Release 0.9.6

- Fix vitastor-disk partition zeroing (sometimes it was writing garbage instead of zeroes)
- Fix incorrect EC space statistics in `vitastor-cli status`
- Several bug fixes for NFS:
  - Add . and .. in NFS directory listings
  - Return FILE_SYNC from NFS writes if immediate_commit is enabled
  - Return the same "verifier" in NFS COMMIT as in NFS WRITE
  - Make parallel NFS extending writes work correctly, without conflicts
  - Handle parallel NFS extending writes without imposing extra load on etcd
- Support UTF-8 in vitastor-cli table output
- Also allow "0" and "no" as false for inmemory_metadata and inmemory_journal
- Use HDD defaults for HDD-only in automatic `vitastor-disk prepare` mode

CMakeLists.txt

@@ -2,6 +2,6 @@ cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
-set(VERSION "0.9.2")
+set(VERSION "0.9.6")
 
 add_subdirectory(src)

README-ru.md

@@ -15,7 +15,7 @@ Vitastor архитектурно похож на Ceph, что означает
 и автоматическое распределение данных по любому числу дисков любого размера с настраиваемыми схемами
 избыточности - репликацией или с произвольными кодами коррекции ошибок.
 
-Vitastor нацелен на SSD и SSD+HDD кластеры с как минимум 10 Гбит/с сетью, поддерживает
+Vitastor нацелен в первую очередь на SSD и SSD+HDD кластеры с как минимум 10 Гбит/с сетью, поддерживает
 TCP и RDMA и на хорошем железе может достигать задержки 4 КБ чтения и записи на уровне ~0.1 мс,
 что примерно в 10 раз быстрее, чем Ceph и другие популярные программные СХД.
 

README.md

@@ -14,8 +14,8 @@ Vitastor is architecturally similar to Ceph which means strong consistency,
 primary-replication, symmetric clustering and automatic data distribution over any
 number of drives of any size with configurable redundancy (replication or erasure codes/XOR).
 
-Vitastor targets SSD and SSD+HDD clusters with at least 10 Gbit/s network, supports
+Vitastor targets primarily SSD and SSD+HDD clusters with at least 10 Gbit/s network,
-TCP and RDMA and may achieve 4 KB read and write latency as low as ~0.1 ms
+supports TCP and RDMA and may achieve 4 KB read and write latency as low as ~0.1 ms
 with proper hardware which is ~10 times faster than other popular SDS's like Ceph
 or internal systems of public clouds.
 

csi/Makefile

@@ -1,4 +1,4 @@
-VERSION ?= v0.9.2
+VERSION ?= v0.9.6
 
 all: build push
 

csi/deploy/004-csi-nodeplugin.yaml

@@ -49,7 +49,7 @@ spec:
             capabilities:
               add: ["SYS_ADMIN"]
             allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v0.9.2
+          image: vitalif/vitastor-csi:v0.9.6
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/deploy/007-csi-provisioner.yaml

@@ -116,7 +116,7 @@ spec:
             privileged: true
             capabilities:
               add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v0.9.2
+          image: vitalif/vitastor-csi:v0.9.6
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/src/config.go

@@ -5,7 +5,7 @@ package vitastor
 
 const (
     vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "0.9.2"
+    vitastorCSIDriverVersion = "0.9.6"
 )
 
 // Config struct fills the parameters of request or user input

debian/changelog

@@ -1,10 +1,10 @@
-vitastor (0.9.2-1) unstable; urgency=medium
+vitastor (0.9.6-1) unstable; urgency=medium
 
   * Bugfixes
 
  -- Vitaliy Filippov <vitalif@yourcmc.ru>  Fri, 03 Jun 2022 02:09:44 +0300
 
-vitastor (0.9.2-1) unstable; urgency=medium
+vitastor (0.9.6-1) unstable; urgency=medium
 
   * Implement NFS proxy
   * Add documentation

debian/patched-qemu.Dockerfile

@@ -28,13 +28,19 @@ RUN apt-get --download-only source qemu
 
 ADD patches /root/vitastor/patches
 ADD src/qemu_driver.c /root/vitastor/src/qemu_driver.c
+
+#RUN set -e; \
+#    apt-get install -y wget; \
+#    wget -q -O /etc/apt/trusted.gpg.d/vitastor.gpg https://vitastor.io/debian/pubkey.gpg; \
+#    (echo deb http://vitastor.io/debian $REL main > /etc/apt/sources.list.d/vitastor.list); \
+#    (echo "APT::Install-Recommends false;" > /etc/apt/apt.conf) && \
+#    apt-get update; \
+#    apt-get install -y vitastor-client vitastor-client-dev quilt
+
 RUN set -e; \
-    apt-get install -y wget; \
+    dpkg -i /root/packages/vitastor-$REL/vitastor-client_*.deb /root/packages/vitastor-$REL/vitastor-client-dev_*.deb; \
-    wget -q -O /etc/apt/trusted.gpg.d/vitastor.gpg https://vitastor.io/debian/pubkey.gpg; \
-    (echo deb http://vitastor.io/debian $REL main > /etc/apt/sources.list.d/vitastor.list); \
-    (echo "APT::Install-Recommends false;" > /etc/apt/apt.conf) && \
     apt-get update; \
-    apt-get install -y vitastor-client vitastor-client-dev quilt; \
+    apt-get install -y quilt; \
     mkdir -p /root/packages/qemu-$REL; \
     rm -rf /root/packages/qemu-$REL/*; \
     cd /root/packages/qemu-$REL; \
@@ -48,7 +54,7 @@ RUN set -e; \
     quilt add block/vitastor.c; \
     cp /root/vitastor/src/qemu_driver.c block/vitastor.c; \
     quilt refresh; \
-    V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)(~bpo[\d\+]*)?\).*$/$1/')+vitastor1; \
+    V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)(~bpo[\d\+]*)?\).*$/$1/')+vitastor3; \
     DEBEMAIL="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v $V 'Plug Vitastor block driver'; \
     DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \
     rm -rf /root/packages/qemu-$REL/qemu-*/

debian/vitastor.Dockerfile

@@ -35,8 +35,8 @@ RUN set -e -x; \
     mkdir -p /root/packages/vitastor-$REL; \
     rm -rf /root/packages/vitastor-$REL/*; \
     cd /root/packages/vitastor-$REL; \
-    cp -r /root/vitastor vitastor-0.9.2; \
+    cp -r /root/vitastor vitastor-0.9.6; \
-    cd vitastor-0.9.2; \
+    cd vitastor-0.9.6; \
     ln -s /root/fio-build/fio-*/ ./fio; \
     FIO=$(head -n1 fio/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     ls /usr/include/linux/raw.h || cp ./debian/raw.h /usr/include/linux/raw.h; \
@@ -49,8 +49,8 @@ RUN set -e -x; \
     rm -rf a b; \
     echo "dep:fio=$FIO" > debian/fio_version; \
     cd /root/packages/vitastor-$REL; \
-    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.9.2.orig.tar.xz vitastor-0.9.2; \
+    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.9.6.orig.tar.xz vitastor-0.9.6; \
-    cd vitastor-0.9.2; \
+    cd vitastor-0.9.6; \
     V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v "$V""$REL" "Rebuild for $REL"; \
     DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \

docs/config/layout-cluster.en.md

@@ -33,12 +33,13 @@ Size of objects (data blocks) into which all physical and virtual drives
 in Vitastor, affects memory usage, write amplification and I/O load
 distribution effectiveness.
 
-Recommended default block size is 128 KB for SSD and 4 MB for HDD. In fact,
+Recommended default block size is 128 KB for SSD and 1 MB for HDD. In fact,
-it's possible to use 4 MB for SSD too - it will lower memory usage, but
+it's possible to use 1 MB for SSD too - it will lower memory usage, but
 may increase average WA and reduce linear performance.
 
 OSD memory usage is roughly (SIZE / BLOCK * 68 bytes) which is roughly
 544 MB per 1 TB of used disk space with the default 128 KB block size.
+With 1 MB it's 8 times lower.
 
 ## bitmap_granularity
 

docs/config/layout-cluster.ru.md

@@ -33,14 +33,14 @@ OSD) могут сосуществовать в одном кластере Vita
 настроек, влияет на потребление памяти, объём избыточной записи (write
 amplification) и эффективность распределения нагрузки по OSD.
 
-Рекомендуемые по умолчанию размеры блока - 128 килобайт для SSD и 4
+Рекомендуемые по умолчанию размеры блока - 128 килобайт для SSD и 1 мегабайт
-мегабайта для HDD. В принципе, для SSD можно тоже использовать 4 мегабайта,
+для HDD. В принципе, для SSD можно тоже использовать блок размером 1 мегабайт,
 это понизит использование памяти, но ухудшит распределение нагрузки и в
 среднем увеличит WA.
 
 Потребление памяти OSD составляет примерно (РАЗМЕР / БЛОК * 68 байт),
 т.е. примерно 544 МБ памяти на 1 ТБ занятого места на диске при
-стандартном 128 КБ блоке.
+стандартном 128 КБ блоке. При 1 МБ блоке памяти нужно в 8 раз меньше.
 
 ## bitmap_granularity
 

docs/config/src/layout-cluster.yml

@@ -7,26 +7,27 @@
     in Vitastor, affects memory usage, write amplification and I/O load
     distribution effectiveness.
 
-    Recommended default block size is 128 KB for SSD and 4 MB for HDD. In fact,
+    Recommended default block size is 128 KB for SSD and 1 MB for HDD. In fact,
-    it's possible to use 4 MB for SSD too - it will lower memory usage, but
+    it's possible to use 1 MB for SSD too - it will lower memory usage, but
     may increase average WA and reduce linear performance.
 
     OSD memory usage is roughly (SIZE / BLOCK * 68 bytes) which is roughly
     544 MB per 1 TB of used disk space with the default 128 KB block size.
+    With 1 MB it's 8 times lower.
   info_ru: |
     Размер объектов (блоков данных), на которые делятся физические и виртуальные
     диски в Vitastor (в рамках каждого пула). Одна из ключевых на данный момент
     настроек, влияет на потребление памяти, объём избыточной записи (write
     amplification) и эффективность распределения нагрузки по OSD.
 
-    Рекомендуемые по умолчанию размеры блока - 128 килобайт для SSD и 4
+    Рекомендуемые по умолчанию размеры блока - 128 килобайт для SSD и 1 мегабайт
-    мегабайта для HDD. В принципе, для SSD можно тоже использовать 4 мегабайта,
+    для HDD. В принципе, для SSD можно тоже использовать блок размером 1 мегабайт,
     это понизит использование памяти, но ухудшит распределение нагрузки и в
     среднем увеличит WA.
 
     Потребление памяти OSD составляет примерно (РАЗМЕР / БЛОК * 68 байт),
     т.е. примерно 544 МБ памяти на 1 ТБ занятого места на диске при
-    стандартном 128 КБ блоке.
+    стандартном 128 КБ блоке. При 1 МБ блоке памяти нужно в 8 раз меньше.
 - name: bitmap_granularity
   type: int
   default: 4096

docs/installation/proxmox.ru.md

@@ -1,4 +1,4 @@
-[Документация](../../README-ru.md#документация) → Установка → Proxmox
+[Документация](../../README-ru.md#документация) → Установка → Proxmox VE
 
 -----
 

docs/installation/source.en.md

@@ -21,7 +21,7 @@
 
 ## Basic instructions
 
-Download source, for example using git: `git clone --recurse-submodules https://yourcmc.ru/git/vitalif/vitastor/`
+Download source, for example using git: `git clone --recurse-submodules https://git.yourcmc.ru/vitalif/vitastor/`
 
 Get `fio` source and symlink it into `<vitastor>/fio`. If you don't want to build fio engine,
 you can disable it by passing `-DWITH_FIO=no` to cmake.
@@ -41,7 +41,7 @@ It's recommended to build the QEMU driver (qemu_driver.c) in-tree, as a part of
 QEMU build process. To do that:
 - Install vitastor client library headers (from source or from vitastor-client-dev package)
 - Take a corresponding patch from `patches/qemu-*-vitastor.patch` and apply it to QEMU source
-- Copy `src/qemu_driver.c` to QEMU source directory as `block/block-vitastor.c`
+- Copy `src/qemu_driver.c` to QEMU source directory as `block/vitastor.c`
 - Build QEMU as usual
 
 But it is also possible to build it out-of-tree. To do that:

docs/installation/source.ru.md

@@ -21,7 +21,7 @@
 
 ## Базовая инструкция
 
-Скачайте исходные коды, например, из git: `git clone --recurse-submodules https://yourcmc.ru/git/vitalif/vitastor/`
+Скачайте исходные коды, например, из git: `git clone --recurse-submodules https://git.yourcmc.ru/vitalif/vitastor/`
 
 Скачайте исходные коды пакета `fio`, распакуйте их и создайте символическую ссылку на них
 в директории исходников Vitastor: `<vitastor>/fio`. Либо, если вы не хотите собирать плагин fio,
@@ -41,7 +41,7 @@ cmake .. && make -j8 install
 Драйвер QEMU (qemu_driver.c) рекомендуется собирать вместе с самим QEMU. Для этого:
 - Установите заголовки клиентской библиотеки Vitastor (из исходников или из пакета vitastor-client-dev)
 - Возьмите соответствующий патч из `patches/qemu-*-vitastor.patch` и примените его к исходникам QEMU
-- Скопируйте [src/qemu_driver.c](../../src/qemu_driver.c) в директорию исходников QEMU как `block/block-vitastor.c`
+- Скопируйте [src/qemu_driver.c](../../src/qemu_driver.c) в директорию исходников QEMU как `block/vitastor.c`
 - Соберите QEMU как обычно
 
 Однако в целях отладки драйвер также можно собирать отдельно от QEMU. Для этого:
@@ -60,7 +60,7 @@ cmake .. && make -j8 install
       * Для QEMU 2.0+: `<qemu>/qapi-types.h` &rarr; `<vitastor>/qemu/b/qemu/qapi-types.h`
    - `config-host.h` и `qapi` нужны, т.к. в них содержатся автогенерируемые заголовки
 - Сконфигурируйте cmake Vitastor с `WITH_QEMU=yes` (`cmake .. -DWITH_QEMU=yes`) и, если вы
-  используете RHEL-подобый дистрибутив, также с `QEMU_PLUGINDIR=qemu-kvm`.
+  используете RHEL-подобный дистрибутив, также с `QEMU_PLUGINDIR=qemu-kvm`.
 - После этого в процессе сборки Vitastor также будет собираться подходящий для вашей
   версии QEMU `block-vitastor.so`.
 - Таким образом можно использовать драйвер даже с немодифицированным QEMU, но в этом случае

docs/intro/quickstart.en.md

@@ -7,6 +7,7 @@
 # Quick Start
 
 - [Preparation](#preparation)
+- [Recommended drives](#recommended-drives)
 - [Configure monitors](#configure-monitors)
 - [Configure OSDs](#configure-osds)
 - [Create a pool](#create-a-pool)
@@ -19,10 +20,20 @@
 - Get some SATA or NVMe SSDs with capacitors (server-grade drives). You can use desktop SSDs
   with lazy fsync, but prepare for inferior single-thread latency. Read more about capacitors
   [here](../config/layout-cluster.en.md#immediate_commit).
+- If you want to use HDDs, get modern HDDs with Media Cache or SSD Cache: HGST Ultrastar,
+  Toshiba MG08, Seagate EXOS or something similar. If your drives don't have such cache then
+  you also need small SSDs for journal and metadata (even 2 GB per 1 TB of HDD space is enough).
 - Get a fast network (at least 10 Gbit/s). Something like Mellanox ConnectX-4 with RoCEv2 is ideal.
 - Disable CPU powersaving: `cpupower idle-set -D 0 && cpupower frequency-set -g performance`.
 - [Install Vitastor packages](../installation/packages.en.md).
 
+## Recommended drives
+
+- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel D3-S4510/4520/4610/4620, Kingston DC500M
+- NVMe: Micron 9100/9200/9300/9400, Micron 7300/7450, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
+  Intel DC-P3700/P4500/P4600, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
+- HDD: HGST Ultrastar, Toshiba MG06/MG07/MG08, Seagate EXOS
+
 ## Configure monitors
 
 On the monitor hosts:
@@ -45,9 +56,10 @@ On the monitor hosts:
   }
   ```
 - Initialize OSDs:
-  - SSD-only: `vitastor-disk prepare /dev/sdXXX [/dev/sdYYY ...]`. You can add
+  - SSD-only or HDD-only: `vitastor-disk prepare /dev/sdXXX [/dev/sdYYY ...]`.
-    `--disable_data_fsync off` to leave disk cache enabled if you use desktop
+    Add `--disable_data_fsync off` to leave disk write cache enabled if you use
-    SSDs without capacitors.
+    desktop SSDs without capacitors. Do NOT add `--disable_data_fsync off` if you
+    use HDDs or SSD+HDD.
   - Hybrid, SSD+HDD: `vitastor-disk prepare --hybrid /dev/sdXXX [/dev/sdYYY ...]`.
     Pass all your devices (HDD and SSD) to this script — it will partition disks and initialize journals on its own.
     This script skips HDDs which are already partitioned so if you want to use non-empty disks for

docs/intro/quickstart.ru.md

@@ -7,6 +7,7 @@
 # Быстрый старт
 
 - [Подготовка](#подготовка)
+- [Рекомендуемые диски](#рекомендуемые-диски)
 - [Настройте мониторы](#настройте-мониторы)
 - [Настройте OSD](#настройте-osd)
 - [Создайте пул](#создайте-пул)
@@ -19,10 +20,20 @@
 - Возьмите серверы с SSD (SATA или NVMe), желательно с конденсаторами (серверные SSD). Можно
   использовать и десктопные SSD, включив режим отложенного fsync, но производительность будет хуже.
   О конденсаторах читайте [здесь](../config/layout-cluster.ru.md#immediate_commit).
+- Если хотите использовать HDD, берите современные модели с Media или SSD кэшем - HGST Ultrastar,
+  Toshiba MG08, Seagate EXOS или что-то похожее. Если такого кэша у ваших дисков нет,
+  обязательно возьмите SSD под метаданные и журнал (маленькие, буквально 2 ГБ на 1 ТБ HDD-места).
 - Возьмите быструю сеть, минимум 10 гбит/с. Идеал - что-то вроде Mellanox ConnectX-4 с RoCEv2.
 - Для лучшей производительности отключите энергосбережение CPU: `cpupower idle-set -D 0 && cpupower frequency-set -g performance`.
 - [Установите пакеты Vitastor](../installation/packages.ru.md).
 
+## Рекомендуемые диски
+
+- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel D3-S4510/4520/4610/4620, Kingston DC500M
+- NVMe: Micron 9100/9200/9300/9400, Micron 7300/7450, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
+  Intel DC-P3700/P4500/P4600, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
+- HDD: HGST Ultrastar, Toshiba MG06/MG07/MG08, Seagate EXOS
+
 ## Настройте мониторы
 
 На хостах, выделенных под мониторы:
@@ -45,9 +56,10 @@
   }
   ```
 - Инициализуйте OSD:
-  - SSD: `vitastor-disk prepare /dev/sdXXX [/dev/sdYYY ...]`. Если вы используете
+  - Только SSD или только HDD: `vitastor-disk prepare /dev/sdXXX [/dev/sdYYY ...]`.
-    десктопные SSD без конденсаторов, можете оставить кэш включённым, добавив
+    Если вы используете десктопные SSD без конденсаторов, добавьте опцию `--disable_data_fsync off`,
-    опцию `--disable_data_fsync off`.
+    чтобы оставить кэш записи диска включённым. НЕ добавляйте эту опцию, если используете
+    жёсткие диски (HDD).
   - Гибридные, SSD+HDD: `vitastor-disk prepare --hybrid /dev/sdXXX [/dev/sdYYY ...]`.
     Передайте все ваши SSD и HDD скрипту в командной строке подряд, скрипт автоматически выделит
     разделы под журналы на SSD и данные на HDD. Скрипт пропускает HDD, на которых уже есть разделы

docs/usage/nbd.en.md

@@ -13,6 +13,8 @@ remains decent (see an example [here](../performance/comparison1.en.md#vitastor-
 
 Vitastor Kubernetes CSI driver is based on NBD.
 
+See also [VDUSE](qemu.en.md#vduse).
+
 ## Map image
 
 To create a local block device for a Vitastor image run:

docs/usage/nbd.ru.md

@@ -16,6 +16,8 @@ NBD немного снижает производительность из-за
 
 CSI-драйвер Kubernetes Vitastor основан на NBD.
 
+Смотрите также [VDUSE](qemu.ru.md#vduse).
+
 ## Подключить устройство
 
 Чтобы создать локальное блочное устройство для образа, выполните команду:

docs/usage/qemu.en.md

@@ -83,3 +83,44 @@ qemu-img rebase -u -b '' testimg.qcow2
 This can be used for backups. Just note that exporting an image that is currently being written to
 is of course unsafe and doesn't produce a consistent result, so only export snapshots if you do this
 on a live VM.
+
+## VDUSE
+
+Linux kernel, starting with version 5.15, supports a new interface for attaching virtual disks
+to the host - VDUSE (vDPA Device in Userspace). QEMU, starting with 7.2, has support for
+exporting QEMU block devices over this protocol using qemu-storage-daemon.
+
+VDUSE has the same problem as other FUSE-like interfaces in Linux: if a userspace process hangs,
+for example, if it loses connectivity with Vitastor cluster - active processes doing I/O may
+hang in the D state (uninterruptible sleep) and you won't be able to kill them even with kill -9.
+In this case reboot will be the only way to remove VDUSE devices from system.
+
+On the other hand, VDUSE is faster than [NBD](nbd.en.md), so you may prefer to use it if
+performance is important for you. Approximate performance numbers:
+direct fio benchmark - 115000 iops, NBD - 60000 iops, VDUSE - 90000 iops.
+
+To try VDUSE you need at least Linux 5.15, built with VDUSE support
+(CONFIG_VIRTIO_VDPA=m and CONFIG_VDPA_USER=m). Debian Linux kernels have these options
+disabled by now, so if you want to try it on Debian, use a kernel from Ubuntu
+[kernel-ppa/mainline](https://kernel.ubuntu.com/~kernel-ppa/mainline/) or Proxmox.
+
+Commands to attach Vitastor image as a VDUSE device:
+
+```
+modprobe vduse
+modprobe virtio-vdpa
+qemu-storage-daemon --daemonize --blockdev '{"node-name":"test1","driver":"vitastor",\
+  "etcd-host":"192.168.7.2:2379/v3","image":"testosd1","cache":{"direct":true,"no-flush":false},"discard":"unmap"}' \
+  --export vduse-blk,id=test1,node-name=test1,name=test1,num-queues=16,queue-size=128,writable=true
+vdpa dev add name test1 mgmtdev vduse
+```
+
+After running these commands /dev/vda device will appear in the system and you'll be able to
+use it as a normal disk.
+
+To remove the device:
+
+```
+vdpa dev del test1
+kill <qemu-storage-daemon_process_PID>
+```

docs/usage/qemu.ru.md

@@ -87,3 +87,44 @@ qemu-img rebase -u -b '' testimg.qcow2
 Это можно использовать для резервного копирования. Только помните, что экспортировать образ, в который
 в то же время идёт запись, небезопасно - результат чтения не будет целостным. Так что если вы работаете
 с активными виртуальными машинами, экспортируйте только их снимки, но не сам образ.
+
+## VDUSE
+
+В Linux, начиная с версии ядра 5.15, доступен новый интерфейс для подключения виртуальных дисков
+к системе - VDUSE (vDPA Device in Userspace), а в QEMU, начиная с версии 7.2, есть поддержка
+экспорта блочных устройств QEMU по этому протоколу через qemu-storage-daemon.
+
+VDUSE страдает общей проблемой FUSE-подобных интерфейсов в Linux: если пользовательский процесс
+подвиснет, например, если будет потеряна связь с кластером Vitastor - читающие/пишущие в кластер
+процессы могут "залипнуть" в состоянии D (непрерываемый сон) и их будет невозможно убить даже
+через kill -9. В этом случае удалить из системы устройство можно только перезагрузившись.
+
+С другой стороны, VDUSE быстрее по сравнению с [NBD](nbd.ru.md), поэтому его может
+быть предпочтительно использовать там, где производительность важнее. Порядок показателей:
+прямое тестирование через fio - 115000 iops, NBD - 60000 iops, VDUSE - 90000 iops.
+
+Чтобы использовать VDUSE, вам нужно ядро Linux версии хотя бы 5.15, собранное с поддержкой
+VDUSE (CONFIG_VIRTIO_VDPA=m и CONFIG_VDPA_USER=m). В ядрах в Debian Linux поддержка пока
+отключена - если хотите попробовать эту функцию на Debian, поставьте ядро из Ubuntu
+[kernel-ppa/mainline](https://kernel.ubuntu.com/~kernel-ppa/mainline/) или из Proxmox.
+
+Команды для подключения виртуального диска через VDUSE:
+
+```
+modprobe vduse
+modprobe virtio-vdpa
+qemu-storage-daemon --daemonize --blockdev '{"node-name":"test1","driver":"vitastor",\
+  "etcd-host":"192.168.7.2:2379/v3","image":"testosd1","cache":{"direct":true,"no-flush":false},"discard":"unmap"}' \
+  --export vduse-blk,id=test1,node-name=test1,name=test1,num-queues=16,queue-size=128,writable=true
+vdpa dev add name test1 mgmtdev vduse
+```
+
+После этого в системе появится устройство /dev/vda, которое можно будет использовать как
+обычный диск.
+
+Для удаления устройства из системы:
+
+```
+vdpa dev del test1
+kill <PID_процесса_qemu-storage-daemon>
+```

mon/make-etcd

@@ -63,8 +63,9 @@ Wants=network-online.target local-fs.target time-sync.target
 
 [Service]
 Restart=always
+Environment=GOGC=50
 ExecStart=etcd -name etcd${num} --data-dir /var/lib/etcd${num}.etcd \\
-    --advertise-client-urls http://${etcds[num]}:2379 --listen-client-urls http://${etcds[num]}:2379 \\
+    --snapshot-count 10000 --advertise-client-urls http://${etcds[num]}:2379 --listen-client-urls http://${etcds[num]}:2379 \\
     --initial-advertise-peer-urls http://${etcds[num]}:2380 --listen-peer-urls http://${etcds[num]}:2380 \\
     --initial-cluster-token vitastor-etcd-1 --initial-cluster ${etcd_cluster} \\
     --initial-cluster-state new --max-txn-ops=100000 --max-request-bytes=104857600 \\

mon/mon.js

@@ -1497,10 +1497,14 @@ class Mon
                     break;
                 }
             }
+            const pool_cfg = (this.state.config.pools[pool_id]||{});
             if (!object_size)
             {
-                object_size = (this.state.config.pools[pool_id]||{}).block_size ||
+                object_size = pool_cfg.block_size || this.config.block_size || 131072;
-                    this.config.block_size || 131072;
+            }
+            if (pool_cfg.scheme !== 'replicated')
+            {
+                object_size *= ((pool_cfg.pg_size||0) - (pool_cfg.parity_chunks||0));
             }
             object_size = BigInt(object_size);
             for (const pg_num in this.state.pg.stats[pool_id])
@@ -1608,7 +1612,7 @@ class Mon
                 }
             }
         }
-        return inode_stats;
+        return { inode_stats, seen_pools };
     }
 
     serialize_bigints(obj)
@@ -1634,7 +1638,7 @@ class Mon
         const timestamp = Date.now();
         const { object_counts, object_bytes } = this.sum_object_counts();
         let stats = this.sum_op_stats(timestamp, this.prev_stats);
-        let inode_stats = this.sum_inode_stats(
+        let { inode_stats, seen_pools } = this.sum_inode_stats(
             this.prev_stats ? this.prev_stats.inode_stats : null,
             timestamp, this.prev_stats ? this.prev_stats.timestamp : null
         );
@@ -1669,12 +1673,22 @@ class Mon
         }
         for (const pool_id in this.state.pool.stats)
         {
-            const pool_stats = { ...this.state.pool.stats[pool_id] };
+            if (!seen_pools[pool_id])
-            this.serialize_bigints(pool_stats);
+            {
-            txn.push({ requestPut: {
+                txn.push({ requestDeleteRange: {
-                key: b64(this.etcd_prefix+'/pool/stats/'+pool_id),
+                    key: b64(this.etcd_prefix+'/pool/stats/'+pool_id),
-                value: b64(JSON.stringify(pool_stats)),
+                } });
-            } });
+                delete this.state.pool.stats[pool_id];
+            }
+            else
+            {
+                const pool_stats = { ...this.state.pool.stats[pool_id] };
+                this.serialize_bigints(pool_stats);
+                txn.push({ requestPut: {
+                    key: b64(this.etcd_prefix+'/pool/stats/'+pool_id),
+                    value: b64(JSON.stringify(pool_stats)),
+                } });
+            }
         }
         if (txn.length)
         {

patches/cinder-vitastor.py

@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils
 
-VERSION = '0.9.2'
+VERSION = '0.9.6'
 
 LOG = logging.getLogger(__name__)
 

patches/qemu-2.12-vitastor.patch

@@ -0,0 +1,176 @@
+diff --git a/block/Makefile.objs b/block/Makefile.objs
+index d644bac60a..e404236291 100644
+--- a/block/Makefile.objs
++++ b/block/Makefile.objs
+@@ -19,6 +19,7 @@ block-obj-$(if $(CONFIG_LIBISCSI),y,n) += iscsi-opts.o
+ block-obj-$(CONFIG_LIBNFS) += nfs.o
+ block-obj-$(CONFIG_CURL) += curl.o
+ block-obj-$(CONFIG_RBD) += rbd.o
++block-obj-$(CONFIG_VITASTOR) += vitastor.o
+ block-obj-$(CONFIG_GLUSTERFS) += gluster.o
+ block-obj-$(CONFIG_VXHS) += vxhs.o
+ block-obj-$(CONFIG_LIBSSH2) += ssh.o
+@@ -39,6 +40,8 @@ curl.o-cflags      := $(CURL_CFLAGS)
+ curl.o-libs        := $(CURL_LIBS)
+ rbd.o-cflags       := $(RBD_CFLAGS)
+ rbd.o-libs         := $(RBD_LIBS)
++vitastor.o-cflags  := $(VITASTOR_CFLAGS)
++vitastor.o-libs    := $(VITASTOR_LIBS)
+ gluster.o-cflags   := $(GLUSTERFS_CFLAGS)
+ gluster.o-libs     := $(GLUSTERFS_LIBS)
+ vxhs.o-libs        := $(VXHS_LIBS)
+diff --git a/configure b/configure
+index 0a19b033bc..58b7fbf24c 100755
+--- a/configure
++++ b/configure
+@@ -398,6 +398,7 @@ trace_backends="log"
+ trace_file="trace"
+ spice=""
+ rbd=""
++vitastor=""
+ smartcard=""
+ libusb=""
+ usb_redir=""
+@@ -1213,6 +1214,10 @@ for opt do
+   ;;
+   --enable-rbd) rbd="yes"
+   ;;
++  --disable-vitastor) vitastor="no"
++  ;;
++  --enable-vitastor) vitastor="yes"
++  ;;
+   --disable-xfsctl) xfs="no"
+   ;;
+   --enable-xfsctl) xfs="yes"
+@@ -1601,6 +1606,7 @@ disabled with --disable-FEATURE, default is enabled if available:
+   vhost-crypto    vhost-crypto acceleration support
+   spice           spice
+   rbd             rados block device (rbd)
++  vitastor        vitastor block device
+   libiscsi        iscsi support
+   libnfs          nfs support
+   smartcard       smartcard support (libcacard)
+@@ -3594,6 +3600,27 @@ EOF
+   fi
+ fi
+ 
++##########################################
++# vitastor probe
++if test "$vitastor" != "no" ; then
++  cat > $TMPC <<EOF
++#include <vitastor_c.h>
++int main(void) {
++  vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
++  return 0;
++}
++EOF
++  vitastor_libs="-lvitastor_client"
++  if compile_prog "" "$vitastor_libs" ; then
++    vitastor=yes
++  else
++    if test "$vitastor" = "yes" ; then
++      feature_not_found "vitastor block device" "Install vitastor-client-dev"
++    fi
++    vitastor=no
++  fi
++fi
++
+ ##########################################
+ # libssh2 probe
+ min_libssh2_version=1.2.8
+@@ -5837,6 +5864,7 @@ echo "Trace output file $trace_file-<pid>"
+ fi
+ echo "spice support     $spice $(echo_version $spice $spice_protocol_version/$spice_server_version)"
+ echo "rbd support       $rbd"
++echo "vitastor support  $vitastor"
+ echo "xfsctl support    $xfs"
+ echo "smartcard support $smartcard"
+ echo "libusb            $libusb"
+@@ -6416,6 +6444,11 @@ if test "$rbd" = "yes" ; then
+   echo "RBD_CFLAGS=$rbd_cflags" >> $config_host_mak
+   echo "RBD_LIBS=$rbd_libs" >> $config_host_mak
+ fi
++if test "$vitastor" = "yes" ; then
++  echo "CONFIG_VITASTOR=m" >> $config_host_mak
++  echo "VITASTOR_CFLAGS=$vitastor_cflags" >> $config_host_mak
++  echo "VITASTOR_LIBS=$vitastor_libs" >> $config_host_mak
++fi
+ 
+ echo "CONFIG_COROUTINE_BACKEND=$coroutine" >> $config_host_mak
+ if test "$coroutine_pool" = "yes" ; then
+diff --git a/qapi/block-core.json b/qapi/block-core.json
+index c50517bff3..c780bb2c1c 100644
+--- a/qapi/block-core.json
++++ b/qapi/block-core.json
+@@ -2514,7 +2514,7 @@
+             'dmg', 'file', 'ftp', 'ftps', 'gluster', 'host_cdrom',
+             'host_device', 'http', 'https', 'iscsi', 'luks', 'nbd', 'nfs',
+             'null-aio', 'null-co', 'nvme', 'parallels', 'qcow', 'qcow2', 'qed',
+-            'quorum', 'raw', 'rbd', 'replication', 'sheepdog', 'ssh',
++            'quorum', 'raw', 'rbd', 'vitastor', 'replication', 'sheepdog', 'ssh',
+             'throttle', 'vdi', 'vhdx', 'vmdk', 'vpc', 'vvfat', 'vxhs' ] }
+ 
+ ##
+@@ -3217,6 +3217,28 @@
+             '*snap-id': 'uint32',
+             '*tag': 'str' } }
+ 
++##
++# @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:
+ #
+@@ -3547,6 +3569,7 @@
+       'rbd':        'BlockdevOptionsRbd',
+       'replication':'BlockdevOptionsReplication',
+       'sheepdog':   'BlockdevOptionsSheepdog',
++      'vitastor':   'BlockdevOptionsVitastor',
+       'ssh':        'BlockdevOptionsSsh',
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+@@ -3991,6 +4014,17 @@
+             '*subformat':           'BlockdevVhdxSubformat',
+             '*block-state-zero':    'bool' } }
+ 
++##
++# @BlockdevCreateOptionsVitastor:
++#
++# Driver specific image creation options for Vitastor.
++#
++# @size: Size of the virtual disk in bytes
++##
++{ 'struct': 'BlockdevCreateOptionsVitastor',
++  'data': { 'location':         'BlockdevOptionsVitastor',
++            'size':             'size' } }
++
+ ##
+ # @BlockdevVpcSubformat:
+ #
+@@ -4074,6 +4108,7 @@
+       'rbd':            'BlockdevCreateOptionsRbd',
+       'replication':    'BlockdevCreateNotSupported',
+       'sheepdog':       'BlockdevCreateOptionsSheepdog',
++      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'throttle':       'BlockdevCreateNotSupported',
+       'vdi':            'BlockdevCreateOptionsVdi',

patches/qemu-5.2-vitastor.patch

@@ -0,0 +1,181 @@
+Index: qemu-5.2+dfsg/qapi/block-core.json
+===================================================================
+--- qemu-5.2+dfsg.orig/qapi/block-core.json
++++ qemu-5.2+dfsg/qapi/block-core.json
+@@ -2831,7 +2831,7 @@
+             'luks', 'nbd', 'nfs', 'null-aio', 'null-co', 'nvme', 'parallels',
+             'qcow', 'qcow2', 'qed', 'quorum', 'raw', 'rbd',
+             { 'name': 'replication', 'if': 'defined(CONFIG_REPLICATION)' },
+-            'sheepdog',
++            'sheepdog', 'vitastor',
+             'ssh', 'throttle', 'vdi', 'vhdx', 'vmdk', 'vpc', 'vvfat' ] }
+ 
+ ##
+@@ -3668,6 +3668,28 @@
+             '*tag': 'str' } }
+ 
+ ##
++# @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:
+ #
+ # An enumeration of replication modes.
+@@ -4015,6 +4037,7 @@
+       'replication': { 'type': 'BlockdevOptionsReplication',
+                        'if': 'defined(CONFIG_REPLICATION)' },
+       'sheepdog':   'BlockdevOptionsSheepdog',
++      'vitastor':   'BlockdevOptionsVitastor',
+       'ssh':        'BlockdevOptionsSsh',
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+@@ -4404,6 +4427,17 @@
+             '*cluster-size' :   'size' } }
+ 
+ ##
++# @BlockdevCreateOptionsVitastor:
++#
++# Driver specific image creation options for Vitastor.
++#
++# @size: Size of the virtual disk in bytes
++##
++{ 'struct': 'BlockdevCreateOptionsVitastor',
++  'data': { 'location':         'BlockdevOptionsVitastor',
++            'size':             'size' } }
++
++##
+ # @BlockdevVmdkSubformat:
+ #
+ # Subformat options for VMDK images
+@@ -4665,6 +4699,7 @@
+       'qed':            'BlockdevCreateOptionsQed',
+       'rbd':            'BlockdevCreateOptionsRbd',
+       'sheepdog':       'BlockdevCreateOptionsSheepdog',
++      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'vdi':            'BlockdevCreateOptionsVdi',
+       'vhdx':           'BlockdevCreateOptionsVhdx',
+Index: qemu-5.2+dfsg/block/meson.build
+===================================================================
+--- qemu-5.2+dfsg.orig/block/meson.build
++++ qemu-5.2+dfsg/block/meson.build
+@@ -76,6 +76,7 @@ foreach m : [
+   ['CONFIG_LIBNFS', 'nfs', libnfs, 'nfs.c'],
+   ['CONFIG_LIBSSH', 'ssh', libssh, 'ssh.c'],
+   ['CONFIG_RBD', 'rbd', rbd, 'rbd.c'],
++  ['CONFIG_VITASTOR', 'vitastor', vitastor, 'vitastor.c'],
+ ]
+   if config_host.has_key(m[0])
+     if enable_modules
+Index: qemu-5.2+dfsg/configure
+===================================================================
+--- qemu-5.2+dfsg.orig/configure
++++ qemu-5.2+dfsg/configure
+@@ -372,6 +372,7 @@ trace_backends="log"
+ trace_file="trace"
+ spice=""
+ rbd=""
++vitastor=""
+ smartcard=""
+ u2f="auto"
+ libusb=""
+@@ -1263,6 +1264,10 @@ for opt do
+   ;;
+   --enable-rbd) rbd="yes"
+   ;;
++  --disable-vitastor) vitastor="no"
++  ;;
++  --enable-vitastor) vitastor="yes"
++  ;;
+   --disable-xfsctl) xfs="no"
+   ;;
+   --enable-xfsctl) xfs="yes"
+@@ -1827,6 +1832,7 @@ disabled with --disable-FEATURE, default
+   vhost-vdpa      vhost-vdpa kernel backend support
+   spice           spice
+   rbd             rados block device (rbd)
++  vitastor        vitastor block device
+   libiscsi        iscsi support
+   libnfs          nfs support
+   smartcard       smartcard support (libcacard)
+@@ -3719,6 +3725,27 @@ EOF
+ fi
+ 
+ ##########################################
++# vitastor probe
++if test "$vitastor" != "no" ; then
++  cat > $TMPC <<EOF
++#include <vitastor_c.h>
++int main(void) {
++  vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
++  return 0;
++}
++EOF
++  vitastor_libs="-lvitastor_client"
++  if compile_prog "" "$vitastor_libs" ; then
++    vitastor=yes
++  else
++    if test "$vitastor" = "yes" ; then
++      feature_not_found "vitastor block device" "Install vitastor-client-dev"
++    fi
++    vitastor=no
++  fi
++fi
++
++##########################################
+ # libssh probe
+ if test "$libssh" != "no" ; then
+   if $pkg_config --exists libssh; then
+@@ -6456,6 +6483,10 @@ if test "$rbd" = "yes" ; then
+   echo "CONFIG_RBD=y" >> $config_host_mak
+   echo "RBD_LIBS=$rbd_libs" >> $config_host_mak
+ fi
++if test "$vitastor" = "yes" ; then
++  echo "CONFIG_VITASTOR=y" >> $config_host_mak
++  echo "VITASTOR_LIBS=$vitastor_libs" >> $config_host_mak
++fi
+ 
+ echo "CONFIG_COROUTINE_BACKEND=$coroutine" >> $config_host_mak
+ if test "$coroutine_pool" = "yes" ; then
+Index: qemu-5.2+dfsg/meson.build
+===================================================================
+--- qemu-5.2+dfsg.orig/meson.build
++++ qemu-5.2+dfsg/meson.build
+@@ -596,6 +596,10 @@ rbd = not_found
+ if 'CONFIG_RBD' in config_host
+   rbd = declare_dependency(link_args: config_host['RBD_LIBS'].split())
+ endif
++vitastor = not_found
++if 'CONFIG_VITASTOR' in config_host
++  vitastor = declare_dependency(link_args: config_host['VITASTOR_LIBS'].split())
++endif
+ glusterfs = not_found
+ if 'CONFIG_GLUSTERFS' in config_host
+   glusterfs = declare_dependency(compile_args: config_host['GLUSTERFS_CFLAGS'].split(),
+@@ -2145,6 +2149,7 @@ endif
+ # TODO: add back protocol and server version
+ summary_info += {'spice support':     config_host.has_key('CONFIG_SPICE')}
+ summary_info += {'rbd support':       config_host.has_key('CONFIG_RBD')}
++summary_info += {'vitastor support':  config_host.has_key('CONFIG_VITASTOR')}
+ summary_info += {'xfsctl support':    config_host.has_key('CONFIG_XFS')}
+ summary_info += {'smartcard support': config_host.has_key('CONFIG_SMARTCARD')}
+ summary_info += {'U2F support':       u2f.found()}

rpm/build-tarball.sh

@@ -24,4 +24,4 @@ rm fio
 mv fio-copy fio
 FIO=`rpm -qi fio | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Version[\s:]+(\S+)/ && print $1; /^Release[\s:]+(\S+)/ && print "-$1"; }'`
 perl -i -pe 's/(Requires:\s*fio)([^\n]+)?/$1 = '$FIO'/' $VITASTOR/rpm/vitastor-el$EL.spec
-tar --transform 's#^#vitastor-0.9.2/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.9.2$(rpm --eval '%dist').tar.gz *
+tar --transform 's#^#vitastor-0.9.6/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.9.6$(rpm --eval '%dist').tar.gz *

rpm/qemu-kvm-4.2-el7.spec.patch

@@ -22,7 +22,7 @@
  Name: qemu-kvm
  Version: 4.2.0
 -Release: 29.vitastor%{?dist}.6
-+Release: 32.vitastor%{?dist}.6
++Release: 34.vitastor%{?dist}.6
  # Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped
  Epoch: 15
  License: GPLv2 and GPLv2+ and CC-BY

rpm/qemu-kvm-4.2-el8.spec.patch

@@ -13,7 +13,7 @@
  Name: qemu-kvm
  Version: 4.2.0
 -Release: 29%{?dist}.6
-+Release: 32.vitastor%{?dist}.6
++Release: 33.vitastor%{?dist}.6
  # Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped
  Epoch: 15
  License: GPLv2 and GPLv2+ and CC-BY

rpm/qemu-kvm-6.2-el8.spec.patch

@@ -0,0 +1,103 @@
+--- qemu-kvm-6.2.spec.orig	2023-07-18 13:52:57.636625440 +0000
++++ qemu-kvm-6.2.spec	2023-07-18 13:52:19.011683886 +0000
+@@ -73,6 +73,7 @@ Requires: %{name}-hw-usbredir = %{epoch}
+ %endif                                                           \
+ Requires: %{name}-block-iscsi = %{epoch}:%{version}-%{release}   \
+ Requires: %{name}-block-rbd = %{epoch}:%{version}-%{release}     \
++Requires: %{name}-block-vitastor = %{epoch}:%{version}-%{release}\
+ Requires: %{name}-block-ssh = %{epoch}:%{version}-%{release}
+ 
+ # Macro to properly setup RHEL/RHEV conflict handling
+@@ -83,7 +84,7 @@ Obsoletes: %1-rhev <= %{epoch}:%{version
+ Summary: QEMU is a machine emulator and virtualizer
+ Name: qemu-kvm
+ Version: 6.2.0
+-Release: 32%{?rcrel}%{?dist}
++Release: 32.vitastor%{?rcrel}%{?dist}
+ # Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped
+ Epoch: 15
+ License: GPLv2 and GPLv2+ and CC-BY
+@@ -122,6 +123,7 @@ Source37: tests_data_acpi_pc_SSDT.dimmpx
+ Source38: tests_data_acpi_q35_FACP.slic
+ Source39: tests_data_acpi_q35_SSDT.dimmpxm
+ Source40: tests_data_acpi_virt_SSDT.memhp
++Source41: qemu-vitastor.c
+ 
+ Patch0001: 0001-redhat-Adding-slirp-to-the-exploded-tree.patch
+ Patch0005: 0005-Initial-redhat-build.patch
+@@ -652,6 +654,7 @@ Patch255: kvm-scsi-protect-req-aiocb-wit
+ Patch256: kvm-dma-helpers-prevent-dma_blk_cb-vs-dma_aio_cancel-rac.patch
+ # For bz#2090990 - qemu crash with error scsi_req_unref(SCSIRequest *): Assertion `req->refcount > 0' failed or scsi_dma_complete(void *, int): Assertion `r->req.aiocb != NULL' failed [8.7.0]
+ Patch257: kvm-virtio-scsi-reset-SCSI-devices-from-main-loop-thread.patch
++Patch258: qemu-6.2-vitastor.patch
+ 
+ BuildRequires: wget
+ BuildRequires: rpm-build
+@@ -689,6 +692,7 @@ BuildRequires: libcurl-devel
+ BuildRequires: libssh-devel
+ BuildRequires: librados-devel
+ BuildRequires: librbd-devel
++BuildRequires: vitastor-client-devel
+ %if %{have_gluster}
+ # For gluster block driver
+ BuildRequires: glusterfs-api-devel
+@@ -926,6 +930,14 @@ Install this package if you want to acce
+ using the rbd protocol.
+ 
+ 
++%package  block-vitastor
++Summary: QEMU Vitastor block driver
++Requires: %{name}-common%{?_isa} = %{epoch}:%{version}-%{release}
++
++%description block-vitastor
++This package provides the additional Vitastor block driver for QEMU.
++
++
+ %package  block-ssh
+ Summary: QEMU SSH block driver
+ Requires: %{name}-common%{?_isa} = %{epoch}:%{version}-%{release}
+@@ -979,6 +991,7 @@ This package provides usbredir support.
+ rm -fr slirp
+ mkdir slirp
+ %autopatch -p1
++cp %{SOURCE41} ./block/vitastor.c
+ 
+ %global qemu_kvm_build qemu_kvm_build
+ mkdir -p %{qemu_kvm_build}
+@@ -994,7 +1007,7 @@ cp -f %{SOURCE40} tests/data/acpi/virt/S
+ # --build-id option is used for giving info to the debug packages.
+ buildldflags="VL_LDFLAGS=-Wl,--build-id"
+ 
+-%global block_drivers_list qcow2,raw,file,host_device,nbd,iscsi,rbd,blkdebug,luks,null-co,nvme,copy-on-read,throttle
++%global block_drivers_list qcow2,raw,file,host_device,nbd,iscsi,rbd,vitastor,blkdebug,luks,null-co,nvme,copy-on-read,throttle
+ 
+ %if 0%{have_gluster}
+     %global block_drivers_list %{block_drivers_list},gluster
+@@ -1149,9 +1162,7 @@ pushd %{qemu_kvm_build}
+   --firmwarepath=%{_prefix}/share/qemu-firmware \
+   --meson="git" \
+   --target-list="%{buildarch}" \
+-  --block-drv-rw-whitelist=%{block_drivers_list} \
+   --audio-drv-list= \
+-  --block-drv-ro-whitelist=vmdk,vhdx,vpc,https,ssh \
+   --with-coroutine=ucontext \
+   --with-git=git \
+   --tls-priority=@QEMU,SYSTEM \
+@@ -1197,6 +1208,7 @@ pushd %{qemu_kvm_build}
+ %endif
+   --enable-pie \
+   --enable-rbd \
++  --enable-vitastor \
+ %if 0%{have_librdma}
+   --enable-rdma \
+ %endif
+@@ -1794,6 +1806,9 @@ sh %{_sysconfdir}/sysconfig/modules/kvm.
+ %files block-rbd
+ %{_libdir}/qemu-kvm/block-rbd.so
+ 
++%files block-vitastor
++%{_libdir}/qemu-kvm/block-vitastor.so
++
+ %files block-ssh
+ %{_libdir}/qemu-kvm/block-ssh.so
+ 

rpm/qemu-kvm-7.2-el9.spec.patch

@@ -0,0 +1,93 @@
+--- qemu-kvm-7.2.spec.orig	2023-06-22 13:56:19.000000000 +0000
++++ qemu-kvm-7.2.spec	2023-07-18 07:55:22.347090196 +0000
+@@ -100,8 +100,6 @@
+ %endif
+ 
+ %global target_list %{kvm_target}-softmmu
+-%global block_drivers_rw_list qcow2,raw,file,host_device,nbd,iscsi,rbd,blkdebug,luks,null-co,nvme,copy-on-read,throttle,compress
+-%global block_drivers_ro_list vdi,vmdk,vhdx,vpc,https
+ %define qemudocdir %{_docdir}/%{name}
+ %global firmwaredirs "%{_datadir}/qemu-firmware:%{_datadir}/ipxe/qemu:%{_datadir}/seavgabios:%{_datadir}/seabios"
+ 
+@@ -126,6 +124,7 @@ Requires: %{name}-device-usb-host = %{ep
+ Requires: %{name}-device-usb-redirect = %{epoch}:%{version}-%{release}   \
+ %endif                                                           \
+ Requires: %{name}-block-rbd = %{epoch}:%{version}-%{release}     \
++Requires: %{name}-block-vitastor = %{epoch}:%{version}-%{release}\
+ Requires: %{name}-audio-pa = %{epoch}:%{version}-%{release}
+ 
+ # Since SPICE is removed from RHEL-9, the following Obsoletes:
+@@ -148,7 +147,7 @@ Obsoletes: %{name}-block-ssh <= %{epoch}
+ Summary: QEMU is a machine emulator and virtualizer
+ Name: qemu-kvm
+ Version: 7.2.0
+-Release: 14%{?rcrel}%{?dist}%{?cc_suffix}.1
++Release: 14.vitastor%{?rcrel}%{?dist}%{?cc_suffix}.1
+ # Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped
+ # Epoch 15 used for RHEL 8
+ # Epoch 17 used for RHEL 9 (due to release versioning offset in RHEL 8.5)
+@@ -171,6 +170,7 @@ Source28: 95-kvm-memlock.conf
+ Source30: kvm-s390x.conf
+ Source31: kvm-x86.conf
+ Source36: README.tests
++Source37: qemu-vitastor.c
+ 
+ 
+ Patch0004: 0004-Initial-redhat-build.patch
+@@ -418,6 +418,7 @@ Patch134: kvm-target-i386-Fix-BZHI-instr
+ Patch135: kvm-intel-iommu-fail-DEVIOTLB_UNMAP-without-dt-mode.patch
+ # For bz#2203745 - Disk detach is unsuccessful while the guest is still booting [rhel-9.2.0.z]
+ Patch136: kvm-acpi-pcihp-allow-repeating-hot-unplug-requests.patch
++Patch137: qemu-7.2-vitastor.patch
+ 
+ %if %{have_clang}
+ BuildRequires: clang
+@@ -449,6 +450,7 @@ BuildRequires: libcurl-devel
+ %if %{have_block_rbd}
+ BuildRequires: librbd-devel
+ %endif
++BuildRequires: vitastor-client-devel
+ # We need both because the 'stap' binary is probed for by configure
+ BuildRequires: systemtap
+ BuildRequires: systemtap-sdt-devel
+@@ -642,6 +644,14 @@ using the rbd protocol.
+ %endif
+ 
+ 
++%package  block-vitastor
++Summary: QEMU Vitastor block driver
++Requires: %{name}-common%{?_isa} = %{epoch}:%{version}-%{release}
++
++%description block-vitastor
++This package provides the additional Vitastor block driver for QEMU.
++
++
+ %package  audio-pa
+ Summary: QEMU PulseAudio audio driver
+ Requires: %{name}-common%{?_isa} = %{epoch}:%{version}-%{release}
+@@ -719,6 +729,7 @@ This package provides usbredir support.
+ %prep
+ %setup -q -n qemu-%{version}%{?rcstr}
+ %autopatch -p1
++cp %{SOURCE37} ./block/vitastor.c
+ 
+ %global qemu_kvm_build qemu_kvm_build
+ mkdir -p %{qemu_kvm_build}
+@@ -946,6 +957,7 @@ run_configure \
+ %if %{have_block_rbd}
+   --enable-rbd \
+ %endif
++  --enable-vitastor \
+ %if %{have_librdma}
+   --enable-rdma \
+ %endif
+@@ -1426,6 +1438,9 @@ useradd -r -u 107 -g qemu -G kvm -d / -s
+ %files block-rbd
+ %{_libdir}/%{name}/block-rbd.so
+ %endif
++%files block-vitastor
++%{_libdir}/%{name}/block-vitastor.so
++
+ %files audio-pa
+ %{_libdir}/%{name}/audio-pa.so
+ 

rpm/vitastor-el7.Dockerfile

@@ -35,7 +35,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.9.2.el7.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.9.6.el7.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el7.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.9.2
+Version:        0.9.6
 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-0.9.2.el7.tar.gz
+Source0:        vitastor-0.9.6.el7.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

rpm/vitastor-el8.Dockerfile

@@ -35,7 +35,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.9.2.el8.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.9.6.el8.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el8.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.9.2
+Version:        0.9.6
 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-0.9.2.el8.tar.gz
+Source0:        vitastor-0.9.6.el8.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

rpm/vitastor-el9.Dockerfile

@@ -18,7 +18,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.9.2.el9.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.9.6.el9.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el9.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el9.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.9.2
+Version:        0.9.6
 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-0.9.2.el9.tar.gz
+Source0:        vitastor-0.9.6.el9.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

src/CMakeLists.txt

@@ -16,7 +16,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 	set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
 endif()
 
-add_definitions(-DVERSION="0.9.2")
+add_definitions(-DVERSION="0.9.6")
 add_definitions(-Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fdiagnostics-color=always -I ${CMAKE_SOURCE_DIR}/src)
 if (${WITH_ASAN})
 	add_definitions(-fsanitize=address -fno-omit-frame-pointer)

src/blockstore_flush.cpp

@@ -714,9 +714,15 @@ resume_1:
                             return false;
                         }
                     }
+                    if (new_trim_pos < bs->journal.used_start
+                        ? (bs->journal.dirty_start >= bs->journal.used_start || bs->journal.dirty_start < new_trim_pos)
+                        : (bs->journal.dirty_start >= bs->journal.used_start && bs->journal.dirty_start < new_trim_pos))
+                    {
+                        bs->journal.dirty_start = new_trim_pos;
+                    }
                     bs->journal.used_start = new_trim_pos;
 #ifdef BLOCKSTORE_DEBUG
-                    printf("Journal trimmed to %08lx (next_free=%08lx)\n", bs->journal.used_start, bs->journal.next_free);
+                    printf("Journal trimmed to %08lx (next_free=%08lx dirty_start=%08lx)\n", bs->journal.used_start, bs->journal.next_free, bs->journal.dirty_start);
 #endif
                     if (bs->journal.flush_journal && !flusher->flush_queue.size())
                     {

src/blockstore_flush.h

@@ -103,6 +103,7 @@ public:
     journal_flusher_t(blockstore_impl_t *bs);
     ~journal_flusher_t();
     void loop();
+    bool is_trim_wanted() { return trim_wanted; }
     bool is_active();
     void mark_trim_possible();
     void request_trim();

src/blockstore_open.cpp

@@ -85,11 +85,13 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
         immediate_commit = IMMEDIATE_SMALL;
     }
     metadata_buf_size = strtoull(config["meta_buf_size"].c_str(), NULL, 10);
-    inmemory_meta = config["inmemory_metadata"] != "false";
+    inmemory_meta = config["inmemory_metadata"] != "false" && config["inmemory_metadata"] != "0" &&
+        config["inmemory_metadata"] != "no";
     journal.sector_count = strtoull(config["journal_sector_buffer_count"].c_str(), NULL, 10);
     journal.no_same_sector_overwrites = config["journal_no_same_sector_overwrites"] == "true" ||
         config["journal_no_same_sector_overwrites"] == "1" || config["journal_no_same_sector_overwrites"] == "yes";
-    journal.inmemory = config["inmemory_journal"] != "false";
+    journal.inmemory = config["inmemory_journal"] != "false" && config["inmemory_journal"] != "0" &&
+        config["inmemory_journal"] != "no";
     // Validate
     if (journal.sector_count < 2)
     {

src/blockstore_rollback.cpp

@@ -218,7 +218,7 @@ void blockstore_impl_t::erase_dirty(blockstore_dirty_db_t::iterator dirty_start,
         auto used = --journal.used_sectors[dirty_it->second.journal_sector];
 #ifdef BLOCKSTORE_DEBUG
         printf(
-            "remove usage of journal offset %08lx by %lx:%lx v%lu (%d refs)\n", dirty_it->second.journal_sector,
+            "remove usage of journal offset %08lx by %lx:%lx v%lu (%lu refs)\n", dirty_it->second.journal_sector,
             dirty_it->first.oid.inode, dirty_it->first.oid.stripe, dirty_it->first.version, used
         );
 #endif

src/blockstore_write.cpp

@@ -661,8 +661,13 @@ void blockstore_impl_t::release_journal_sectors(blockstore_op_t *op)
         uint64_t s = PRIV(op)->min_flushed_journal_sector;
         while (1)
         {
-            if (s != (1+journal.cur_sector) && journal.sector_info[s-1].flush_count == 0)
+            if (!journal.sector_info[s-1].dirty && journal.sector_info[s-1].flush_count == 0)
             {
+                if (s == (1+journal.cur_sector))
+                {
+                    // Forcibly move to the next sector and move dirty position
+                    journal.in_sector_pos = journal.block_size;
+                }
                 // We know for sure that we won't write into this sector anymore
                 uint64_t new_ds = journal.sector_info[s-1].offset + journal.block_size;
                 if (new_ds >= journal.len)

src/cli_ls.cpp

@@ -56,14 +56,15 @@ struct image_lister_t
             {
                 continue;
             }
-            auto & pool_cfg = parent->cli->st_cli.pool_config.at(INODE_POOL(ic.second.num));
+            auto pool_it = parent->cli->st_cli.pool_config.find(INODE_POOL(ic.second.num));
+            bool good_pool = pool_it != parent->cli->st_cli.pool_config.end();
             auto item = json11::Json::object {
                 { "name", ic.second.name },
                 { "size", ic.second.size },
                 { "used_size", 0 },
                 { "readonly", ic.second.readonly },
                 { "pool_id", (uint64_t)INODE_POOL(ic.second.num) },
-                { "pool_name", pool_cfg.name },
+                { "pool_name", good_pool ? pool_it->second.name : "? (ID:"+std::to_string(INODE_POOL(ic.second.num))+")" },
                 { "inode_num", INODE_NO_POOL(ic.second.num) },
                 { "inode_id", ic.second.num },
             };
@@ -247,6 +248,8 @@ resume_1:
         if (state == 1)
             goto resume_1;
         get_list();
+        if (state == 100)
+            return;
         if (show_stats)
         {
 resume_1:
@@ -269,7 +272,7 @@ resume_1:
             { "key", "name" },
             { "title", "NAME" },
         });
-        if (!list_pool_id)
+        if (list_pool_name == "")
         {
             cols.push_back(json11::Json::object{
                 { "key", "pool_name" },
@@ -376,16 +379,18 @@ resume_1:
 
 std::string print_table(json11::Json items, json11::Json header, bool use_esc)
 {
+    int header_sizes[header.array_items().size()];
     std::vector<int> sizes;
     for (int i = 0; i < header.array_items().size(); i++)
     {
-        sizes.push_back(header[i]["title"].string_value().length());
+        header_sizes[i] = utf8_length(header[i]["title"].string_value());
+        sizes.push_back(header_sizes[i]);
     }
     for (auto & item: items.array_items())
     {
         for (int i = 0; i < header.array_items().size(); i++)
         {
-            int l = item[header[i]["key"].string_value()].as_string().length();
+            int l = utf8_length(item[header[i]["key"].string_value()].as_string());
             sizes[i] = sizes[i] < l ? l : sizes[i];
         }
     }
@@ -397,7 +402,7 @@ std::string print_table(json11::Json items, json11::Json header, bool use_esc)
             // Separator
             str += "  ";
         }
-        int pad = sizes[i]-header[i]["title"].string_value().length();
+        int pad = sizes[i]-header_sizes[i];
         if (header[i]["right"].bool_value())
         {
             // Align right
@@ -425,7 +430,7 @@ std::string print_table(json11::Json items, json11::Json header, bool use_esc)
                 // Separator
                 str += "  ";
             }
-            int pad = sizes[i] - item[header[i]["key"].string_value()].as_string().length();
+            int pad = sizes[i] - utf8_length(item[header[i]["key"].string_value()].as_string());
             if (header[i]["right"].bool_value())
             {
                 // Align right

src/cli_modify.cpp

@@ -13,7 +13,7 @@ struct image_changer_t
     std::string image_name;
     std::string new_name;
     uint64_t new_size = 0;
-    bool force_size = false;
+    bool force_size = false, inc_size = false;
     bool set_readonly = false, set_readwrite = false, force = false;
     // interval between fsyncs
     int fsync_interval = 128;
@@ -81,14 +81,14 @@ struct image_changer_t
         }
         if ((!set_readwrite || !cfg.readonly) &&
             (!set_readonly || cfg.readonly) &&
-            (!new_size && !force_size || cfg.size == new_size) &&
+            (!new_size && !force_size || cfg.size == new_size || cfg.size >= new_size && inc_size) &&
             (new_name == "" || new_name == image_name))
         {
             result = (cli_result_t){ .text = "No change" };
             state = 100;
             return;
         }
-        if (new_size != 0 || force_size)
+        if ((new_size != 0 || force_size) && (cfg.size < new_size || !inc_size))
         {
             if (cfg.size >= new_size)
             {
@@ -233,6 +233,7 @@ std::function<bool(cli_result_t &)> cli_tool_t::start_modify(json11::Json cfg)
     changer->new_name = cfg["rename"].string_value();
     changer->new_size = parse_size(cfg["resize"].as_string());
     changer->force_size = cfg["force_size"].bool_value();
+    changer->inc_size = cfg["inc_size"].bool_value();
     changer->force = cfg["force"].bool_value();
     changer->set_readonly = cfg["readonly"].bool_value();
     changer->set_readwrite = cfg["readwrite"].bool_value();

src/disk_tool_prepare.cpp

@@ -99,15 +99,16 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
     if (options["journal_size"] == "")
     {
         if (options["journal_device"] == "")
-            options["journal_size"] = "32M";
+            options["journal_size"] = is_hdd ? "128M" : "32M";
         else if (is_hdd)
             options["journal_size"] = DEFAULT_HYBRID_JOURNAL;
     }
+    bool is_hybrid = is_hdd && options["journal_device"] != "" && options["journal_device"] != options["data_device"];
     if (is_hdd)
     {
         if (options["block_size"] == "")
             options["block_size"] = "1M";
-        if (options["throttle_small_writes"] == "")
+        if (is_hybrid && options["throttle_small_writes"] == "")
             options["throttle_small_writes"] = "1";
     }
     json11::Json::object sb;
@@ -134,7 +135,7 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
             { "meta_offset", 4096 + (dsk.meta_device == dsk.journal_device ? dsk.journal_len : 0) },
             { "data_offset", 4096 + (dsk.data_device == dsk.meta_device ? dsk.meta_len : 0) +
                 (dsk.data_device == dsk.journal_device ? dsk.journal_len : 0) },
-            { "journal_no_same_sector_overwrites", true },
+            { "journal_no_same_sector_overwrites", !is_hdd || is_hybrid },
             { "journal_sector_buffer_count", 1024 },
             { "disable_data_fsync", json_is_true(options["disable_data_fsync"]) },
             { "disable_meta_fsync", json_is_true(options["disable_meta_fsync"]) },
@@ -146,7 +147,7 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
         for (int i = 0; i < sizeof(allow_additional_params)/sizeof(allow_additional_params[0]); i++)
         {
             auto it = options.find(allow_additional_params[i]);
-            if (it != options.end())
+            if (it != options.end() && it->second != "")
             {
                 sb[it->first] = it->second;
             }
@@ -620,7 +621,7 @@ int disk_tool_t::prepare(std::vector<std::string> devices)
                     }
                 }
                 // Treat all disks as SSDs if not in the hybrid mode
-                prepare_one(options, hybrid && dev.is_hdd ? 1 : 0);
+                prepare_one(options, dev.is_hdd ? 1 : 0);
                 if (hybrid)
                 {
                     options.erase("journal_device");

src/disk_tool_utils.cpp

@@ -264,6 +264,7 @@ int write_zero(int fd, uint64_t offset, uint64_t size)
 {
     uint64_t buf_len = 1024*1024;
     void *zero_buf = memalign_or_die(MEM_ALIGNMENT, buf_len);
+    memset(zero_buf, 0, buf_len);
     ssize_t r;
     while (size > 0)
     {

src/etcd_state_client.cpp

@@ -187,22 +187,30 @@ void etcd_state_client_t::add_etcd_url(std::string addr)
             check_addr = addr;
         if (pos == std::string::npos)
             addr += "/v3";
+        bool local = false;
         int i;
         for (i = 0; i < local_ips.size(); i++)
         {
             if (local_ips[i] == check_addr)
             {
-                this->etcd_local.push_back(addr);
+                local = true;
                 break;
             }
         }
-        if (i >= local_ips.size())
+        auto & to = local ? this->etcd_local : this->etcd_addresses;
-            this->etcd_addresses.push_back(addr);
+        for (i = 0; i < to.size(); i++)
+        {
+            if (to[i] == addr)
+                break;
+        }
+        if (i >= to.size())
+            to.push_back(addr);
     }
 }
 
 void etcd_state_client_t::parse_config(const json11::Json & config)
 {
+    this->etcd_local.clear();
     this->etcd_addresses.clear();
     if (config["etcd_address"].is_string())
     {
@@ -349,7 +357,7 @@ void etcd_state_client_t::start_etcd_watcher()
                         watch_id == ETCD_OSD_STATE_WATCH_ID)
                         etcd_watches_initialised++;
                     if (etcd_watches_initialised == 4 && this->log_level > 0)
-                        fprintf(stderr, "Successfully subscribed to etcd at %s\n", selected_etcd_address.c_str());
+                        fprintf(stderr, "Successfully subscribed to etcd at %s\n", cur_addr.c_str());
                 }
                 if (data["result"]["canceled"].bool_value())
                 {
@@ -360,15 +368,17 @@ void etcd_state_client_t::start_etcd_watcher()
                         // so we should restart from the beginning if we can
                         if (on_reload_hook != NULL)
                         {
-                            fprintf(stderr, "Revisions before %lu were compacted by etcd, reloading state\n",
+                            // check to not trigger on_reload_hook multiple times
-                                data["result"]["compact_revision"].uint64_value());
+                            if (etcd_watch_ws != NULL)
-                            if (etcd_watch_ws)
                             {
+                                fprintf(stderr, "Revisions before %lu 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 = 0;
+                                on_reload_hook();
                             }
-                            etcd_watch_revision = 0;
+                            return;
-                            on_reload_hook();
                         }
                         else
                         {
@@ -415,13 +425,9 @@ void etcd_state_client_t::start_etcd_watcher()
         }
         if (msg->eof)
         {
+            fprintf(stderr, "Disconnected from etcd %s\n", cur_addr.c_str());
             if (cur_addr == selected_etcd_address)
-            {
-                fprintf(stderr, "Disconnected from etcd %s\n", selected_etcd_address.c_str());
                 selected_etcd_address = "";
-            }
-            else
-                fprintf(stderr, "Disconnected from etcd\n");
             if (etcd_watch_ws)
             {
                 http_close(etcd_watch_ws);
@@ -438,6 +444,7 @@ void etcd_state_client_t::start_etcd_watcher()
             else if (etcd_watches_initialised > 0)
             {
                 // Connection was live, retry immediately
+                etcd_watches_initialised = 0;
                 start_etcd_watcher();
             }
         }

src/nfs_conn.cpp

@@ -190,7 +190,15 @@ static int nfs3_setattr_proc(void *opaque, rpc_op_t *rop)
     {
         if (handle == "roothandle" || self->parent->dir_by_hash.find(handle) != self->parent->dir_by_hash.end())
         {
-            *reply = (SETATTR3res){ .status = NFS3ERR_ISDIR };
+            if (args->new_attributes.size.set_it)
+            {
+                *reply = (SETATTR3res){ .status = NFS3ERR_ISDIR };
+            }
+            else
+            {
+                // Silently ignore mode, uid, gid, atime, mtime changes
+                *reply = (SETATTR3res){ .status = NFS3_OK };
+            }
         }
         else
         {
@@ -358,7 +366,6 @@ static int nfs3_read_proc(void *opaque, rpc_op_t *rop)
 }
 
 static void nfs_resize_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint64_t new_size, uint64_t offset, uint64_t count, void *buf);
-static void nfs_do_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint64_t offset, uint64_t count, void *buf);
 
 static int nfs3_write_proc(void *opaque, rpc_op_t *rop)
 {
@@ -392,7 +399,6 @@ static int nfs3_write_proc(void *opaque, rpc_op_t *rop)
         .resok = (WRITE3resok){
             //.file_wcc = ...,
             .count = (unsigned)count,
-            .committed = args->stable,
         },
     };
     if ((args->offset % alignment) != 0 || (count % alignment) != 0)
@@ -436,42 +442,101 @@ static int nfs3_write_proc(void *opaque, rpc_op_t *rop)
     return 1;
 }
 
-static void nfs_resize_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint64_t new_size, uint64_t offset, uint64_t count, void *buf)
+static void complete_extend_write(nfs_client_t *self, rpc_op_t *rop, inode_t inode, int res)
 {
-    // Check if we have to resize the inode before writing
+    WRITE3args *args = (WRITE3args*)rop->request;
+    WRITE3res *reply = (WRITE3res*)rop->reply;
+    if (res < 0)
+    {
+        *reply = (WRITE3res){ .status = vitastor_nfs_map_err(res) };
+        rpc_queue_reply(rop);
+        return;
+    }
+    bool imm = self->parent->cli->get_immediate_commit(inode);
+    reply->resok.committed = args->stable != UNSTABLE || imm ? FILE_SYNC : UNSTABLE;
+    *(uint64_t*)reply->resok.verf = self->parent->server_id;
+    if (args->stable != UNSTABLE && !imm)
+    {
+        // Client requested a stable write. Add an fsync
+        auto op = new cluster_op_t;
+        op->opcode = OSD_OP_SYNC;
+        op->callback = [rop](cluster_op_t *op)
+        {
+            if (op->retval != 0)
+            {
+                WRITE3res *reply = (WRITE3res*)rop->reply;
+                *reply = (WRITE3res){ .status = vitastor_nfs_map_err(-op->retval) };
+            }
+            delete op;
+            rpc_queue_reply(rop);
+        };
+        self->parent->cli->execute(op);
+    }
+    else
+    {
+        rpc_queue_reply(rop);
+    }
+}
+
+static void complete_extend_inode(nfs_client_t *self, uint64_t inode, uint64_t new_size, int err)
+{
+    auto ext_it = self->extend_writes.lower_bound((extend_size_t){ .inode = inode, .new_size = 0 });
+    while (ext_it != self->extend_writes.end() &&
+        ext_it->first.inode == inode &&
+        ext_it->first.new_size <= new_size)
+    {
+        ext_it->second.resize_res = err;
+        if (ext_it->second.write_res <= 0)
+        {
+            complete_extend_write(self, ext_it->second.rop, inode, ext_it->second.write_res < 0
+                ? ext_it->second.write_res : ext_it->second.resize_res);
+            self->extend_writes.erase(ext_it++);
+        }
+        else
+            ext_it++;
+    }
+}
+
+static void extend_inode(nfs_client_t *self, uint64_t inode, uint64_t new_size)
+{
+    // Send an extend request
+    auto & ext = self->extends[inode];
+    ext.cur_extend = new_size;
     auto inode_it = self->parent->cli->st_cli.inode_config.find(inode);
     if (inode_it != self->parent->cli->st_cli.inode_config.end() &&
         inode_it->second.size < new_size)
     {
         self->parent->cmd->loop_and_wait(self->parent->cmd->start_modify(json11::Json::object {
+            // FIXME: Resizing by ID is probably more correct
             { "image", inode_it->second.name },
             { "resize", new_size },
+            { "inc_size", true },
             { "force_size", true },
         }), [=](const cli_result_t & r)
         {
+            auto & ext = self->extends[inode];
             if (r.err)
             {
-                if (r.err == EAGAIN)
+                fprintf(stderr, "Error extending inode %lu to %lu bytes: %s\n", inode, new_size, r.text.c_str());
-                {
+            }
-                    // Multiple concurrent resize requests received, try to repeat
+            if (r.err == EAGAIN || ext.next_extend > ext.cur_extend)
-                    nfs_resize_write(self, rop, inode, new_size, offset, count, buf);
+            {
-                    return;
+                // Multiple concurrent resize requests received, try to repeat
-                }
+                extend_inode(self, inode, ext.next_extend > ext.cur_extend ? ext.next_extend : ext.cur_extend);
-                WRITE3res *reply = (WRITE3res*)rop->reply;
-                *reply = (WRITE3res){ .status = vitastor_nfs_map_err(r.err) };
-                rpc_queue_reply(rop);
                 return;
             }
-            nfs_do_write(self, rop, inode, offset, count, buf);
+            ext.cur_extend = ext.next_extend = 0;
+            complete_extend_inode(self, inode, new_size, r.err);
         });
     }
     else
     {
-        nfs_do_write(self, rop, inode, offset, count, buf);
+        complete_extend_inode(self, inode, new_size, 0);
     }
 }
 
-static void nfs_do_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint64_t offset, uint64_t count, void *buf)
+static void nfs_do_write(nfs_client_t *self, std::multimap<extend_size_t, extend_write_t>::iterator ewr_it,
+    rpc_op_t *rop, uint64_t inode, uint64_t offset, uint64_t count, void *buf)
 {
     cluster_op_t *op = new cluster_op_t;
     op->opcode = OSD_OP_WRITE;
@@ -479,48 +544,61 @@ static void nfs_do_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint
     op->offset = offset;
     op->len = count;
     op->iov.push_back(buf, count);
-    op->callback = [self, rop](cluster_op_t *op)
+    op->callback = [self, ewr_it, rop](cluster_op_t *op)
     {
-        uint64_t inode = op->inode;
+        auto inode = op->inode;
-        WRITE3args *args = (WRITE3args*)rop->request;
+        int write_res = op->retval < 0 ? op->retval : (op->retval != op->len ? -ERANGE : 0);
-        WRITE3res *reply = (WRITE3res*)rop->reply;
+        if (ewr_it == self->extend_writes.end())
-        if (op->retval != op->len)
         {
-            *reply = (WRITE3res){ .status = vitastor_nfs_map_err(-op->retval) };
+            complete_extend_write(self, rop, inode, write_res);
-            delete op;
-            rpc_queue_reply(rop);
         }
         else
         {
-            *(uint64_t*)reply->resok.verf = self->parent->server_id;
+            ewr_it->second.write_res = write_res;
-            delete op;
+            if (ewr_it->second.resize_res <= 0)
-            if (args->stable != UNSTABLE &&
-                !self->parent->cli->get_immediate_commit(inode))
             {
-                // Client requested a stable write. Add an fsync
+                complete_extend_write(self, rop, inode, write_res < 0 ? write_res : ewr_it->second.resize_res);
-                op = new cluster_op_t;
+                self->extend_writes.erase(ewr_it);
-                op->opcode = OSD_OP_SYNC;
-                op->callback = [rop](cluster_op_t *op)
-                {
-                    if (op->retval != 0)
-                    {
-                        WRITE3res *reply = (WRITE3res*)rop->reply;
-                        *reply = (WRITE3res){ .status = vitastor_nfs_map_err(-op->retval) };
-                    }
-                    delete op;
-                    rpc_queue_reply(rop);
-                };
-                self->parent->cli->execute(op);
-            }
-            else
-            {
-                rpc_queue_reply(rop);
             }
         }
     };
     self->parent->cli->execute(op);
 }
 
+static void nfs_resize_write(nfs_client_t *self, rpc_op_t *rop, uint64_t inode, uint64_t new_size, uint64_t offset, uint64_t count, void *buf)
+{
+    // Check if we have to resize the inode during write
+    auto inode_it = self->parent->cli->st_cli.inode_config.find(inode);
+    if (inode_it != self->parent->cli->st_cli.inode_config.end() &&
+        inode_it->second.size < new_size)
+    {
+        auto ewr_it = self->extend_writes.emplace((extend_size_t){
+            .inode = inode,
+            .new_size = new_size,
+        }, (extend_write_t){
+            .rop = rop,
+            .resize_res = 1,
+            .write_res = 1,
+        });
+        auto & ext = self->extends[inode];
+        if (ext.cur_extend > 0)
+        {
+            // Already resizing, just wait
+            if (ext.next_extend < new_size)
+                ext.next_extend = new_size;
+        }
+        else
+        {
+            extend_inode(self, inode, new_size);
+        }
+        nfs_do_write(self, ewr_it, rop, inode, offset, count, buf);
+    }
+    else
+    {
+        nfs_do_write(self, self->extend_writes.end(), rop, inode, offset, count, buf);
+    }
+}
+
 static int nfs3_create_proc(void *opaque, rpc_op_t *rop)
 {
     nfs_client_t *self = (nfs_client_t*)opaque;
@@ -881,6 +959,27 @@ static int nfs3_link_proc(void *opaque, rpc_op_t *rop)
     return 0;
 }
 
+static void fill_dir_entry(nfs_client_t *self, rpc_op_t *rop,
+    std::map<std::string, nfs_dir_t>::iterator dir_id_it, struct entryplus3 *entry, bool is_plus)
+{
+    if (dir_id_it == self->parent->dir_info.end())
+    {
+        return;
+    }
+    entry->fileid = dir_id_it->second.id;
+    if (is_plus)
+    {
+        entry->name_attributes = (post_op_attr){
+            .attributes_follow = 1,
+            .attributes = get_dir_attributes(self, dir_id_it->first),
+        };
+        entry->name_handle = (post_op_fh3){
+            .handle_follows = 1,
+            .handle = xdr_copy_string(rop->xdrs, "S"+base64_encode(sha256(dir_id_it->first))),
+        };
+    }
+}
+
 static void nfs3_readdir_common(void *opaque, rpc_op_t *rop, bool is_plus)
 {
     nfs_client_t *self = (nfs_client_t*)opaque;
@@ -958,17 +1057,17 @@ static void nfs3_readdir_common(void *opaque, rpc_op_t *rop, bool is_plus)
             continue;
         std::string subname = dir_id_it->first.substr(prefix.size());
         // for directories, fileid changes when the user restarts proxy
-        entries[subname].fileid = dir_id_it->second.id;
+        fill_dir_entry(self, rop, dir_id_it, &entries[subname], is_plus);
-        if (is_plus)
+    }
+    // Add . and ..
+    {
+        auto dir_id_it = self->parent->dir_info.find(dir);
+        fill_dir_entry(self, rop, dir_id_it, &entries["."], is_plus);
+        auto sl = dir.rfind("/");
+        if (sl != std::string::npos)
         {
-            entries[subname].name_attributes = (post_op_attr){
+            auto dir_id_it = self->parent->dir_info.find(dir.substr(0, sl));
-                .attributes_follow = 1,
+            fill_dir_entry(self, rop, dir_id_it, &entries[".."], is_plus);
-                .attributes = get_dir_attributes(self, dir_id_it->first),
-            };
-            entries[subname].name_handle = (post_op_fh3){
-                .handle_follows = 1,
-                .handle = xdr_copy_string(rop->xdrs, "S"+base64_encode(sha256(dir_id_it->first))),
-            };
         }
     }
     // Offset results by the continuation cookie (equal to index in the listing)
@@ -1193,10 +1292,11 @@ static int nfs3_commit_proc(void *opaque, rpc_op_t *rop)
     cluster_op_t *op = new cluster_op_t;
     // fsync. we don't know how to fsync a single inode, so just fsync everything
     op->opcode = OSD_OP_SYNC;
-    op->callback = [rop](cluster_op_t *op)
+    op->callback = [self, rop](cluster_op_t *op)
     {
         COMMIT3res *reply = (COMMIT3res*)rop->reply;
         *reply = (COMMIT3res){ .status = vitastor_nfs_map_err(op->retval) };
+        *(uint64_t*)reply->resok.verf = self->parent->server_id;
         rpc_queue_reply(rop);
     };
     self->parent->cli->execute(op);

src/nfs_proxy.h

@@ -86,6 +86,28 @@ struct rpc_free_buffer_t
     unsigned size;
 };
 
+struct extend_size_t
+{
+    inode_t inode;
+    uint64_t new_size;
+};
+
+inline bool operator < (const extend_size_t &a, const extend_size_t &b)
+{
+    return a.inode < b.inode || a.inode == b.inode && a.new_size < b.new_size;
+}
+
+struct extend_write_t
+{
+    rpc_op_t *rop;
+    int resize_res, write_res; // 1 = started, 0 = completed OK, -errno = completed with error
+};
+
+struct extend_inode_t
+{
+    uint64_t cur_extend = 0, next_extend = 0;
+};
+
 class nfs_client_t
 {
 public:
@@ -100,6 +122,8 @@ public:
     rpc_cur_buffer_t cur_buffer = { 0 };
     std::map<uint8_t*, rpc_used_buffer_t> used_buffers;
     std::vector<rpc_free_buffer_t> free_buffers;
+    std::map<inode_t, extend_inode_t> extends;
+    std::multimap<extend_size_t, extend_write_t> extend_writes;
 
     iovec read_iov;
     msghdr read_msg = { 0 };

src/osd.h

@@ -198,13 +198,14 @@ class osd_t
     void on_change_pg_history_hook(pool_id_t pool_id, pg_num_t pg_num);
     void on_change_etcd_state_hook(std::map<std::string, etcd_kv_t> & changes);
     void on_load_config_hook(json11::Json::object & changes);
+    void on_reload_config_hook(json11::Json::object & changes);
     json11::Json on_load_pgs_checks_hook();
     void on_load_pgs_hook(bool success);
     void bind_socket();
     void acquire_lease();
     json11::Json get_osd_state();
     void create_osd_state();
-    void renew_lease();
+    void renew_lease(bool reload);
     void print_stats();
     void print_slow();
     void reset_stats();

src/osd_cluster.cpp

@@ -70,6 +70,7 @@ void osd_t::init_cluster()
         st_cli.on_load_config_hook = [this](json11::Json::object & cfg) { on_load_config_hook(cfg); };
         st_cli.load_pgs_checks_hook = [this]() { return on_load_pgs_checks_hook(); };
         st_cli.on_load_pgs_hook = [this](bool success) { on_load_pgs_hook(success); };
+        st_cli.on_reload_hook = [this]() { st_cli.load_global_config(); };
         peering_state = OSD_LOADING_PGS;
         st_cli.load_global_config();
     }
@@ -395,6 +396,14 @@ void osd_t::on_load_config_hook(json11::Json::object & global_config)
     parse_config(true);
     bind_socket();
     acquire_lease();
+    st_cli.on_load_config_hook = [this](json11::Json::object & cfg) { on_reload_config_hook(cfg); };
+}
+
+void osd_t::on_reload_config_hook(json11::Json::object & global_config)
+{
+    etcd_global_config = global_config;
+    parse_config(false);
+    renew_lease(true);
 }
 
 // Acquire lease
@@ -424,7 +433,7 @@ void osd_t::acquire_lease()
     );
     tfd->set_timer(etcd_report_interval*1000, true, [this](int timer_id)
     {
-        renew_lease();
+        renew_lease(false);
     });
 }
 
@@ -499,11 +508,11 @@ void osd_t::create_osd_state()
 }
 
 // Renew lease
-void osd_t::renew_lease()
+void osd_t::renew_lease(bool reload)
 {
     st_cli.etcd_call("/lease/keepalive", json11::Json::object {
         { "ID", etcd_lease_id }
-    }, st_cli.etcd_quick_timeout, 0, 0, [this](std::string err, json11::Json data)
+    }, st_cli.etcd_quick_timeout, 0, 0, [this, reload](std::string err, json11::Json data)
     {
         if (err == "" && data["result"]["TTL"].string_value() == "")
         {
@@ -522,15 +531,20 @@ void osd_t::renew_lease()
                 force_stop(1);
             }
             // Retry
-            tfd->set_timer(st_cli.etcd_quick_timeout, false, [this](int timer_id)
+            tfd->set_timer(st_cli.etcd_quick_timeout, false, [this, reload](int timer_id)
             {
-                renew_lease();
+                renew_lease(reload);
             });
         }
         else
         {
             etcd_failed_attempts = 0;
             report_statistics();
+            // Reload PGs
+            if (reload && run_primary)
+            {
+                st_cli.load_pgs();
+            }
         }
     });
 }
@@ -560,7 +574,6 @@ void osd_t::force_stop(int exitcode)
 
 json11::Json osd_t::on_load_pgs_checks_hook()
 {
-    assert(this->pgs.size() == 0);
     json11::Json::array checks = {
         json11::Json::object {
             { "target", "LEASE" },

src/qemu_driver.c

@@ -58,6 +58,8 @@ typedef struct VitastorFdData VitastorFdData;
 typedef struct VitastorClient
 {
     void *proxy;
+    int uring_eventfd;
+
     void *watch;
     char *config_path;
     char *etcd_host;
@@ -77,6 +79,7 @@ typedef struct VitastorClient
     AioContext *ctx;
     VitastorFdData **fds;
     int fd_count, fd_alloc;
+    int bh_uring_scheduled;
 
     uint64_t last_bitmap_inode, last_bitmap_offset, last_bitmap_len;
     uint32_t last_bitmap_granularity;
@@ -106,6 +109,14 @@ typedef struct VitastorRPC
 #endif
 } VitastorRPC;
 
+#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR < 8
+typedef struct VitastorBH
+{
+    VitastorClient *cli;
+    QEMUBH *bh;
+} VitastorBH;
+#endif
+
 static void vitastor_co_init_task(BlockDriverState *bs, VitastorRPC *task);
 static void vitastor_co_generic_cb(void *opaque, long retval);
 static void vitastor_co_read_cb(void *opaque, long retval, uint64_t version);
@@ -223,6 +234,54 @@ out:
     return;
 }
 
+#if defined VITASTOR_C_API_VERSION && VITASTOR_C_API_VERSION >= 2
+static void vitastor_uring_handler(void *opaque)
+{
+    VitastorClient *client = (VitastorClient*)opaque;
+    qemu_mutex_lock(&client->mutex);
+    client->bh_uring_scheduled = 0;
+    vitastor_c_uring_handle_events(client->proxy);
+    qemu_mutex_unlock(&client->mutex);
+}
+
+#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR < 8
+static void vitastor_bh_uring_handler(void *opaque)
+{
+    VitastorBH *vbh = opaque;
+    vitastor_bh_handler(vbh->cli);
+    qemu_bh_delete(vbh->bh);
+    free(vbh);
+}
+#endif
+
+static void vitastor_schedule_uring_handler(VitastorClient *client)
+{
+    void *opaque = client;
+    if (client->uring_eventfd >= 0 && !client->bh_uring_scheduled)
+    {
+        client->bh_uring_scheduled = 1;
+#if QEMU_VERSION_MAJOR > 4 || QEMU_VERSION_MAJOR == 4 && QEMU_VERSION_MINOR >= 2
+        replay_bh_schedule_oneshot_event(client->ctx, vitastor_uring_handler, opaque);
+#elif QEMU_VERSION_MAJOR >= 3 || QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 8
+        aio_bh_schedule_oneshot(client->ctx, vitastor_uring_handler, opaque);
+#else
+        VitastorBH *vbh = (VitastorBH*)malloc(sizeof(VitastorBH));
+        vbh->cli = client;
+#if QEMU_VERSION_MAJOR >= 2
+        vbh->bh = aio_bh_new(bdrv_get_aio_context(task->bs), vitastor_bh_uring_handler, vbh);
+#else
+        vbh->bh = qemu_bh_new(vitastor_bh_uring_handler, vbh);
+#endif
+        qemu_bh_schedule(vbh->bh);
+#endif
+    }
+}
+#else
+static void vitastor_schedule_uring_handler(VitastorClient *client)
+{
+}
+#endif
+
 static void coroutine_fn vitastor_co_get_metadata(VitastorRPC *task)
 {
     BlockDriverState *bs = task->bs;
@@ -231,6 +290,7 @@ static void coroutine_fn vitastor_co_get_metadata(VitastorRPC *task)
 
     qemu_mutex_lock(&client->mutex);
     vitastor_c_watch_inode(client->proxy, client->image, vitastor_co_generic_cb, task);
+    vitastor_schedule_uring_handler(client);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task->complete)
@@ -244,6 +304,7 @@ static void vitastor_aio_fd_read(void *fddv)
     VitastorFdData *fdd = (VitastorFdData*)fddv;
     qemu_mutex_lock(&fdd->cli->mutex);
     fdd->fd_read(fdd->opaque);
+    vitastor_schedule_uring_handler(fdd->cli);
     qemu_mutex_unlock(&fdd->cli->mutex);
 }
 
@@ -252,9 +313,30 @@ static void vitastor_aio_fd_write(void *fddv)
     VitastorFdData *fdd = (VitastorFdData*)fddv;
     qemu_mutex_lock(&fdd->cli->mutex);
     fdd->fd_write(fdd->opaque);
+    vitastor_schedule_uring_handler(fdd->cli);
     qemu_mutex_unlock(&fdd->cli->mutex);
 }
 
+static void universal_aio_set_fd_handler(AioContext *ctx, int fd, IOHandler *fd_read, IOHandler *fd_write, void *opaque)
+{
+    aio_set_fd_handler(ctx, fd,
+#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 5 || QEMU_VERSION_MAJOR >= 3
+        0 /*is_external*/,
+#endif
+        fd_read,
+        fd_write,
+#if QEMU_VERSION_MAJOR == 1 && QEMU_VERSION_MINOR <= 6 || QEMU_VERSION_MAJOR < 1
+        NULL /*io_flush*/,
+#endif
+#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 9 || QEMU_VERSION_MAJOR >= 3
+        NULL /*io_poll*/,
+#endif
+#if QEMU_VERSION_MAJOR >= 7
+        NULL /*io_poll_ready*/,
+#endif
+        opaque);
+}
+
 static void vitastor_aio_set_fd_handler(void *vcli, int fd, int unused1, IOHandler *fd_read, IOHandler *fd_write, void *unused2, void *opaque)
 {
     VitastorClient *client = (VitastorClient*)vcli;
@@ -297,22 +379,9 @@ static void vitastor_aio_set_fd_handler(void *vcli, int fd, int unused1, IOHandl
         }
         client->fds[client->fd_count++] = fdd;
     }
-    aio_set_fd_handler(client->ctx, fd,
+    universal_aio_set_fd_handler(
-#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 5 || QEMU_VERSION_MAJOR >= 3
+        client->ctx, fd, fd_read ? vitastor_aio_fd_read : NULL, fd_write ? vitastor_aio_fd_write : NULL, fdd
-        0 /*is_external*/,
+    );
-#endif
-        fd_read ? vitastor_aio_fd_read : NULL,
-        fd_write ? vitastor_aio_fd_write : NULL,
-#if QEMU_VERSION_MAJOR == 1 && QEMU_VERSION_MINOR <= 6 || QEMU_VERSION_MAJOR < 1
-        NULL /*io_flush*/,
-#endif
-#if QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 9 || QEMU_VERSION_MAJOR >= 3
-        NULL /*io_poll*/,
-#endif
-#if QEMU_VERSION_MAJOR >= 7
-        NULL /*io_poll_ready*/,
-#endif
-        fdd);
 }
 
 static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, Error **errp)
@@ -333,10 +402,35 @@ static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, E
     client->rdma_gid_index = qdict_get_try_int(options, "rdma-gid-index", 0);
     client->rdma_mtu = qdict_get_try_int(options, "rdma-mtu", 0);
     client->ctx = bdrv_get_aio_context(bs);
-    client->proxy = vitastor_c_create_qemu(
+#if defined VITASTOR_C_API_VERSION && VITASTOR_C_API_VERSION >= 2
+    client->proxy = vitastor_c_create_qemu_uring(
         vitastor_aio_set_fd_handler, client, client->config_path, client->etcd_host, client->etcd_prefix,
         client->use_rdma, client->rdma_device, client->rdma_port_num, client->rdma_gid_index, client->rdma_mtu, 0
     );
+    if (!client->proxy)
+    {
+        fprintf(stderr, "vitastor: failed to create io_uring: %s - I/O will be slower\n", strerror(errno));
+        client->uring_eventfd = -1;
+#endif
+        client->proxy = vitastor_c_create_qemu(
+            vitastor_aio_set_fd_handler, client, client->config_path, client->etcd_host, client->etcd_prefix,
+            client->use_rdma, client->rdma_device, client->rdma_port_num, client->rdma_gid_index, client->rdma_mtu, 0
+        );
+#if defined VITASTOR_C_API_VERSION && VITASTOR_C_API_VERSION >= 2
+    }
+    else
+    {
+        client->uring_eventfd = vitastor_c_uring_register_eventfd(client->proxy);
+        if (client->uring_eventfd < 0)
+        {
+            fprintf(stderr, "vitastor: failed to create io_uring eventfd: %s\n", strerror(errno));
+            error_setg(errp, "failed to create io_uring eventfd");
+            vitastor_close(bs);
+            return -1;
+        }
+        universal_aio_set_fd_handler(client->ctx, client->uring_eventfd, vitastor_uring_handler, NULL, client);
+    }
+#endif
     image = client->image = g_strdup(qdict_get_try_str(options, "image"));
     client->readonly = (flags & BDRV_O_RDWR) ? 1 : 0;
     // Get image metadata (size and readonly flag) or just wait until the client is ready
@@ -568,7 +662,8 @@ static void vitastor_co_generic_cb(void *opaque, long retval)
     task->bh = aio_bh_new(bdrv_get_aio_context(task->bs), vitastor_co_generic_bh_cb, opaque);
     qemu_bh_schedule(task->bh);
 #else
-    vitastor_co_generic_bh_cb(opaque);
+    task->bh = qemu_bh_new(vitastor_co_generic_bh_cb, opaque);
+    qemu_bh_schedule(task->bh);
 #endif
 }
 
@@ -593,6 +688,7 @@ static int coroutine_fn vitastor_co_preadv(BlockDriverState *bs,
     uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;
     qemu_mutex_lock(&client->mutex);
     vitastor_c_read(client->proxy, inode, offset, bytes, iov->iov, iov->niov, vitastor_co_read_cb, &task);
+    vitastor_schedule_uring_handler(client);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)
@@ -626,6 +722,7 @@ static int coroutine_fn vitastor_co_pwritev(BlockDriverState *bs,
     uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;
     qemu_mutex_lock(&client->mutex);
     vitastor_c_write(client->proxy, inode, offset, bytes, 0, iov->iov, iov->niov, vitastor_co_generic_cb, &task);
+    vitastor_schedule_uring_handler(client);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)
@@ -643,7 +740,6 @@ static void vitastor_co_read_bitmap_cb(void *opaque, long retval, uint8_t *bitma
     VitastorRPC *task = opaque;
     VitastorClient *client = task->bs->opaque;
     task->ret = retval;
-    task->complete = 1;
     if (retval >= 0)
     {
         task->bitmap = bitmap;
@@ -655,15 +751,17 @@ static void vitastor_co_read_bitmap_cb(void *opaque, long retval, uint8_t *bitma
             client->last_bitmap = bitmap;
         }
     }
-    if (qemu_coroutine_self() != task->co)
+#if QEMU_VERSION_MAJOR > 4 || QEMU_VERSION_MAJOR == 4 && QEMU_VERSION_MINOR >= 2
-    {
+    replay_bh_schedule_oneshot_event(bdrv_get_aio_context(task->bs), vitastor_co_generic_bh_cb, opaque);
-#if QEMU_VERSION_MAJOR >= 3 || QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR > 8
+#elif QEMU_VERSION_MAJOR >= 3 || QEMU_VERSION_MAJOR == 2 && QEMU_VERSION_MINOR >= 8
-        aio_co_wake(task->co);
+    aio_bh_schedule_oneshot(bdrv_get_aio_context(task->bs), vitastor_co_generic_bh_cb, opaque);
+#elif QEMU_VERSION_MAJOR >= 2
+    task->bh = aio_bh_new(bdrv_get_aio_context(task->bs), vitastor_co_generic_bh_cb, opaque);
+    qemu_bh_schedule(task->bh);
 #else
-        qemu_coroutine_enter(task->co, NULL);
+    task->bh = qemu_bh_new(vitastor_co_generic_bh_cb, opaque);
-        qemu_aio_release(task);
+    qemu_bh_schedule(task->bh);
 #endif
-    }
 }
 
 static int coroutine_fn vitastor_co_block_status(
@@ -704,6 +802,7 @@ static int coroutine_fn vitastor_co_block_status(
         task.bitmap = client->last_bitmap = NULL;
         qemu_mutex_lock(&client->mutex);
         vitastor_c_read_bitmap(client->proxy, task.inode, task.offset, task.len, !client->skip_parents, vitastor_co_read_bitmap_cb, &task);
+        vitastor_schedule_uring_handler(client);
         qemu_mutex_unlock(&client->mutex);
         while (!task.complete)
         {
@@ -790,6 +889,7 @@ static int coroutine_fn vitastor_co_flush(BlockDriverState *bs)
 
     qemu_mutex_lock(&client->mutex);
     vitastor_c_sync(client->proxy, vitastor_co_generic_cb, &task);
+    vitastor_schedule_uring_handler(client);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)

src/ringloop.cpp

@@ -2,9 +2,12 @@
 // License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
 
 #include <stdlib.h>
+#include <unistd.h>
 
 #include <stdexcept>
 
+#include <sys/eventfd.h>
+
 #include "ringloop.h"
 
 ring_loop_t::ring_loop_t(int qd)
@@ -32,6 +35,10 @@ ring_loop_t::~ring_loop_t()
     free(free_ring_data);
     free(ring_datas);
     io_uring_queue_exit(&ring);
+    if (ring_eventfd)
+    {
+        close(ring_eventfd);
+    }
 }
 
 void ring_loop_t::register_consumer(ring_consumer_t *consumer)
@@ -59,6 +66,16 @@ void ring_loop_t::unregister_consumer(ring_consumer_t *consumer)
 
 void ring_loop_t::loop()
 {
+    if (ring_eventfd >= 0)
+    {
+        // Reset eventfd counter
+        uint64_t ctr = 0;
+        int r = read(ring_eventfd, &ctr, 8);
+        if (r < 0 && errno != EAGAIN && errno != EINTR)
+        {
+            fprintf(stderr, "Error resetting eventfd: %s\n", strerror(errno));
+        }
+    }
     struct io_uring_cqe *cqe;
     while (!io_uring_peek_cqe(&ring, &cqe))
     {
@@ -77,7 +94,7 @@ void ring_loop_t::loop()
         }
         else
         {
-            printf("Warning: empty callback in SQE\n");
+            fprintf(stderr, "Warning: empty callback in SQE\n");
             free_ring_data[free_ring_data_ptr++] = d - ring_datas;
         }
         io_uring_cqe_seen(&ring, cqe);
@@ -127,3 +144,24 @@ int ring_loop_t::sqes_left()
     }
     return left;
 }
+
+int ring_loop_t::register_eventfd()
+{
+    if (ring_eventfd >= 0)
+    {
+        return ring_eventfd;
+    }
+    ring_eventfd = eventfd(0, EFD_CLOEXEC|EFD_NONBLOCK);
+    if (ring_eventfd < 0)
+    {
+        return -errno;
+    }
+    int r = io_uring_register_eventfd(&ring, ring_eventfd);
+    if (r < 0)
+    {
+        close(ring_eventfd);
+        ring_eventfd = -1;
+        return r;
+    }
+    return ring_eventfd;
+}

src/ringloop.h

@@ -126,11 +126,13 @@ class ring_loop_t
     unsigned free_ring_data_ptr;
     bool loop_again;
     struct io_uring ring;
+    int ring_eventfd = -1;
 public:
     ring_loop_t(int qd);
     ~ring_loop_t();
     void register_consumer(ring_consumer_t *consumer);
     void unregister_consumer(ring_consumer_t *consumer);
+    int register_eventfd();
 
     inline struct io_uring_sqe* get_sqe()
     {

src/str_util.cpp

@@ -308,3 +308,19 @@ std::string str_repeat(const std::string & str, int times)
         r += str;
     return r;
 }
+
+size_t utf8_length(const std::string & s)
+{
+    size_t len = 0;
+    for (size_t i = 0; i < s.size(); i++)
+        len += (s[i] & 0xC0) != 0x80;
+    return len;
+}
+
+size_t utf8_length(const char *s)
+{
+    size_t len = 0;
+    for (; *s; s++)
+        len += (*s & 0xC0) != 0x80;
+    return len;
+}

src/str_util.h

@@ -18,3 +18,5 @@ void print_help(const char *help_text, std::string exe_name, std::string cmd, bo
 uint64_t parse_time(std::string time_str, bool *ok = NULL);
 std::string read_all_fd(int fd);
 std::string str_repeat(const std::string & str, int times);
+size_t utf8_length(const std::string & s);
+size_t utf8_length(const char *s);

src/vitastor.pc.in

@@ -6,7 +6,7 @@ includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@
 
 Name: Vitastor
 Description: Vitastor client library
-Version: 0.9.2
+Version: 0.9.6
 Libs: -L${libdir} -lvitastor_client
 Cflags: -I${includedir}
 

src/vitastor_c.cpp

@@ -5,6 +5,7 @@
 // Also acts as a C-C++ proxy for the QEMU driver (QEMU headers don't compile with g++)
 
 #include <sys/epoll.h>
+#include <sys/eventfd.h>
 
 #include "ringloop.h"
 #include "epoll_manager.h"
@@ -25,6 +26,7 @@ struct vitastor_c
     epoll_manager_t *epmgr = NULL;
     timerfd_manager_t *tfd = NULL;
     cluster_client_t *cli = NULL;
+    int uring_eventfd = -1;
 
     QEMUSetFDHandler *aio_set_fd_handler = NULL;
     void *aio_ctx = NULL;
@@ -70,14 +72,8 @@ static void vitastor_c_write_handler(void *opaque)
     data->callback(data->fd, EPOLLOUT);
 }
 
-vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+static vitastor_c *vitastor_c_create_qemu_common(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context)
-    const char *config_path, const char *etcd_host, const char *etcd_prefix,
-    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
 {
-    json11::Json cfg_json = vitastor_c_common_config(
-        config_path, etcd_host, etcd_prefix, use_rdma,
-        rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
-    );
     vitastor_c *self = new vitastor_c;
     self->aio_set_fd_handler = aio_set_fd_handler;
     self->aio_ctx = aio_context;
@@ -95,24 +91,77 @@ vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *a
             self->aio_set_fd_handler(self->aio_ctx, fd, false, NULL, NULL, NULL, NULL);
         }
     });
-    self->cli = new cluster_client_t(NULL, self->tfd, cfg_json);
     return self;
 }
 
-vitastor_c *vitastor_c_create_uring(const char *config_path, const char *etcd_host, const char *etcd_prefix,
+vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+    const char *config_path, const char *etcd_host, const char *etcd_prefix,
     int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
 {
+    json11::Json cfg_json = vitastor_c_common_config(
+        config_path, etcd_host, etcd_prefix, use_rdma,
+        rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
+    );
+    auto self = vitastor_c_create_qemu_common(aio_set_fd_handler, aio_context);
+    self->cli = new cluster_client_t(NULL, self->tfd, cfg_json);
+    return self;
+}
+
+vitastor_c *vitastor_c_create_qemu_uring(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+    const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
+{
+    ring_loop_t *ringloop = NULL;
+    try
+    {
+        ringloop = new ring_loop_t(512);
+    }
+    catch (std::exception & e)
+    {
+        return NULL;
+    }
+    json11::Json cfg_json = vitastor_c_common_config(
+        config_path, etcd_host, etcd_prefix, use_rdma,
+        rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
+    );
+    auto self = vitastor_c_create_qemu_common(aio_set_fd_handler, aio_context);
+    self->ringloop = ringloop;
+    self->cli = new cluster_client_t(self->ringloop, self->tfd, cfg_json);
+    return self;
+}
+
+vitastor_c *vitastor_c_create_uring(const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
+{
+    ring_loop_t *ringloop = NULL;
+    try
+    {
+        ringloop = new ring_loop_t(512);
+    }
+    catch (std::exception & e)
+    {
+        return NULL;
+    }
     json11::Json cfg_json = vitastor_c_common_config(
         config_path, etcd_host, etcd_prefix, use_rdma,
         rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
     );
     vitastor_c *self = new vitastor_c;
-    self->ringloop = new ring_loop_t(512);
+    self->ringloop = ringloop;
     self->epmgr = new epoll_manager_t(self->ringloop);
     self->cli = new cluster_client_t(self->ringloop, self->epmgr->tfd, cfg_json);
     return self;
 }
 
+int vitastor_c_uring_register_eventfd(vitastor_c *client)
+{
+    if (!client->ringloop)
+    {
+        return -EINVAL;
+    }
+    return client->ringloop->register_eventfd();
+}
+
 vitastor_c *vitastor_c_create_uring_json(const char **options, int options_len)
 {
     json11::Json::object cfg;
@@ -166,6 +215,11 @@ void vitastor_c_uring_wait_events(vitastor_c *client)
     client->ringloop->wait();
 }
 
+int vitastor_c_uring_has_work(vitastor_c *client)
+{
+    return client->ringloop->has_work();
+}
+
 void vitastor_c_read(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len,
     struct iovec *iov, int iovcnt, VitastorReadHandler cb, void *opaque)
 {

src/vitastor_c.h

@@ -7,7 +7,7 @@
 #define VITASTOR_QEMU_PROXY_H
 
 // C API wrapper version
-#define VITASTOR_C_API_VERSION 1
+#define VITASTOR_C_API_VERSION 2
 
 #ifndef POOL_ID_BITS
 #define POOL_ID_BITS 16
@@ -34,14 +34,19 @@ typedef void QEMUSetFDHandler(void *ctx, int fd, int is_external, IOHandler *fd_
 vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
     const char *config_path, const char *etcd_host, const char *etcd_prefix,
     int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level);
+vitastor_c *vitastor_c_create_qemu_uring(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+    const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level);
 vitastor_c *vitastor_c_create_uring(const char *config_path, const char *etcd_host, const char *etcd_prefix,
     int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level);
 vitastor_c *vitastor_c_create_uring_json(const char **options, int options_len);
 void vitastor_c_destroy(vitastor_c *client);
 int vitastor_c_is_ready(vitastor_c *client);
+int vitastor_c_uring_register_eventfd(vitastor_c *client);
 void vitastor_c_uring_wait_ready(vitastor_c *client);
 void vitastor_c_uring_handle_events(vitastor_c *client);
 void vitastor_c_uring_wait_events(vitastor_c *client);
+int vitastor_c_uring_has_work(vitastor_c *client);
 void vitastor_c_read(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len,
     struct iovec *iov, int iovcnt, VitastorReadHandler cb, void *opaque);
 void vitastor_c_write(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len, uint64_t check_version,

tests/run_3osds.sh

@@ -82,7 +82,9 @@ wait_up()
     done
 }
 
-wait_up 60
+if [[ $OSD_COUNT -gt 0 ]]; then
+    wait_up 60
+fi
 
 try_reweight()
 {