thebestvibedev/tromcho.net
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove block_order and replace << >> by / * data_block_size

Browse Source
This commit is contained in:
Vitaliy Filippov
2025-11-22 18:10:32 +03:00
parent d75b1cb2d2
commit be1858848e
7 changed files with 29 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/blockstore/blockstore_disk.cpp
+1 -1
View File
@@ -95,7 +95,7 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
{
data_block_size = (1 << DEFAULT_DATA_BLOCK_ORDER);
}
if ((block_order = is_power_of_two(data_block_size)) >= 64 || data_block_size < MIN_DATA_BLOCK_SIZE || data_block_size >= MAX_DATA_BLOCK_SIZE)
if (is_power_of_two(data_block_size) >= 64 || data_block_size < MIN_DATA_BLOCK_SIZE || data_block_size >= MAX_DATA_BLOCK_SIZE)
{
throw std::runtime_error("Bad block size");
}
src/blockstore/blockstore_disk.h
-1
View File
@@ -46,7 +46,6 @@ struct blockstore_disk_t
uint64_t data_offset, data_device_sect, data_device_size, data_len;
uint64_t journal_offset, journal_device_sect, journal_device_size, journal_len;
uint32_t block_order = 0;
uint64_t block_count = 0;
uint32_t clean_entry_bitmap_size = 0, clean_entry_size = 0, clean_dyn_size = 0;
src/blockstore/blockstore_flush.cpp
+12 -12
View File
@@ -542,7 +542,7 @@ resume_2:
if (old_entry->oid.inode != 0 && old_entry->oid != cur.oid)
{
printf("Fatal error (metadata corruption or bug): tried to wipe metadata entry %ju (%jx:%jx v%ju) as old location of %jx:%jx\n",
old_clean_loc >> bs->dsk.block_order, old_entry->oid.inode, old_entry->oid.stripe,
old_clean_loc / bs->dsk.data_block_size, old_entry->oid.inode, old_entry->oid.stripe,
old_entry->version, cur.oid.inode, cur.oid.stripe);
exit(1);
}
@@ -645,7 +645,7 @@ void journal_flusher_co::update_metadata_entry()
has_delete
? "Fatal error (metadata corruption or bug): tried to delete metadata entry %ju (%jx:%jx v%ju) while deleting %jx:%jx v%ju\n"
: "Fatal error (metadata corruption or bug): tried to overwrite non-zero metadata entry %ju (%jx:%jx v%ju) with %jx:%jx v%ju\n",
clean_loc >> bs->dsk.block_order, new_entry->oid.inode, new_entry->oid.stripe,
clean_loc / bs->dsk.data_block_size, new_entry->oid.inode, new_entry->oid.stripe,
new_entry->version, cur.oid.inode, cur.oid.stripe, cur.version
);
exit(1);
@@ -695,7 +695,7 @@ void journal_flusher_co::update_metadata_entry()
new_entry->version = cur.version;
if (!bs->inmemory_meta)
{
auto inmem_bmp = (uint8_t*)bs->clean_bitmaps + (clean_loc >> bs->dsk.block_order)*2*bs->dsk.clean_entry_bitmap_size;
auto inmem_bmp = (uint8_t*)bs->clean_bitmaps + (clean_loc / bs->dsk.data_block_size)*2*bs->dsk.clean_entry_bitmap_size;
memcpy(inmem_bmp, new_clean_bitmap, 2*bs->dsk.clean_entry_bitmap_size);
}
if (bs->dsk.meta_format >= BLOCKSTORE_META_FORMAT_V2)
@@ -848,7 +848,7 @@ bool journal_flusher_co::clear_incomplete_csum_block_bits(int wait_base)
{
printf(
"Fatal error (metadata corruption or bug): tried to make holes in %ju (%jx:%jx v%ju) with %jx:%jx v%ju\n",
clean_loc >> bs->dsk.block_order, new_entry->oid.inode, new_entry->oid.stripe,
clean_loc / bs->dsk.data_block_size, new_entry->oid.inode, new_entry->oid.stripe,
new_entry->version, cur.oid.inode, cur.oid.stripe, cur.version
);
}
@@ -864,7 +864,7 @@ bool journal_flusher_co::clear_incomplete_csum_block_bits(int wait_base)
calc_block_checksums(new_data_csums, true);
if (!bs->inmemory_meta)
{
auto inmem_bmp = (uint8_t*)bs->clean_bitmaps + (clean_loc >> bs->dsk.block_order)*2*bs->dsk.clean_entry_bitmap_size;
auto inmem_bmp = (uint8_t*)bs->clean_bitmaps + (clean_loc / bs->dsk.data_block_size)*2*bs->dsk.clean_entry_bitmap_size;
memcpy(inmem_bmp, new_clean_bitmap, 2*bs->dsk.clean_entry_bitmap_size);
}
if (bs->dsk.meta_format >= BLOCKSTORE_META_FORMAT_V2)
@@ -1231,8 +1231,8 @@ bool journal_flusher_co::modify_meta_read(uint64_t meta_loc, flusher_meta_write_
// And yet another option is to use LSM trees for metadata, but it sophisticates everything a lot,
// so I'll avoid it as long as I can.
wr.submitted = false;
wr.sector = ((meta_loc >> bs->dsk.block_order) / (bs->dsk.meta_block_size / bs->dsk.clean_entry_size)) * bs->dsk.meta_block_size;
wr.pos = ((meta_loc >> bs->dsk.block_order) % (bs->dsk.meta_block_size / bs->dsk.clean_entry_size));
wr.sector = ((meta_loc / bs->dsk.data_block_size) / (bs->dsk.meta_block_size / bs->dsk.clean_entry_size)) * bs->dsk.meta_block_size;
wr.pos = ((meta_loc / bs->dsk.data_block_size) % (bs->dsk.meta_block_size / bs->dsk.clean_entry_size));
if (bs->inmemory_meta)
{
wr.buf = (uint8_t*)bs->metadata_buffer + wr.sector;
@@ -1292,14 +1292,14 @@ void journal_flusher_co::free_data_blocks()
#ifdef BLOCKSTORE_DEBUG
printf("%s block %ju from %jx:%jx v%ju (new location is %ju)\n",
used ? "Postpone free" : "Free",
old_clean_loc >> bs->dsk.block_order,
old_clean_loc / bs->dsk.data_block_size,
cur.oid.inode, cur.oid.stripe, cur.version,
clean_loc >> bs->dsk.block_order);
clean_loc / bs->dsk.data_block_size);
#endif
if (used)
uo_it->second.was_freed = true;
else
bs->data_alloc->set(old_clean_loc >> bs->dsk.block_order, false);
bs->data_alloc->set(old_clean_loc / bs->dsk.data_block_size, false);
}
if (has_delete)
{
@@ -1309,13 +1309,13 @@ void journal_flusher_co::free_data_blocks()
#ifdef BLOCKSTORE_DEBUG
printf("%s block %ju from %jx:%jx v%ju (delete)\n",
used ? "Postpone free" : "Free",
old_clean_loc >> bs->dsk.block_order,
old_clean_loc / bs->dsk.data_block_size,
cur.oid.inode, cur.oid.stripe, cur.version);
#endif
if (used)
uo_it->second.was_freed = true;
else
bs->data_alloc->set(old_clean_loc >> bs->dsk.block_order, false);
bs->data_alloc->set(old_clean_loc / bs->dsk.data_block_size, false);
}
}
src/blockstore/blockstore_init.cpp
+7 -7
View File
@@ -372,7 +372,7 @@ bool blockstore_init_meta::handle_meta_block(uint8_t *buf, uint64_t entries_per_
// free the previous block
// here we have to zero out the previous entry because otherwise we'll hit
// "tried to overwrite non-zero metadata entry" later
uint64_t old_clean_loc = clean_it->second.location >> bs->dsk.block_order;
uint64_t old_clean_loc = clean_it->second.location / bs->dsk.data_block_size;
if (bs->inmemory_meta)
{
uint64_t sector = (old_clean_loc / entries_per_block) * bs->dsk.meta_block_size;
@@ -390,7 +390,7 @@ bool blockstore_init_meta::handle_meta_block(uint8_t *buf, uint64_t entries_per_
}
else
{
entries_to_zero.push_back(clean_it->second.location >> bs->dsk.block_order);
entries_to_zero.push_back(clean_it->second.location / bs->dsk.data_block_size);
}
#ifdef BLOCKSTORE_DEBUG
printf("Free block %ju from %jx:%jx v%ju (new location is %ju)\n",
@@ -412,7 +412,7 @@ bool blockstore_init_meta::handle_meta_block(uint8_t *buf, uint64_t entries_per_
bs->data_alloc->set(done_cnt+i, true);
clean_db[entry->oid] = (struct clean_entry){
.version = entry->version,
.location = (done_cnt+i) << bs->dsk.block_order,
.location = (done_cnt+i) * bs->dsk.data_block_size,
};
}
else
@@ -1001,7 +1001,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
printf(
"je_big_write%s oid=%jx:%jx ver=%ju loc=%ju\n",
je->type == JE_BIG_WRITE_INSTANT ? "_instant" : "",
je->big_write.oid.inode, je->big_write.oid.stripe, je->big_write.version, je->big_write.location >> bs->dsk.block_order
je->big_write.oid.inode, je->big_write.oid.stripe, je->big_write.version, je->big_write.location / bs->dsk.data_block_size
);
#endif
auto dirty_it = bs->dirty_db.upper_bound((obj_ver_id){
@@ -1064,7 +1064,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
.journal_sector = proc_pos,
.dyn_data = dyn,
}).first;
if (bs->data_alloc->get(je->big_write.location >> bs->dsk.block_order))
if (bs->data_alloc->get(je->big_write.location / bs->dsk.data_block_size))
{
// This is probably a big_write that's already flushed and freed, but it may
// also indicate a bug. So we remember such entries and recheck them afterwards.
@@ -1077,11 +1077,11 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
#ifdef BLOCKSTORE_DEBUG
printf(
"Allocate block (journal) %ju: %jx:%jx v%ju\n",
je->big_write.location >> bs->dsk.block_order,
je->big_write.location / bs->dsk.data_block_size,
ov.oid.inode, ov.oid.stripe, ov.version
);
#endif
bs->data_alloc->set(je->big_write.location >> bs->dsk.block_order, true);
bs->data_alloc->set(je->big_write.location / bs->dsk.data_block_size, true);
}
bs->journal.used_sectors[proc_pos]++;
#ifdef BLOCKSTORE_DEBUG
src/blockstore/blockstore_read.cpp
+5 -5
View File
@@ -151,7 +151,7 @@ int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op,
uint8_t* blockstore_impl_t::get_clean_entry_bitmap(uint64_t block_loc, int offset)
{
uint8_t *clean_entry_bitmap;
uint64_t meta_loc = block_loc >> dsk.block_order;
uint64_t meta_loc = block_loc / dsk.data_block_size;
if (inmemory_meta)
{
uint64_t sector = (meta_loc / (dsk.meta_block_size / dsk.clean_entry_size)) * dsk.meta_block_size;
@@ -691,8 +691,8 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
uint8_t* blockstore_impl_t::read_clean_meta_block(blockstore_op_t *op, uint64_t clean_loc, int rv_pos)
{
auto & rv = PRIV(op)->read_vec;
auto sector = ((clean_loc >> dsk.block_order) / (dsk.meta_block_size / dsk.clean_entry_size)) * dsk.meta_block_size;
auto pos = ((clean_loc >> dsk.block_order) % (dsk.meta_block_size / dsk.clean_entry_size)) * dsk.clean_entry_size;
auto sector = ((clean_loc / dsk.data_block_size) / (dsk.meta_block_size / dsk.clean_entry_size)) * dsk.meta_block_size;
auto pos = ((clean_loc / dsk.data_block_size) % (dsk.meta_block_size / dsk.clean_entry_size)) * dsk.clean_entry_size;
uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.meta_block_size);
rv.insert(rv.begin()+rv_pos, (copy_buffer_t){
.copy_flags = COPY_BUF_META_BLOCK|COPY_BUF_CSUM_FILL,
@@ -806,7 +806,7 @@ bool blockstore_impl_t::verify_clean_padded_checksums(blockstore_op_t *op, uint6
uint32_t offset = clean_loc % dsk.data_block_size;
if (from_journal)
return verify_padded_checksums(dyn_data, dyn_data + dsk.clean_entry_bitmap_size, offset, iov, n_iov, bad_block_cb);
clean_loc = (clean_loc >> dsk.block_order) << dsk.block_order;
clean_loc = (clean_loc / dsk.data_block_size) * dsk.data_block_size;
if (!dyn_data)
{
assert(inmemory_meta);
@@ -866,7 +866,7 @@ void blockstore_impl_t::handle_read_event(ring_data_t *data, blockstore_op_t *op
{
// BIG_WRITE from journal or clean data
// Do not verify checksums if the data location is/was mutated by flushers
auto & uo = used_clean_objects.at((rv[i].disk_offset >> dsk.block_order) << dsk.block_order);
auto & uo = used_clean_objects.at((rv[i].disk_offset / dsk.data_block_size) * dsk.data_block_size);
if (!uo.was_changed)
{
verify_clean_padded_checksums(
src/blockstore/blockstore_rollback.cpp
+2 -2
View File
@@ -210,10 +210,10 @@ void blockstore_impl_t::erase_dirty(blockstore_dirty_db_t::iterator dirty_start,
dirty_it->second.location != UINT64_MAX)
{
#ifdef BLOCKSTORE_DEBUG
printf("Free block %ju from %jx:%jx v%ju\n", dirty_it->second.location >> dsk.block_order,
printf("Free block %ju from %jx:%jx v%ju\n", dirty_it->second.location / dsk.data_block_size,
dirty_it->first.oid.inode, dirty_it->first.oid.stripe, dirty_it->first.version);
#endif
data_alloc->set(dirty_it->second.location >> dsk.block_order, false);
data_alloc->set(dirty_it->second.location / dsk.data_block_size, false);
}
auto used = --journal.used_sectors.at(dirty_it->second.journal_sector);
#ifdef BLOCKSTORE_DEBUG
src/blockstore/blockstore_write.cpp
+2 -2
View File
@@ -340,7 +340,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
}
BS_SUBMIT_GET_SQE(sqe, data);
write_iodepth++;
dirty_it->second.location = loc << dsk.block_order;
dirty_it->second.location = loc * dsk.data_block_size;
dirty_it->second.state = (dirty_it->second.state & ~BS_ST_WORKFLOW_MASK) | BS_ST_SUBMITTED;
#ifdef BLOCKSTORE_DEBUG
printf(
@@ -366,7 +366,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
data->iov.iov_len = op->len + stripe_offset + stripe_end; // to check it in the callback
data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
io_uring_prep_writev(
sqe, dsk.data_fd, PRIV(op)->iov_zerofill, vcnt, dsk.data_offset + (loc << dsk.block_order) + op->offset - stripe_offset
sqe, dsk.data_fd, PRIV(op)->iov_zerofill, vcnt, dsk.data_offset + (loc * dsk.data_block_size) + op->offset - stripe_offset
);
PRIV(op)->pending_ops = 1;
if (!(dirty_it->second.state & BS_ST_INSTANT))