
-    else
-    {
-        read_ready_clients.push_back(cl->client_id);
     }
+    bool full_read = false;
     if (result > 0)
     {
         if (cl->read_iov.iov_base == cl->in_buf)
         {
+            full_read = result >= cl->read_iov.iov_len;
             if (!handle_read_buffer(cl, cl->in_buf, result))
             {
+                clear_immediate_ops(peer_fd);
                 handle_immediate_ops();
-                return false;
+                return;
             }
         }
         else
         {
+            // Reset OSD ping state
+            cl->ping_time_remaining = 0;
+            cl->idle_time_remaining = osd_idle_timeout;
             // Long data
-            cl->read_remaining -= result;
+            size_t i = 0;
-            cl->recv_list.eat(result);
+            while (i < cl->recv_list.size() && result >= cl->recv_list[i].iov_len)
-            if (cl->recv_list.done >= cl->recv_list.count)
             {
-                if (!handle_finished_read(cl))
+                result -= cl->recv_list[i].iov_len;
-                {
+                i++;
-                    handle_immediate_ops();
+            }
-                    return false;
+            if (i < cl->recv_list.size())
-                }
+            {
+                cl->recv_list[i].iov_base += result;
+                cl->recv_list[i].iov_len -= result;
+            }
+            else
+            {
+                full_read = true;
+            }
+            cl->recv_list.erase(cl->recv_list.begin(), cl->recv_list.begin()+i);
+            if (!cl->recv_list.size())
+            {
+                handle_finished_op(cl);
             }
         }
-        if (result >= cl->read_iov.iov_len)
+    }
-        {
+    cl->read_msg.msg_iovlen = 0;
-            ret = true;
+    if (result == -EAGAIN || result == -EINTR || !full_read)
-        }
+    {
+        cl->read_ready--;
+        if (cl->read_ready > 0)
+            read_ready_clients.push_back(cl->peer_fd);
+    }
+    else
+    {
+        read_ready_clients.push_back(cl->peer_fd);
     }
     handle_immediate_ops();
-    return ret;
+}
+
+void osd_messenger_t::clear_immediate_ops(int peer_fd)
+{
+    size_t i = 0, j = 0;
+    while (i < set_immediate_ops.size())
+    {
+        if (set_immediate_ops[i]->peer_fd == peer_fd && set_immediate_ops[i]->op_type == OSD_OP_IN)
+        {
+            delete set_immediate_ops[i];
+        }
+        else
+        {
+            if (i != j)
+                set_immediate_ops[j] = set_immediate_ops[i];
+            j++;
+        }
+        i++;
+    }
+    set_immediate_ops.resize(j);
 }
 
 void osd_messenger_t::handle_immediate_ops()
 {
-    while (set_immediate_ops.size())
+    for (auto op: set_immediate_ops)
     {
-        auto op = set_immediate_ops.front();
-        set_immediate_ops.pop_front();
         if (op->op_type == OSD_OP_IN)
         {
-            auto cl_it = clients.find(op->client_id);
+            exec_op(op);
-            if (cl_it != clients.end() && cl_it->second->peer_state != PEER_STOPPED)
-                exec_op(op);
-            else
-                delete op;
         }
         else
         {
@@ -161,115 +187,98 @@ void osd_messenger_t::handle_immediate_ops()
             std::function<void(osd_op_t*)>(op->callback)(op);
         }
     }
+    set_immediate_ops.clear();
 }
 
-bool osd_messenger_t::handle_read_buffer(osd_client_t *cl, void *curbuf, int remain)
+bool osd_messenger_t::handle_read_buffer(osd_client_t *cl, uint8_t *curbuf, size_t bufsize)
-{
-    // Compose operation(s) from the buffer
-    while (remain > 0)
-    {
-        if (!cl->read_op)
-        {
-            cl->read_op = new osd_op_t;
-            cl->read_op->client_id = cl->client_id;
-            cl->read_op->op_type = OSD_OP_IN;
-            cl->recv_list.push_back(cl->read_op->req.buf, OSD_PACKET_SIZE);
-            cl->read_remaining = OSD_PACKET_SIZE;
-            cl->read_state = CL_READ_HDR;
-        }
-        while (cl->recv_list.done < cl->recv_list.count && remain > 0)
-        {
-            iovec* cur = cl->recv_list.get_iovec();
-            if (cur->iov_len > remain)
-            {
-                memcpy(cur->iov_base, curbuf, remain);
-                cl->read_remaining -= remain;
-                cur->iov_len -= remain;
-                cur->iov_base = (uint8_t*)cur->iov_base + remain;
-                remain = 0;
-            }
-            else
-            {
-                memcpy(cur->iov_base, curbuf, cur->iov_len);
-                curbuf = (uint8_t*)curbuf + cur->iov_len;
-                cl->read_remaining -= cur->iov_len;
-                remain -= cur->iov_len;
-                cur->iov_len = 0;
-                cl->recv_list.done++;
-            }
-        }
-        if (cl->recv_list.done >= cl->recv_list.count)
-        {
-            if (!handle_finished_read(cl))
-            {
-                return false;
-            }
-        }
-    }
-    return true;
-}
-
-bool osd_messenger_t::handle_finished_read(osd_client_t *cl)
 {
     // Reset OSD ping state
     cl->ping_time_remaining = 0;
     cl->idle_time_remaining = osd_idle_timeout;
-    cl->recv_list.reset();
+    // Compose operation(s) from the buffer
-    if (cl->read_state == CL_READ_HDR)
+    size_t done = 0;
+    while (done < bufsize)
     {
-        if (cl->read_op->req.hdr.magic == SECONDARY_OSD_REPLY_MAGIC)
+        if (!cl->read_op)
-            return handle_reply_hdr(cl);
-        else if (cl->read_op->req.hdr.magic == SECONDARY_OSD_OP_MAGIC)
         {
-            if (cl->check_sequencing)
+            cl->read_op = new osd_op_t;
+            cl->read_op->peer_fd = cl->peer_fd;
+            cl->read_op->op_type = OSD_OP_IN;
+            cl->read_op_pos = 0;
+            cl->read_op_size = 0;
+            cl->read_op_inline_decrypt_pos = (size_t)-1;
+        }
+        if (cl->read_op_pos < OSD_PACKET_SIZE)
+        {
+            int len = OSD_PACKET_SIZE - cl->read_op_pos;
+            if (len > bufsize-done)
+                len = bufsize-done;
+            memcpy(cl->read_op->req.buf + cl->read_op_pos, curbuf+done, len);
+            done += len;
+            cl->read_op_pos += len;
+            if (cl->read_op_pos < OSD_PACKET_SIZE)
+                return true;
+            if (!handle_hdr(cl))
             {
-                if (cl->read_op->req.hdr.id != cl->read_op_id)
+                stop_client(cl->peer_fd);
-                {
+                return false;
-                    fprintf(stderr, "Warning: operation sequencing is broken on client %ju: expected num %ju, got %ju, stopping client\n", cl->client_id, cl->read_op_id, cl->read_op->req.hdr.id);
-                    stop_client(cl->client_id);
-                    return false;
-                }
-                cl->read_op_id++;
             }
-            handle_op_hdr(cl);
         }
-        else
+        op_copy_from(cl, curbuf, bufsize, done);
-        {
-            fprintf(stderr, "Received garbage: magic=%jx id=%ju opcode=%jx from client %ju\n", cl->read_op->req.hdr.magic, cl->read_op->req.hdr.id, cl->read_op->req.hdr.opcode, cl->client_id);
-            stop_client(cl->client_id);
-            return false;
-        }
-    }
-    else if (cl->read_state == CL_READ_DATA)
-    {
-        // Operation is ready
-        cl->received_ops.push_back(cl->read_op);
-        set_immediate_ops.push_back(cl->read_op);
-        cl->read_op = NULL;
-        cl->read_state = 0;
-    }
-    else if (cl->read_state == CL_READ_REPLY_DATA)
-    {
-        // Reply is ready
-        handle_reply_ready(cl->read_op);
-        cl->read_op = NULL;
-        cl->read_state = 0;
-    }
-    else
-    {
-        assert(0);
     }
     return true;
 }
 
-void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
+bool osd_messenger_t::handle_hdr(osd_client_t *cl)
+{
+    if (cl->read_op->req.hdr.magic == SECONDARY_OSD_REPLY_MAGIC)
+    {
+        auto req_it = cl->sent_ops.find(cl->read_op->req.hdr.id);
+        if (req_it == cl->sent_ops.end())
+        {
+            // Command out of sync. Drop connection
+            fprintf(stderr, "Client %d command out of sync: id %ju\n", cl->peer_fd, cl->read_op->req.hdr.id);
+            return false;
+        }
+        osd_op_t *op = req_it->second;
+        memcpy(op->reply.buf, cl->read_op->req.buf, OSD_PACKET_SIZE);
+        if (!allocate_reply_buffers(cl, op))
+        {
+            return false;
+        }
+        cl->sent_ops.erase(req_it);
+        delete cl->read_op;
+        cl->read_op = op;
+    }
+    else if (cl->read_op->req.hdr.magic == SECONDARY_OSD_OP_MAGIC)
+    {
+        if (cl->check_sequencing)
+        {
+            if (cl->read_op->req.hdr.id != cl->read_op_id)
+            {
+                fprintf(stderr, "Warning: operation sequencing is broken on client %d: expected num %ju, got %ju, stopping client\n", cl->peer_fd, cl->read_op_id, cl->read_op->req.hdr.id);
+                return false;
+            }
+            cl->read_op_id++;
+        }
+        if (!allocate_op_buffers(cl))
+        {
+            return false;
+        }
+    }
+    else
+    {
+        fprintf(stderr, "Received garbage: magic=%jx id=%ju opcode=%jx from %d\n", cl->read_op->req.hdr.magic, cl->read_op->req.hdr.id, cl->read_op->req.hdr.opcode, cl->peer_fd);
+        return false;
+    }
+    return true;
+}
+
+bool osd_messenger_t::allocate_op_buffers(osd_client_t *cl)
 {
     osd_op_t *cur_op = cl->read_op;
-    if (cur_op->req.hdr.opcode == OSD_OP_SEC_READ)
+    cl->read_op_size = 0;
-    {
+    if (cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
-        cl->read_remaining = 0;
-    }
-    else if (cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
         cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE)
     {
         if (cur_op->req.sec_rw.attr_len > 0)
@@ -278,14 +287,12 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
                 cur_op->bitmap = cur_op->rmw_buf = malloc_or_die(cur_op->req.sec_rw.attr_len);
             else
                 cur_op->bitmap = &cur_op->bmp_data;
-            cl->recv_list.push_back(cur_op->bitmap, cur_op->req.sec_rw.attr_len);
         }
         if (cur_op->req.sec_rw.len > 0)
         {
             cur_op->buf = memalign_or_die(MEM_ALIGNMENT, cur_op->req.sec_rw.len);
-            cl->recv_list.push_back(cur_op->buf, cur_op->req.sec_rw.len);
         }
-        cl->read_remaining = cur_op->req.sec_rw.len + cur_op->req.sec_rw.attr_len;
+        cl->read_op_size = cur_op->req.sec_rw.len + cur_op->req.sec_rw.attr_len;
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
         cur_op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK)
@@ -293,27 +300,24 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
         if (cur_op->req.sec_stab.len > 0)
         {
             cur_op->buf = memalign_or_die(MEM_ALIGNMENT, cur_op->req.sec_stab.len);
-            cl->recv_list.push_back(cur_op->buf, cur_op->req.sec_stab.len);
         }
-        cl->read_remaining = cur_op->req.sec_stab.len;
+        cl->read_op_size = cur_op->req.sec_stab.len;
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
     {
         if (cur_op->req.sec_read_bmp.len > 0)
         {
             cur_op->buf = memalign_or_die(MEM_ALIGNMENT, cur_op->req.sec_read_bmp.len);
-            cl->recv_list.push_back(cur_op->buf, cur_op->req.sec_read_bmp.len);
         }
-        cl->read_remaining = cur_op->req.sec_read_bmp.len;
+        cl->read_op_size = cur_op->req.sec_read_bmp.len;
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_WRITE)
     {
         if (cur_op->req.rw.len > 0)
         {
             cur_op->buf = memalign_or_die(MEM_ALIGNMENT, cur_op->req.rw.len);
-            cl->recv_list.push_back(cur_op->buf, cur_op->req.rw.len);
         }
-        cl->read_remaining = cur_op->req.rw.len;
+        cl->read_op_size = cur_op->req.rw.len;
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_SHOW_CONFIG)
     {
@@ -321,44 +325,15 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
         {
             cur_op->buf = malloc_or_die(cur_op->req.show_conf.json_len+1);
             ((uint8_t*)cur_op->buf)[cur_op->req.show_conf.json_len] = 0;
-            cl->recv_list.push_back(cur_op->buf, cur_op->req.show_conf.json_len);
         }
-        cl->read_remaining = cur_op->req.show_conf.json_len;
+        cl->read_op_size = cur_op->req.show_conf.json_len;
-    }
-    /*else if (cur_op->req.hdr.opcode == OSD_OP_READ ||
-        cur_op->req.hdr.opcode == OSD_OP_SCRUB ||
-        cur_op->req.hdr.opcode == OSD_OP_DESCRIBE)
-    {
-        cl->read_remaining = 0;
-    }*/
-    if (cl->read_remaining > 0)
-    {
-        // Read data
-        cl->read_state = CL_READ_DATA;
-    }
-    else
-    {
-        // Operation is ready
-        cl->received_ops.push_back(cur_op);
-        set_immediate_ops.push_back(cur_op);
-        cl->read_op = NULL;
-        cl->read_state = 0;
     }
+    return true;
 }
 
-bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
+bool osd_messenger_t::allocate_reply_buffers(osd_client_t *cl, osd_op_t *op)
 {
-    auto req_it = cl->sent_ops.find(cl->read_op->req.hdr.id);
+    cl->read_op_size = 0;
-    if (req_it == cl->sent_ops.end() || req_it->second->req.hdr.opcode != cl->read_op->req.hdr.opcode)
-    {
-        // Command out of sync. Drop connection
-        fprintf(stderr, "Client %ju command out of sync: id %ju\n", cl->client_id, cl->read_op->req.hdr.id);
-        stop_client(cl->client_id);
-        return false;
-    }
-    osd_op_t *op = req_it->second;
-    memcpy(op->reply.buf, cl->read_op->req.buf, OSD_PACKET_SIZE);
-    cl->sent_ops.erase(req_it);
     if (op->reply.hdr.opcode == OSD_OP_SEC_READ || op->reply.hdr.opcode == OSD_OP_READ)
     {
         // Read data. In this case we assume that the buffer is preallocated by the caller (!)
@@ -367,99 +342,297 @@ bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
         if (op->reply.hdr.retval >= 0 && (op->reply.hdr.retval != expected_size || bmp_len > op->bitmap_len))
         {
             // Check reply length to not overflow the buffer
-            fprintf(stderr, "Client %ju read reply of different length: expected %u+%u, got %jd+%u\n",
+            fprintf(stderr, "Client %d read reply of different length: expected %u+%u, got %jd+%u\n",
-                cl->client_id, expected_size, op->bitmap_len, op->reply.hdr.retval, bmp_len);
+                cl->peer_fd, expected_size, op->bitmap_len, op->reply.hdr.retval, bmp_len);
-            cl->sent_ops[op->req.hdr.id] = op;
-            stop_client(cl->client_id);
             return false;
         }
-        if (op->reply.hdr.retval >= 0 && bmp_len > 0)
+        if (bmp_len > 0)
         {
             assert(op->bitmap);
-            cl->recv_list.push_back(op->bitmap, bmp_len);
+            cl->read_op_size += bmp_len;
-            cl->read_remaining += bmp_len;
         }
         if (op->reply.hdr.retval > 0)
         {
             assert(op->iov.count > 0);
-            cl->recv_list.append(op->iov);
+            cl->read_op_size += op->reply.hdr.retval;
-            cl->read_remaining += op->reply.hdr.retval;
         }
-        if (cl->read_remaining == 0)
-        {
-            goto reuse;
-        }
-        delete cl->read_op;
-        cl->read_op = op;
-        cl->read_state = CL_READ_REPLY_DATA;
     }
     else if (op->reply.hdr.opcode == OSD_OP_SEC_LIST && op->reply.hdr.retval > 0)
     {
         assert(!op->iov.count);
-        delete cl->read_op;
+        cl->read_op_size = sizeof(obj_ver_id) * op->reply.hdr.retval;
-        cl->read_op = op;
+        op->buf = memalign_or_die(MEM_ALIGNMENT, cl->read_op_size);
-        cl->read_state = CL_READ_REPLY_DATA;
-        cl->read_remaining = sizeof(obj_ver_id) * op->reply.hdr.retval;
-        op->buf = memalign_or_die(MEM_ALIGNMENT, cl->read_remaining);
-        cl->recv_list.push_back(op->buf, cl->read_remaining);
     }
     else if (op->reply.hdr.opcode == OSD_OP_SEC_READ_BMP && op->reply.hdr.retval > 0)
     {
         assert(!op->iov.count);
-        delete cl->read_op;
+        cl->read_op_size = op->reply.hdr.retval;
-        cl->read_op = op;
-        cl->read_state = CL_READ_REPLY_DATA;
-        cl->read_remaining = op->reply.hdr.retval;
         free(op->buf);
-        op->buf = memalign_or_die(MEM_ALIGNMENT, cl->read_remaining);
+        op->buf = memalign_or_die(MEM_ALIGNMENT, cl->read_op_size);
-        cl->recv_list.push_back(op->buf, cl->read_remaining);
     }
     else if (op->reply.hdr.opcode == OSD_OP_SHOW_CONFIG && op->reply.hdr.retval > 0)
     {
-        delete cl->read_op;
+        cl->read_op_size = op->reply.hdr.retval;
-        cl->read_op = op;
-        cl->read_state = CL_READ_REPLY_DATA;
-        cl->read_remaining = op->reply.hdr.retval;
         free(op->buf);
         op->buf = malloc_or_die(op->reply.hdr.retval);
-        cl->recv_list.push_back(op->buf, op->reply.hdr.retval);
     }
     else if (op->reply.hdr.opcode == OSD_OP_DESCRIBE && op->reply.describe.result_bytes > 0)
     {
-        delete cl->read_op;
+        cl->read_op_size = op->reply.describe.result_bytes;
-        cl->read_op = op;
-        cl->read_state = CL_READ_REPLY_DATA;
-        cl->read_remaining = op->reply.describe.result_bytes;
         free(op->buf);
         op->buf = malloc_or_die(op->reply.describe.result_bytes);
-        cl->recv_list.push_back(op->buf, op->reply.describe.result_bytes);
-    }
-    else
-    {
-reuse:
-        // It's fine to reuse cl->read_op for the next reply
-        handle_reply_ready(op);
-        cl->recv_list.push_back(cl->read_op->req.buf, OSD_PACKET_SIZE);
-        cl->read_remaining = OSD_PACKET_SIZE;
-        cl->read_state = CL_READ_HDR;
     }
     return true;
 }
 
-void osd_messenger_t::handle_reply_ready(osd_op_t *op)
+size_t osd_messenger_t::op_copy_from(osd_client_t *cl, uint8_t *src, size_t src_len, size_t & done)
 {
-    // Measure subop latency
+    osd_op_t *op = cl->read_op;
-    timespec tv_end;
+    size_t from = cl->read_op_pos-OSD_PACKET_SIZE;
-    clock_gettime(CLOCK_REALTIME, &tv_end);
+    auto op_read_buf = [&](uint8_t *dst, size_t dst_len)
-    stats.subop_stat_count[op->req.hdr.opcode]++;
-    if (!stats.subop_stat_count[op->req.hdr.opcode])
     {
-        stats.subop_stat_count[op->req.hdr.opcode]++;
+        if (from < dst_len)
-        stats.subop_stat_sum[op->req.hdr.opcode] = 0;
+        {
+            size_t n = dst_len-from;
+            if (n > src_len-done)
+                n = src_len-done;
+            memcpy(dst+from, src+done, n);
+            done += n;
+            cl->read_op_pos += n;
+            from += n;
+            if (from < dst_len)
+                return false;
+            from = 0;
+        }
+        else
+            from -= dst_len;
+        return true;
+    };
+    if (op->op_type == OSD_OP_IN)
+    {
+        if (op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
+            op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE)
+        {
+            if (!op_read_buf((uint8_t*)op->bitmap, op->req.sec_rw.attr_len))
+                return done;
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_rw.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
+            op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_stab.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_read_bmp.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_WRITE)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.rw.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SHOW_CONFIG)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.show_conf.json_len))
+                return done;
+        }
     }
-    stats.subop_stat_sum[op->req.hdr.opcode] += (
+    else
-        (tv_end.tv_sec - op->tv_begin.tv_sec)*1000000 +
+    {
-        (tv_end.tv_nsec - op->tv_begin.tv_nsec)/1000
+        if (op->reply.hdr.opcode == OSD_OP_SEC_READ)
-    );
+        {
-    set_immediate_ops.push_back(op);
+            if (op->reply.sec_rw.attr_len > 0)
+            {
+                if (!op_read_buf((uint8_t*)op->bitmap, op->reply.sec_rw.attr_len))
+                    return done;
+            }
+            if (op->reply.hdr.retval > 0)
+            {
+                for (int i = 0; i < op->iov.count; i++)
+                    if (!op_read_buf((uint8_t*)op->iov.buf[i].iov_base, op->iov.buf[i].iov_len))
+                        return done;
+            }
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_READ)
+        {
+            if (op->reply.rw.bitmap_len > 0)
+            {
+                if (!op_read_buf((uint8_t*)op->bitmap, op->reply.rw.bitmap_len))
+                    return done;
+            }
+            if (op->reply.hdr.retval > 0)
+            {
+                if (op->enc)
+                {
+                    if (!op_decrypted_copy_data_from(cl, src, src_len, from, done))
+                        return done;
+                }
+                else
+                {
+                    for (int i = 0; i < op->iov.count; i++)
+                        if (!op_read_buf((uint8_t*)op->iov.buf[i].iov_base, op->iov.buf[i].iov_len))
+                            return done;
+                }
+            }
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_SEC_LIST && op->reply.hdr.retval > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, sizeof(obj_ver_id) * op->reply.hdr.retval))
+                return done;
+        }
+        else if ((op->reply.hdr.opcode == OSD_OP_SEC_READ_BMP ||
+            op->reply.hdr.opcode == OSD_OP_SHOW_CONFIG) && op->reply.hdr.retval > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->reply.hdr.retval))
+                return done;
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_DESCRIBE && op->reply.describe.result_bytes > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->reply.describe.result_bytes))
+                return done;
+        }
+    }
+    handle_finished_op(cl);
+    return done;
+}
+
+size_t osd_messenger_t::op_get_read_buffers(osd_client_t *cl, std::vector<iovec> & lst)
+{
+    osd_op_t *op = cl->read_op;
+    size_t from = cl->read_op_pos-OSD_PACKET_SIZE;
+    size_t done = 0;
+    auto op_read_buf = [&](uint8_t *dst, size_t dst_len)
+    {
+        if (lst.size() >= IOV_MAX)
+            return false;
+        if (from < dst_len)
+        {
+            lst.push_back((iovec){ .iov_base = dst+from, .iov_len = dst_len-from });
+            cl->read_op_pos += dst_len-from;
+            done += dst_len-from;
+            from = 0;
+        }
+        else
+            from -= dst_len;
+        return true;
+    };
+    if (op->op_type == OSD_OP_IN)
+    {
+        if (op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
+            op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE)
+        {
+            if (!op_read_buf((uint8_t*)op->bitmap, op->req.sec_rw.attr_len))
+                return done;
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_rw.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
+            op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_stab.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.sec_read_bmp.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_WRITE)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.rw.len))
+                return done;
+        }
+        else if (op->req.hdr.opcode == OSD_OP_SHOW_CONFIG)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->req.show_conf.json_len))
+                return done;
+        }
+    }
+    else
+    {
+        if (op->reply.hdr.opcode == OSD_OP_SEC_READ)
+        {
+            if (op->reply.sec_rw.attr_len > 0)
+            {
+                if (!op_read_buf((uint8_t*)op->bitmap, op->reply.sec_rw.attr_len))
+                    return done;
+            }
+            if (op->reply.hdr.retval > 0)
+            {
+                for (int i = 0; i < op->iov.count; i++)
+                    if (!op_read_buf((uint8_t*)op->iov.buf[i].iov_base, op->iov.buf[i].iov_len))
+                        return done;
+            }
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_READ)
+        {
+            if (op->reply.rw.bitmap_len > 0)
+            {
+                if (!op_read_buf((uint8_t*)op->bitmap, op->reply.rw.bitmap_len))
+                    return done;
+            }
+            if (op->reply.hdr.retval > 0)
+            {
+                if (op->enc)
+                    cl->read_op_inline_decrypt_pos = cl->read_op_pos;
+                for (int i = 0; i < op->iov.count; i++)
+                    if (!op_read_buf((uint8_t*)op->iov.buf[i].iov_base, op->iov.buf[i].iov_len))
+                        return done;
+            }
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_SEC_LIST && op->reply.hdr.retval > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, sizeof(obj_ver_id) * op->reply.hdr.retval))
+                return done;
+        }
+        else if ((op->reply.hdr.opcode == OSD_OP_SEC_READ_BMP ||
+            op->reply.hdr.opcode == OSD_OP_SHOW_CONFIG) && op->reply.hdr.retval > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->reply.hdr.retval))
+                return done;
+        }
+        else if (op->reply.hdr.opcode == OSD_OP_DESCRIBE && op->reply.describe.result_bytes > 0)
+        {
+            if (!op_read_buf((uint8_t*)op->buf, op->reply.describe.result_bytes))
+                return done;
+        }
+    }
+    return done;
+}
+
+void osd_messenger_t::handle_finished_op(osd_client_t *cl)
+{
+    osd_op_t *op = cl->read_op;
+    if (op->op_type == OSD_OP_IN)
+    {
+        // Operation is ready
+        cl->received_ops.push_back(op);
+    }
+    else
+    {
+        // Inline decryption
+        if (cl->read_op_inline_decrypt_pos != (size_t)-1)
+        {
+            op_decrypt_inline(cl);
+            cl->read_op_inline_decrypt_pos = (size_t)-1;
+        }
+        // Measure subop (outbound op) latency
+        timespec tv_end;
+        clock_gettime(CLOCK_REALTIME, &tv_end);
+        stats.subop_stat_count[op->req.hdr.opcode]++;
+        if (!stats.subop_stat_count[op->req.hdr.opcode])
+        {
+            stats.subop_stat_count[op->req.hdr.opcode]++;
+            stats.subop_stat_sum[op->req.hdr.opcode] = 0;
+        }
+        stats.subop_stat_sum[op->req.hdr.opcode] += (
+            (tv_end.tv_sec - op->tv_begin.tv_sec)*1000000 +
+            (tv_end.tv_nsec - op->tv_begin.tv_nsec)/1000
+        );
+    }
+    set_immediate_ops.push_back(op);
+    cl->read_op = NULL;
 }

src/client/msgr_send.cpp

@@ -6,26 +6,21 @@
 #include <sys/epoll.h>
 
 #include "messenger.h"
-#include "msgr_iothread.h"
 
 void osd_messenger_t::outbox_push(osd_op_t *cur_op)
 {
-    assert(cur_op->client_id);
+    assert(cur_op->peer_fd);
-    auto cl_it = clients.find(cur_op->client_id);
+    osd_client_t *cl = clients.at(cur_op->peer_fd);
-    if (cl_it == clients.end() || cl_it->second->peer_state == PEER_STOPPED)
-    {
-        delete cur_op;
-        return;
-    }
-    osd_client_t *cl = cl_it->second;
     if (cur_op->op_type == OSD_OP_OUT)
     {
         clock_gettime(CLOCK_REALTIME, &cur_op->tv_begin);
         cur_op->req.hdr.id = ++cl->send_op_id;
+        cl->sent_ops[cur_op->req.hdr.id] = cur_op;
     }
     else
     {
-        // Remove the operation from received op list
+        // Check that operation actually belongs to this client
+        // FIXME: Review if this is still needed
         bool found = false;
         for (auto it = cl->received_ops.begin(); it != cl->received_ops.end(); it++)
         {
@@ -36,84 +31,14 @@ void osd_messenger_t::outbox_push(osd_op_t *cur_op)
                 break;
             }
         }
-        // Can't be not found because client IDs are unique
+        if (!found)
-        assert(found);
+        {
-    }
+            delete cur_op;
-    auto & to_send_list = cl->write_msg.msg_iovlen ? cl->next_send_list : cl->send_list;
+            return;
-    auto & to_outbox = cl->write_msg.msg_iovlen ? cl->next_outbox : cl->outbox;
+        }
-    if (cur_op->op_type == OSD_OP_IN)
-    {
         measure_exec(cur_op);
-        to_send_list.push_back((iovec){ .iov_base = cur_op->reply.buf, .iov_len = OSD_PACKET_SIZE });
-    }
-    else
-    {
-        to_send_list.push_back((iovec){ .iov_base = cur_op->req.buf, .iov_len = OSD_PACKET_SIZE });
-        cl->sent_ops[cur_op->req.hdr.id] = cur_op;
-    }
-    to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = MSGR_SENDP_HDR });
-    // Bitmap
-    if (cur_op->op_type == OSD_OP_IN &&
-        cur_op->req.hdr.opcode == OSD_OP_SEC_READ &&
-        cur_op->reply.sec_rw.attr_len > 0)
-    {
-        to_send_list.push_back((iovec){
-            .iov_base = cur_op->bitmap,
-            .iov_len = cur_op->reply.sec_rw.attr_len,
-        });
-        to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
-    }
-    else if (cur_op->op_type == OSD_OP_OUT &&
-        (cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE || cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE) &&
-        cur_op->req.sec_rw.attr_len > 0)
-    {
-        to_send_list.push_back((iovec){
-            .iov_base = cur_op->bitmap,
-            .iov_len = cur_op->req.sec_rw.attr_len,
-        });
-        to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
-    }
-    // Operation data
-    if ((cur_op->op_type == OSD_OP_IN
-        ? (cur_op->req.hdr.opcode == OSD_OP_READ ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_READ ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_LIST ||
-        cur_op->req.hdr.opcode == OSD_OP_SHOW_CONFIG ||
-        cur_op->req.hdr.opcode == OSD_OP_DESCRIBE)
-        : (cur_op->req.hdr.opcode == OSD_OP_WRITE ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
-        cur_op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK ||
-        cur_op->req.hdr.opcode == OSD_OP_SHOW_CONFIG)) && cur_op->iov.count > 0)
-    {
-        for (int i = 0; i < cur_op->iov.count; i++)
-        {
-            if (cur_op->iov.buf[i].iov_len > 0)
-            {
-                assert(cur_op->iov.buf[i].iov_base);
-                to_send_list.push_back(cur_op->iov.buf[i]);
-                to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
-            }
-        }
-    }
-    if (cur_op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
-    {
-        if (cur_op->op_type == OSD_OP_IN && cur_op->reply.hdr.retval > 0)
-        {
-            to_send_list.push_back((iovec){ .iov_base = cur_op->buf, .iov_len = (size_t)cur_op->reply.hdr.retval });
-            to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
-        }
-        else if (cur_op->op_type == OSD_OP_OUT && cur_op->req.sec_read_bmp.len > 0)
-        {
-            to_send_list.push_back((iovec){ .iov_base = cur_op->buf, .iov_len = (size_t)cur_op->req.sec_read_bmp.len });
-            to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
-        }
-    }
-    if (cur_op->op_type == OSD_OP_IN)
-    {
-        to_outbox[to_outbox.size()-1].flags |= MSGR_SENDP_FREE;
     }
+    cl->write_ops.push_back(cur_op);
 #ifdef WITH_RDMA
     if (cl->peer_state == PEER_RDMA)
     {
@@ -124,7 +49,7 @@ void osd_messenger_t::outbox_push(osd_op_t *cur_op)
     if (!ringloop)
     {
         // FIXME: It's worse because it doesn't allow batching
-        while (cl->outbox.size())
+        while (cl->write_ops.size())
         {
             try_send(cl);
         }
@@ -134,7 +59,7 @@ void osd_messenger_t::outbox_push(osd_op_t *cur_op)
         if ((cl->write_msg.msg_iovlen > 0 || !try_send(cl)) && (cl->write_state == 0))
         {
             cl->write_state = CL_WRITE_READY;
-            write_ready_clients.push_back(cur_op->client_id);
+            write_ready_clients.push_back(cur_op->peer_fd);
         }
         ringloop->wakeup();
     }
@@ -191,14 +116,29 @@ void osd_messenger_t::measure_exec(osd_op_t *cur_op)
 
 bool osd_messenger_t::try_send(osd_client_t *cl)
 {
-    if (!cl->send_list.size() || cl->write_msg.msg_iovlen > 0 || cl->peer_state == PEER_STOPPED || cl->peer_fd < 0)
+    int peer_fd = cl->peer_fd;
+    if (!cl->write_op && !cl->write_ops.size() || cl->write_msg.msg_iovlen > 0)
     {
         return true;
     }
     assert(cl->peer_state != PEER_RDMA);
+    while ((cl->write_op || cl->write_ops.size()) && cl->send_list.size() < IOV_MAX)
+    {
+        if (!cl->write_op)
+        {
+            cl->write_op = cl->write_ops.front();
+            cl->write_ops.pop_front();
+        }
+        osd_op_t *op = cl->write_op;
+        op_get_write_buffers(cl, cl->send_list);
+        if (!cl->write_op && op->op_type == OSD_OP_IN)
+        {
+            cl->send_free_ops.push_back(op);
+        }
+    }
     if (ringloop && !use_sync_send_recv)
     {
-        auto iothread = iothreads.size() ? iothreads[cl->peer_fd % iothreads.size()] : NULL;
+        auto iothread = iothreads.size() ? iothreads[peer_fd % iothreads.size()] : NULL;
         io_uring_sqe sqe_local;
         ring_data_t data_local;
         io_uring_sqe* sqe = (iothread ? &sqe_local : ringloop->get_sqe());
@@ -208,28 +148,32 @@ bool osd_messenger_t::try_send(osd_client_t *cl)
             data_local = {};
         }
         if (!sqe)
+        {
             return false;
+        }
+        cl->send_list_size = 0;
+        for (auto & iov: cl->send_list)
+        {
+            cl->send_list_size += iov.iov_len;
+        }
         cl->write_msg.msg_iov = cl->send_list.data();
         cl->write_msg.msg_iovlen = cl->send_list.size() < IOV_MAX ? cl->send_list.size() : IOV_MAX;
         cl->refs++;
         ring_data_t* data = ((ring_data_t*)sqe->user_data);
         data->callback = [this, cl](ring_data_t *data) { handle_send(data->res, data->prev, data->more, cl); };
         bool use_zc = has_sendmsg_zc && min_zerocopy_send_size >= 0;
-        if (use_zc && min_zerocopy_send_size > 0)
+        if (use_zc && min_zerocopy_send_size > 0 &&
+            cl->send_list_size/cl->write_msg.msg_iovlen < min_zerocopy_send_size)
         {
-            size_t avg_size = 0;
+            use_zc = false;
-            for (size_t i = 0; i < cl->write_msg.msg_iovlen; i++)
-                avg_size += cl->write_msg.msg_iov[i].iov_len;
-            if (avg_size/cl->write_msg.msg_iovlen < min_zerocopy_send_size)
-                use_zc = false;
         }
         if (use_zc)
         {
-            io_uring_prep_sendmsg_zc(sqe, cl->peer_fd, &cl->write_msg, MSG_WAITALL);
+            io_uring_prep_sendmsg_zc(sqe, peer_fd, &cl->write_msg, MSG_WAITALL);
         }
         else
         {
-            io_uring_prep_sendmsg(sqe, cl->peer_fd, &cl->write_msg, MSG_WAITALL);
+            io_uring_prep_sendmsg(sqe, peer_fd, &cl->write_msg, MSG_WAITALL);
         }
         if (iothread)
         {
@@ -241,7 +185,7 @@ bool osd_messenger_t::try_send(osd_client_t *cl)
         cl->write_msg.msg_iov = cl->send_list.data();
         cl->write_msg.msg_iovlen = cl->send_list.size() < IOV_MAX ? cl->send_list.size() : IOV_MAX;
         cl->refs++;
-        int result = sendmsg(cl->peer_fd, &cl->write_msg, MSG_NOSIGNAL);
+        int result = sendmsg(peer_fd, &cl->write_msg, MSG_NOSIGNAL);
         if (result < 0)
         {
             result = -errno;
@@ -256,9 +200,9 @@ void osd_messenger_t::send_replies()
 {
     for (int i = 0; i < write_ready_clients.size(); i++)
     {
-        uint64_t client_id = write_ready_clients[i];
+        int peer_fd = write_ready_clients[i];
-        auto cl_it = clients.find(client_id);
+        auto cl_it = clients.find(peer_fd);
-        if (cl_it != clients.end() && cl_it->second->peer_state != PEER_RDMA && !try_send(cl_it->second))
+        if (cl_it != clients.end() && !try_send(cl_it->second))
         {
             write_ready_clients.erase(write_ready_clients.begin(), write_ready_clients.begin() + i);
             return;
@@ -272,6 +216,7 @@ void osd_messenger_t::handle_send(int result, bool prev, bool more, osd_client_t
     if (!prev)
     {
         cl->write_msg.msg_iovlen = 0;
+        cl->send_list.clear();
     }
     if (!more)
     {
@@ -281,15 +226,15 @@ void osd_messenger_t::handle_send(int result, bool prev, bool more, osd_client_t
     {
         if (cl->refs <= 0)
         {
-            destroy_client(cl);
+            delete cl;
         }
         return;
     }
     if (result < 0 && result != -EAGAIN && result != -EINTR)
     {
         // this is a client socket, so don't panic. just disconnect it
-        fprintf(stderr, "Client %ju socket write error: %d (%s). Disconnecting client\n", cl->client_id, -result, strerror(-result));
+        fprintf(stderr, "Client %d socket write error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
-        stop_client(cl->client_id);
+        stop_client(cl->peer_fd);
         return;
     }
     if (result >= 0)
@@ -304,64 +249,42 @@ void osd_messenger_t::handle_send(int result, bool prev, bool more, osd_client_t
                 cl->zc_free_list.erase(cl->zc_free_list.begin(), cl->zc_free_list.begin()+i+1);
             return;
         }
-        int done = 0;
+        if (cl->send_list_size > result)
-        while (result > 0 && done < cl->send_list.size())
         {
-            iovec & iov = cl->send_list[done];
+            fprintf(stderr, "Client %d socket write error: expected to send "
-            if (iov.iov_len <= result)
+                "%zu bytes with MSG_WAITALL but sent %u. Disconnecting client\n", cl->peer_fd, cl->send_list_size, result);
-            {
+            stop_client(cl->peer_fd);
-                if (cl->outbox[done].flags & MSGR_SENDP_FREE)
+            return;
-                {
+        }
-                    // Reply fully sent
+        for (auto op: cl->send_free_ops)
-                    if (more)
+        {
-                        cl->zc_free_list.push_back(cl->outbox[done].op);
+            if (more)
-                    else
+                cl->zc_free_list.push_back(op);
-                        delete cl->outbox[done].op;
-                }
-                result -= iov.iov_len;
-                done++;
-            }
             else
-            {
+                delete op;
-                iov.iov_len -= result;
-                iov.iov_base = (uint8_t*)iov.iov_base + result;
-                break;
-            }
         }
         if (more)
-        {
-            int expected = cl->send_list.size() < IOV_MAX ? cl->send_list.size() : IOV_MAX;
-            if (done != expected)
-            {
-                fprintf(stderr, "Client %ju socket write error: expected to send "
-                    "%d iovecs with MSG_WAITALL but sent %d. Disconnecting client\n", cl->client_id, expected, done);
-                stop_client(cl->client_id);
-                return;
-            }
             cl->zc_free_list.push_back(NULL); // end marker
-        }
+        cl->send_free_ops.clear();
-        if (done > 0)
+        cl->write_state = cl->write_op || cl->write_ops.size() ? CL_WRITE_READY : 0;
-        {
-            cl->send_list.erase(cl->send_list.begin(), cl->send_list.begin()+done);
-            cl->outbox.erase(cl->outbox.begin(), cl->outbox.begin()+done);
-        }
-        if (cl->next_send_list.size())
-        {
-            cl->send_list.insert(cl->send_list.end(), cl->next_send_list.begin(), cl->next_send_list.end());
-            cl->outbox.insert(cl->outbox.end(), cl->next_outbox.begin(), cl->next_outbox.end());
-            cl->next_send_list.clear();
-            cl->next_outbox.clear();
-        }
-        cl->write_state = cl->outbox.size() > 0 ? CL_WRITE_READY : 0;
 #ifdef WITH_RDMA
-        if (cl->rdma_conn && !cl->outbox.size() && cl->peer_state == PEER_RDMA_CONNECTING)
+        if (cl->rdma_conn && !cl->write_op && !cl->write_ops.size() && cl->peer_state == PEER_RDMA_CONNECTING)
         {
+            // FIXME: Do something better than just forgetting the FD
             // FIXME: Ignore pings during RDMA state transition
             if (log_level > 0)
             {
-                fprintf(stderr, "Successfully connected with client %ju using RDMA\n", cl->client_id);
+                fprintf(stderr, "Successfully connected with client %d using RDMA\n", cl->peer_fd);
             }
             cl->peer_state = PEER_RDMA;
+            tfd->set_fd_handler(cl->peer_fd, false, [this](int peer_fd, int epoll_events)
+            {
+                // Do not miss the disconnection!
+                if (epoll_events & EPOLLRDHUP)
+                {
+                    handle_peer_epoll(peer_fd, epoll_events);
+                }
+            });
             // Add the initial receive request
             init_recv_rdma(cl);
         }
@@ -369,6 +292,161 @@ void osd_messenger_t::handle_send(int result, bool prev, bool more, osd_client_t
     }
     if (cl->write_state != 0)
     {
-        write_ready_clients.push_back(cl->client_id);
+        write_ready_clients.push_back(cl->peer_fd);
     }
 }
+
+static inline bool op_write_headers(osd_op_t *op, std::function<bool(uint8_t*, size_t)> op_write_buf)
+{
+    // Header
+    if (!op_write_buf((op->op_type == OSD_OP_IN ? op->reply.buf : op->req.buf), OSD_PACKET_SIZE))
+        return false;
+    // Bitmap
+    if (op->op_type == OSD_OP_IN &&
+        op->req.hdr.opcode == OSD_OP_SEC_READ &&
+        op->reply.sec_rw.attr_len > 0)
+    {
+        if (!op_write_buf((uint8_t*)op->bitmap, op->reply.sec_rw.attr_len))
+            return false;
+    }
+    else if (op->op_type == OSD_OP_OUT &&
+        (op->req.hdr.opcode == OSD_OP_SEC_WRITE || op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE) &&
+        op->req.sec_rw.attr_len > 0)
+    {
+        if (!op_write_buf((uint8_t*)op->bitmap, op->req.sec_rw.attr_len))
+            return false;
+    }
+    if (op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
+    {
+        if (op->op_type == OSD_OP_IN && op->reply.hdr.retval > 0)
+        {
+            if (!op_write_buf((uint8_t*)op->buf, (size_t)op->reply.hdr.retval))
+                return false;
+        }
+        else if (op->op_type == OSD_OP_OUT && op->req.sec_read_bmp.len > 0)
+        {
+            if (!op_write_buf((uint8_t*)op->buf, (size_t)op->req.sec_read_bmp.len))
+                return false;
+        }
+    }
+    return true;
+}
+
+static inline bool op_has_data(osd_op_t *op)
+{
+    return (op->op_type == OSD_OP_IN
+        ? (op->req.hdr.opcode == OSD_OP_READ ||
+        op->req.hdr.opcode == OSD_OP_SEC_READ ||
+        op->req.hdr.opcode == OSD_OP_SEC_LIST ||
+        op->req.hdr.opcode == OSD_OP_SHOW_CONFIG ||
+        op->req.hdr.opcode == OSD_OP_DESCRIBE)
+        : (op->req.hdr.opcode == OSD_OP_WRITE ||
+        op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
+        op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE ||
+        op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
+        op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK ||
+        op->req.hdr.opcode == OSD_OP_SHOW_CONFIG)) && op->iov.count > 0;
+}
+
+size_t osd_messenger_t::op_copy_to(osd_client_t *cl, uint8_t *dst, size_t dst_len)
+{
+    size_t done = 0;
+    size_t from = cl->write_op_pos;
+    auto op_write_buf = [&](uint8_t *src, size_t src_len)
+    {
+        if (from < src_len)
+        {
+            size_t n = src_len-from;
+            if (n > dst_len-done)
+                n = dst_len-done;
+            memcpy(dst+done, src+from, n);
+            done += n;
+            cl->write_op_pos += n;
+            from += n;
+            if (from < src_len)
+                return false;
+            from = 0;
+        }
+        else
+            from -= src_len;
+        return true;
+    };
+    if (!op_write_headers(cl->write_op, op_write_buf))
+    {
+        return done;
+    }
+    // Operation data
+    if (op_has_data(cl->write_op))
+    {
+        if (cl->write_op->enc)
+        {
+            if (!op_encrypted_copy_data_to(cl, dst, dst_len, from, done))
+            {
+                return done;
+            }
+        }
+        else
+        {
+            for (int i = 0; i < cl->write_op->iov.count; i++)
+            {
+                if (!op_write_buf((uint8_t*)cl->write_op->iov.buf[i].iov_base, cl->write_op->iov.buf[i].iov_len))
+                    return done;
+            }
+        }
+    }
+    cl->write_op = NULL;
+    cl->write_op_pos = 0;
+    return done;
+}
+
+void osd_messenger_t::op_get_write_buffers(osd_client_t *cl, std::vector<iovec> & lst)
+{
+    size_t from = cl->write_op_pos;
+    auto op_write_buf = [&](uint8_t *src, size_t src_len)
+    {
+        if (lst.size() >= IOV_MAX)
+            return false;
+        if (from < src_len)
+        {
+            lst.push_back((iovec){ .iov_base = src+from, .iov_len = src_len-from });
+            cl->write_op_pos += src_len-from;
+            from = 0;
+        }
+        else
+            from -= src_len;
+        return true;
+    };
+    if (!op_write_headers(cl->write_op, op_write_buf))
+    {
+        return;
+    }
+    // Operation data
+    if (op_has_data(cl->write_op))
+    {
+        if (cl->write_op->enc)
+        {
+            if (lst.size() >= IOV_MAX)
+                return;
+            // No way except to allocate a temporary buffer and encrypt data to it
+            assert(cl->write_op->req.hdr.opcode == OSD_OP_WRITE);
+            size_t remsize = cl->write_op->req.rw.len - from + (from % 16);
+            assert(remsize > 0);
+            assert(!cl->write_op->enc_buf);
+            cl->write_op->enc_buf = (uint8_t*)malloc_or_die(remsize);
+            size_t done = 0;
+            bool end = op_encrypted_copy_data_to(cl, cl->write_op->enc_buf, remsize, from, done);
+            assert(end);
+            lst.push_back((iovec){ .iov_base = cl->write_op->enc_buf, .iov_len = remsize });
+        }
+        else
+        {
+            for (int i = 0; i < cl->write_op->iov.count; i++)
+            {
+                if (!op_write_buf((uint8_t*)cl->write_op->iov.buf[i].iov_base, cl->write_op->iov.buf[i].iov_len))
+                    return;
+            }
+        }
+    }
+    cl->write_op = NULL;
+    cl->write_op_pos = 0;
+}

src/client/msgr_stop.cpp

@@ -5,6 +5,9 @@
 #include <assert.h>
 
 #include "messenger.h"
+#ifdef WITH_RDMA
+#include "msgr_rdma.h"
+#endif
 
 void osd_client_t::cancel_ops()
 {
@@ -40,60 +43,99 @@ void osd_op_t::cancel()
     }
 }
 
-// force_delete means stop the client anyway, even if there are refs to it in the event loop.
+void osd_messenger_t::stop_client(int peer_fd, bool force, bool force_delete)
-// the flag should be used in the destructor.
-// why? - because yes, we could close the FD first and let it fail all requests in the event loop,
-// but in that case it can be quickly reopened and we can get old failed responses for the new FD.
-void osd_messenger_t::stop_client(uint64_t client_id, bool force_delete)
 {
-    auto it = clients.find(client_id);
+    assert(peer_fd != 0);
-    if (!client_id || it == clients.end())
+    auto it = clients.find(peer_fd);
+    if (it == clients.end())
     {
         return;
     }
     osd_client_t *cl = it->second;
-    if (cl->peer_state == PEER_STOPPED)
+    // FIXME: This 'force' flag is probably an ugly reenterability hack - check its logic and maybe remove it
+    if (cl->peer_state == PEER_CONNECTING && !force || cl->peer_state == PEER_STOPPED)
     {
-        if (force_delete)
-        {
-            destroy_client(cl);
-        }
         return;
     }
-    cl->received_ops.clear();
     if (log_level > 0)
     {
         if (cl->osd_num)
         {
-            fprintf(stderr, "[OSD %ju] Stopping client %ju (OSD peer %ju)\n", osd_num, client_id, cl->osd_num);
+            fprintf(stderr, "[OSD %ju] Stopping client %d (OSD peer %ju)\n", osd_num, peer_fd, cl->osd_num);
         }
         else if (cl->in_osd_num)
         {
-            fprintf(stderr, "[OSD %ju] Stopping client %ju (incoming OSD peer %ju)\n", osd_num, client_id, cl->in_osd_num);
+            fprintf(stderr, "[OSD %ju] Stopping client %d (incoming OSD peer %ju)\n", osd_num, peer_fd, cl->in_osd_num);
         }
         else
         {
-            fprintf(stderr, "[OSD %ju] Stopping client %ju (regular client)\n", osd_num, client_id);
+            fprintf(stderr, "[OSD %ju] Stopping client %d (regular client)\n", osd_num, peer_fd);
         }
     }
+    if (cl->encrypt_ctx)
+    {
+        if (encrypt_ctx_pool.size() > max_aes_xts_pool_size)
+            destroy_aes_xts_encrypt(cl->encrypt_ctx);
+        else
+            encrypt_ctx_pool.push_back(cl->encrypt_ctx);
+        cl->encrypt_ctx = NULL;
+    }
+    if (cl->decrypt_ctx)
+    {
+        if (decrypt_ctx_pool.size() > max_aes_xts_pool_size)
+            destroy_aes_xts_decrypt(cl->decrypt_ctx);
+        else
+            decrypt_ctx_pool.push_back(cl->decrypt_ctx);
+        cl->decrypt_ctx = NULL;
+    }
     // First set state to STOPPED so another stop_client() call doesn't try to free it again
     cl->refs++;
     int prev_state = cl->peer_state;
     cl->peer_state = PEER_STOPPED;
     if (cl->osd_num)
     {
-        auto osd_it = osd_peers.find(cl->osd_num);
+        auto osd_it = osd_peer_fds.find(cl->osd_num);
-        if (osd_it != osd_peers.end() && osd_it->second == cl)
+        if (osd_it != osd_peer_fds.end() && osd_it->second == cl->peer_fd)
         {
             // ...and forget OSD peer
-            osd_peers.erase(osd_it);
+            osd_peer_fds.erase(osd_it);
         }
     }
+#ifdef WITH_RDMA
+    if (cl->rdma_conn && cl->rdma_conn->cmid)
+    {
+        auto rdma_it = rdmacm_connections.find(cl->rdma_conn->cmid);
+        if (rdma_it != rdmacm_connections.end() && rdma_it->second == cl)
+        {
+            rdmacm_connections.erase(rdma_it);
+        }
+    }
+#endif
+#ifndef __MOCK__
+    // Then remove FD from the eventloop so we don't accidentally read something
+    tfd->set_fd_handler(peer_fd, false, NULL);
     if (cl->connect_timeout_id >= 0)
     {
         tfd->clear_timer(cl->connect_timeout_id);
         cl->connect_timeout_id = -1;
     }
+    for (auto rit = read_ready_clients.begin(); rit != read_ready_clients.end(); rit++)
+    {
+        if (*rit == peer_fd)
+        {
+            read_ready_clients.erase(rit);
+            break;
+        }
+    }
+    for (auto wit = write_ready_clients.begin(); wit != write_ready_clients.end(); wit++)
+    {
+        if (*wit == peer_fd)
+        {
+            write_ready_clients.erase(wit);
+            break;
+        }
+    }
+#endif
     if (cl->in_osd_num && break_pg_locks)
     {
         // Break PG locks
@@ -107,56 +149,19 @@ void osd_messenger_t::stop_client(uint64_t client_id, bool force_delete)
         // so do not repeer on it.
         repeer_pgs(cl->osd_num);
     }
-    if (cl->peer_fd >= 0)
+    // Find the item again because it can be invalidated at this point
+    it = clients.find(peer_fd);
+    if (it != clients.end())
     {
-        int r = shutdown(cl->peer_fd, SHUT_RDWR);
+        clients.erase(it);
-        if (r != 0 && errno != ENOTCONN)
-        {
-            fprintf(stderr, "[OSD %ju] failed to shutdown a socket: %s (code %d)\n", osd_num, strerror(errno), errno);
-        }
     }
     cl->refs--;
     if (cl->refs <= 0 || force_delete)
     {
-        destroy_client(cl);
+        delete cl;
     }
 }
 
-void osd_messenger_t::destroy_client(osd_client_t *cl)
-{
-    // Find the item again because it can be invalidated at this point
-    clients.erase(cl->client_id);
-    if (cl->peer_fd >= 0)
-    {
-        tfd->set_fd_handler(cl->peer_fd, false, NULL);
-        for (auto rit = read_ready_clients.begin(); rit != read_ready_clients.end(); rit++)
-        {
-            if (*rit == cl->client_id)
-            {
-                read_ready_clients.erase(rit);
-                break;
-            }
-        }
-        for (auto wit = write_ready_clients.begin(); wit != write_ready_clients.end(); wit++)
-        {
-            if (*wit == cl->client_id)
-            {
-                write_ready_clients.erase(wit);
-                break;
-            }
-        }
-        clients_by_fd.erase(cl->peer_fd);
-    }
-#ifdef WITH_RDMA
-    if (cl->rdma_conn)
-    {
-        destroy_rdma_conn(cl->rdma_conn);
-        cl->rdma_conn = NULL;
-    }
-#endif
-    delete cl;
-}
-
 osd_client_t::~osd_client_t()
 {
     free(in_buf);
@@ -179,6 +184,13 @@ osd_client_t::~osd_client_t()
     }
     // Cancel outbound ops
     cancel_ops();
+    for (osd_op_t *op: send_free_ops)
+    {
+        if (op)
+        {
+            delete op;
+        }
+    }
     for (osd_op_t *op: zc_free_list)
     {
         if (op)
@@ -186,4 +198,13 @@ osd_client_t::~osd_client_t()
             delete op;
         }
     }
+#ifndef __MOCK__
+#ifdef WITH_RDMA
+    if (rdma_conn)
+    {
+        delete rdma_conn;
+        rdma_conn = NULL;
+    }
+#endif
+#endif
 }

src/client/object_id.h

@@ -20,15 +20,6 @@ typedef uint64_t inode_t;
 // Pool ID is 16 bits long
 typedef uint32_t pool_id_t;
 
-typedef uint64_t osd_num_t;
-typedef uint32_t pg_num_t;
-
-struct pool_pg_num_t
-{
-    pool_id_t pool_id;
-    pg_num_t pg_num;
-};
-
 // 16 bytes per object/stripe id
 // stripe = (start of the parity stripe + peer role)
 // i.e. for example (256KB + one of 0,1,2)
@@ -70,21 +61,6 @@ inline bool operator < (const obj_ver_id & a, const obj_ver_id & b)
     return a.oid < b.oid || a.oid == b.oid && a.version < b.version;
 }
 
-inline bool operator < (const pool_pg_num_t & a, const pool_pg_num_t & b)
-{
-    return a.pool_id < b.pool_id || a.pool_id == b.pool_id && a.pg_num < b.pg_num;
-}
-
-inline bool operator == (const pool_pg_num_t & a, const pool_pg_num_t & b)
-{
-    return a.pool_id == b.pool_id && a.pg_num == b.pg_num;
-}
-
-inline bool operator != (const pool_pg_num_t & a, const pool_pg_num_t & b)
-{
-    return a.pool_id != b.pool_id || a.pg_num != b.pg_num;
-}
-
 namespace std
 {
     template<> struct hash<object_id>

src/client/osd_ops.h

@@ -4,6 +4,7 @@
 #pragma once
 
 #include "object_id.h"
+#include "osd_id.h"
 
 // Magic numbers
 #define SECONDARY_OSD_OP_MAGIC      0x2bd7b10325434553l

src/client/qemu_driver.c

@@ -1049,7 +1049,7 @@ static int coroutine_fn vitastor_co_block_status(BlockDriverState *bs,
     {
         // Get larger allocated extents, possibly with false positives
         uint64_t bmp_pos = (offset-task.offset) / task.bitmap_granularity;
-        uint64_t bmp_end = (offset+bytes-task.offset) / task.bitmap_granularity;
+        uint64_t bmp_end = (offset+bytes-task.offset) / task.bitmap_granularity - bmp_pos;
         while (bmp_pos < bmp_end)
         {
             if (!(bmp_pos & 7) && bmp_end >= bmp_pos+8)