vitalif
/
vitastor
5
4
Fork 9
Code Issues 1 Pull Requests 2 Actions Releases 69 Wiki Activity

Sadly we have to refcount dyn_data...

hotfix-1.0.0
Vitaliy Filippov 2023-06-10 19:17:02 +03:00
parent a8ee391e05
commit 45e07d6294
8 changed files with 89 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
src/blockstore_flush.cpp
View File

@ -642,7 +642,8 @@ void journal_flusher_co::update_metadata_entry()
// Copy latest external bitmap/attributes // Copy latest external bitmap/attributes
{ {
uint64_t dyn_size = bs->dsk.dirty_dyn_size(dirty_end->second.offset, dirty_end->second.len); uint64_t dyn_size = bs->dsk.dirty_dyn_size(dirty_end->second.offset, dirty_end->second.len);
void *dyn_ptr = dyn_size > sizeof(void*) ? dirty_end->second.dyn_data : &dirty_end->second.dyn_data; void *dyn_ptr = dyn_size > sizeof(void*)
? (uint8_t*)dirty_end->second.dyn_data+sizeof(int) : (uint8_t*)&dirty_end->second.dyn_data;
memcpy(new_clean_bitmap + bs->dsk.clean_entry_bitmap_size, dyn_ptr, bs->dsk.clean_entry_bitmap_size); memcpy(new_clean_bitmap + bs->dsk.clean_entry_bitmap_size, dyn_ptr, bs->dsk.clean_entry_bitmap_size);
} }
// Copy initial (big_write) data checksums // Copy initial (big_write) data checksums
@ -955,7 +956,8 @@ void journal_flusher_co::scan_dirty()
// FIXME Remove this > sizeof(void*) inline perversion from everywhere. // FIXME Remove this > sizeof(void*) inline perversion from everywhere.
// I think it doesn't matter but I couldn't stop myself from implementing it :) // I think it doesn't matter but I couldn't stop myself from implementing it :)
uint64_t dyn_size = bs->dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len); uint64_t dyn_size = bs->dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len);
uint8_t* dyn_from = (uint8_t*)(dyn_size > sizeof(void*) ? dirty_it->second.dyn_data : &dirty_it->second.dyn_data) + uint8_t* dyn_from = (uint8_t*)(dyn_size > sizeof(void*)
? (uint8_t*)dirty_it->second.dyn_data+sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data) +
bs->dsk.clean_entry_bitmap_size; bs->dsk.clean_entry_bitmap_size;
it->csum_buf = dyn_from + (it->offset/bs->dsk.csum_block_size - it->csum_buf = dyn_from + (it->offset/bs->dsk.csum_block_size -
dirty_it->second.offset/bs->dsk.csum_block_size) * (bs->dsk.data_csum_type & 0xFF); dirty_it->second.offset/bs->dsk.csum_block_size) * (bs->dsk.data_csum_type & 0xFF);
@ -967,7 +969,7 @@ void journal_flusher_co::scan_dirty()
{ {
bs->pad_journal_read(v, *it, dirty_it->second.offset, bs->pad_journal_read(v, *it, dirty_it->second.offset,
dirty_it->second.offset + dirty_it->second.len, dirty_it->second.location, dirty_it->second.offset + dirty_it->second.len, dirty_it->second.location,
dyn_from, offset, submit_len, blk_begin, blk_end, blk_buf); dyn_from, NULL, offset, submit_len, blk_begin, blk_end, blk_buf);
} }
} }
} }
@ -985,7 +987,7 @@ void journal_flusher_co::scan_dirty()
clean_bitmap_offset = dirty_it->second.offset; clean_bitmap_offset = dirty_it->second.offset;
clean_bitmap_len = dirty_it->second.len; clean_bitmap_len = dirty_it->second.len;
clean_init_dyn_ptr = bs->dsk.dirty_dyn_size(clean_bitmap_offset, clean_bitmap_len) > sizeof(void*) clean_init_dyn_ptr = bs->dsk.dirty_dyn_size(clean_bitmap_offset, clean_bitmap_len) > sizeof(void*)
? (uint8_t*)dirty_it->second.dyn_data : (uint8_t*)&dirty_it->second.dyn_data; ? (uint8_t*)dirty_it->second.dyn_data+sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data;
skip_copy = true; skip_copy = true;
} }
else if (IS_DELETE(dirty_it->second.state) && !skip_copy) else if (IS_DELETE(dirty_it->second.state) && !skip_copy)
@ -1028,7 +1030,7 @@ void journal_flusher_co::scan_dirty()
uint8_t *bmp_ptr = bs->get_clean_entry_bitmap(old_clean_loc, 0); uint8_t *bmp_ptr = bs->get_clean_entry_bitmap(old_clean_loc, 0);
uint64_t fulfilled = 0; uint64_t fulfilled = 0;
read_to_fill_incomplete = bs->fill_partial_checksum_blocks( read_to_fill_incomplete = bs->fill_partial_checksum_blocks(
v, fulfilled, bmp_ptr, false, NULL, v[0].offset/bs->dsk.csum_block_size * bs->dsk.csum_block_size, v, fulfilled, bmp_ptr, NULL, false, NULL, v[0].offset/bs->dsk.csum_block_size * bs->dsk.csum_block_size,
((v[v.size()-1].offset+v[v.size()-1].len-1) / bs->dsk.csum_block_size + 1) * bs->dsk.csum_block_size ((v[v.size()-1].offset+v[v.size()-1].len-1) / bs->dsk.csum_block_size + 1) * bs->dsk.csum_block_size
); );
} }

1
src/blockstore_flush.h
View File

@ -14,6 +14,7 @@ struct copy_buffer_t
uint64_t journal_sector; // only for reads: sector+1 if used and !journal.inmemory, otherwise 0 uint64_t journal_sector; // only for reads: sector+1 if used and !journal.inmemory, otherwise 0
void *buf; void *buf;
uint8_t *csum_buf; uint8_t *csum_buf;
int *dyn_data;
}; };
struct meta_sector_t struct meta_sector_t

13
src/blockstore_impl.h
View File

@ -331,14 +331,17 @@ class blockstore_impl_t
int dequeue_read(blockstore_op_t *read_op); int dequeue_read(blockstore_op_t *read_op);
void find_holes(std::vector<copy_buffer_t> & read_vec, uint32_t item_start, uint32_t item_end, void find_holes(std::vector<copy_buffer_t> & read_vec, uint32_t item_start, uint32_t item_end,
std::function<int(int, bool, uint32_t, uint32_t)> callback); std::function<int(int, bool, uint32_t, uint32_t)> callback);
int fulfill_read(blockstore_op_t *read_op, uint64_t & fulfilled, uint32_t item_start, uint32_t item_end, int fulfill_read(blockstore_op_t *read_op,
uint32_t item_state, uint64_t item_version, uint64_t item_location, uint64_t journal_sector, uint8_t *csum); uint64_t &fulfilled, uint32_t item_start, uint32_t item_end,
uint32_t item_state, uint64_t item_version, uint64_t item_location,
uint64_t journal_sector, uint8_t *csum, int *dyn_data);
bool fulfill_clean_read(blockstore_op_t *read_op, uint64_t & fulfilled, bool fulfill_clean_read(blockstore_op_t *read_op, uint64_t & fulfilled,
uint8_t *clean_entry_bitmap, uint32_t item_start, uint32_t item_end, uint64_t clean_loc, uint64_t clean_ver); uint8_t *clean_entry_bitmap, int *dyn_data,
uint32_t item_start, uint32_t item_end, uint64_t clean_loc, uint64_t clean_ver);
int fill_partial_checksum_blocks(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled, int fill_partial_checksum_blocks(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled,
uint8_t *clean_entry_bitmap, bool from_journal, uint8_t *read_buf, uint64_t read_offset, uint64_t read_end); uint8_t *clean_entry_bitmap, int *dyn_data, bool from_journal, uint8_t *read_buf, uint64_t read_offset, uint64_t read_end);
int pad_journal_read(std::vector<copy_buffer_t> & rv, copy_buffer_t & cp, int pad_journal_read(std::vector<copy_buffer_t> & rv, copy_buffer_t & cp,
uint64_t dirty_offset, uint64_t dirty_end, uint64_t dirty_loc, uint8_t *csum_ptr, uint64_t dirty_offset, uint64_t dirty_end, uint64_t dirty_loc, uint8_t *csum_ptr, int *dyn_data,
uint64_t offset, uint64_t submit_len, uint64_t & blk_begin, uint64_t & blk_end, uint8_t* & blk_buf); uint64_t offset, uint64_t submit_len, uint64_t & blk_begin, uint64_t & blk_end, uint8_t* & blk_buf);
bool read_range_fulfilled(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled, uint8_t *read_buf, bool read_range_fulfilled(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled, uint8_t *read_buf,
uint8_t *clean_entry_bitmap, uint32_t item_start, uint32_t item_end); uint8_t *clean_entry_bitmap, uint32_t item_start, uint32_t item_end);

10
src/blockstore_init.cpp
View File

@ -908,8 +908,9 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
// allocations for entry bitmaps. This can only be fixed by using // allocations for entry bitmaps. This can only be fixed by using
// a patched map with dynamic entry size, but not the btree_map, // a patched map with dynamic entry size, but not the btree_map,
// because it doesn't keep iterators valid all the time. // because it doesn't keep iterators valid all the time.
dyn = malloc_or_die(dyn_size); dyn = malloc_or_die(dyn_size+sizeof(int));
memcpy(dyn, dyn_from, dyn_size); *((int*)dyn) = 1;
memcpy((uint8_t*)dyn+sizeof(int), dyn_from, dyn_size);
} }
bs->dirty_db.emplace(ov, (dirty_entry){ bs->dirty_db.emplace(ov, (dirty_entry){
.state = (BS_ST_SMALL_WRITE | BS_ST_SYNCED), .state = (BS_ST_SMALL_WRITE | BS_ST_SYNCED),
@ -996,8 +997,9 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
// allocations for entry bitmaps. This can only be fixed by using // allocations for entry bitmaps. This can only be fixed by using
// a patched map with dynamic entry size, but not the btree_map, // a patched map with dynamic entry size, but not the btree_map,
// because it doesn't keep iterators valid all the time. // because it doesn't keep iterators valid all the time.
dyn = malloc_or_die(dyn_size); dyn = malloc_or_die(dyn_size+sizeof(int));
memcpy(dyn, dyn_from, dyn_size); *((int*)dyn) = 1;
memcpy((uint8_t*)dyn+sizeof(int), dyn_from, dyn_size);
} }
auto dirty_it = bs->dirty_db.emplace(ov, (dirty_entry){ auto dirty_it = bs->dirty_db.emplace(ov, (dirty_entry){
.state = (BS_ST_BIG_WRITE | BS_ST_SYNCED), .state = (BS_ST_BIG_WRITE | BS_ST_SYNCED),

68
src/blockstore_read.cpp
View File

@ -76,7 +76,7 @@ void blockstore_impl_t::find_holes(std::vector<copy_buffer_t> & read_vec,
int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op, int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op,
uint64_t &fulfilled, uint32_t item_start, uint32_t item_end, uint64_t &fulfilled, uint32_t item_start, uint32_t item_end,
uint32_t item_state, uint64_t item_version, uint64_t item_location, uint32_t item_state, uint64_t item_version, uint64_t item_location,
uint64_t journal_sector, uint8_t *csum) uint64_t journal_sector, uint8_t *csum, int *dyn_data)
{ {
int r = 1; int r = 1;
if (item_start < read_op->offset + read_op->len && item_end > read_op->offset) if (item_start < read_op->offset + read_op->len && item_end > read_op->offset)
@ -97,7 +97,12 @@ int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op,
.disk_offset = item_location + start - item_start, .disk_offset = item_location + start - item_start,
.journal_sector = (IS_JOURNAL(item_state) ? journal_sector : 0), .journal_sector = (IS_JOURNAL(item_state) ? journal_sector : 0),
.csum_buf = !csum ? NULL : (csum + (start - item_start) / dsk.csum_block_size * (dsk.data_csum_type & 0xFF)), .csum_buf = !csum ? NULL : (csum + (start - item_start) / dsk.csum_block_size * (dsk.data_csum_type & 0xFF)),
.dyn_data = dyn_data,
}; };
if (dyn_data)
{
(*dyn_data)++;
}
if (IS_BIG_WRITE(item_state)) if (IS_BIG_WRITE(item_state))
{ {
// If we don't track it then we may IN THEORY read another object's data: // If we don't track it then we may IN THEORY read another object's data:
@ -109,7 +114,7 @@ int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op,
if (!journal.inmemory && dsk.csum_block_size > dsk.bitmap_granularity && IS_JOURNAL(item_state) && !IS_DELETE(item_state)) if (!journal.inmemory && dsk.csum_block_size > dsk.bitmap_granularity && IS_JOURNAL(item_state) && !IS_DELETE(item_state))
{ {
int pad_state = pad_journal_read(rv, rv[pos], item_start, item_end, item_location, int pad_state = pad_journal_read(rv, rv[pos], item_start, item_end, item_location,
csum, start, end-start, blk_begin, blk_end, blk_buf); csum, dyn_data, start, end-start, blk_begin, blk_end, blk_buf);
if (pad_state == 2) if (pad_state == 2)
return 1; return 1;
else if (pad_state == 1) else if (pad_state == 1)
@ -148,7 +153,7 @@ uint8_t* blockstore_impl_t::get_clean_entry_bitmap(uint64_t block_loc, int offse
} }
int blockstore_impl_t::fill_partial_checksum_blocks(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled, int blockstore_impl_t::fill_partial_checksum_blocks(std::vector<copy_buffer_t> & rv, uint64_t & fulfilled,
uint8_t *clean_entry_bitmap, bool from_journal, uint8_t *read_buf, uint64_t read_offset, uint64_t read_end) uint8_t *clean_entry_bitmap, int *dyn_data, bool from_journal, uint8_t *read_buf, uint64_t read_offset, uint64_t read_end)
{ {
if (read_end == read_offset) if (read_end == read_offset)
return 0; return 0;
@ -185,7 +190,12 @@ int blockstore_impl_t::fill_partial_checksum_blocks(std::vector<copy_buffer_t> &
.len = (end_block-start_block)*dsk.csum_block_size, .len = (end_block-start_block)*dsk.csum_block_size,
// save clean_entry_bitmap if we're reading clean data from the journal // save clean_entry_bitmap if we're reading clean data from the journal
.csum_buf = from_journal ? clean_entry_bitmap : NULL, .csum_buf = from_journal ? clean_entry_bitmap : NULL,
.dyn_data = dyn_data,
}); });
if (dyn_data)
{
(*dyn_data)++;
}
start_block = end_block; start_block = end_block;
required++; required++;
} }
@ -325,6 +335,7 @@ bool blockstore_impl_t::read_checksum_block(blockstore_op_t *op, int rv_pos, uin
.disk_offset = clean_loc + item_start, .disk_offset = clean_loc + item_start,
.buf = (uint8_t*)buf, .buf = (uint8_t*)buf,
.csum_buf = vi->csum_buf, .csum_buf = vi->csum_buf,
.dyn_data = vi->dyn_data,
}; };
int submit_fd = (vi->copy_flags & COPY_BUF_JOURNAL ? dsk.journal_fd : dsk.data_fd); int submit_fd = (vi->copy_flags & COPY_BUF_JOURNAL ? dsk.journal_fd : dsk.data_fd);
uint64_t submit_offset = (vi->copy_flags & COPY_BUF_JOURNAL ? journal.offset : dsk.data_offset); uint64_t submit_offset = (vi->copy_flags & COPY_BUF_JOURNAL ? journal.offset : dsk.data_offset);
@ -402,7 +413,9 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *read_op)
return 2; return 2;
} }
size_t dyn_size = dsk.dirty_dyn_size(dirty.offset, dirty.len); size_t dyn_size = dsk.dirty_dyn_size(dirty.offset, dirty.len);
uint8_t *bmp_ptr = (uint8_t*)(dyn_size > sizeof(void*) ? dirty.dyn_data : &dirty.dyn_data); int *dyn_data = (int*)(dsk.csum_block_size > 0 && dyn_size > sizeof(void*) ? dirty.dyn_data : NULL);
uint8_t *bmp_ptr = (dyn_size > sizeof(void*)
? (uint8_t*)dirty.dyn_data + sizeof(int) : (uint8_t*)&dirty.dyn_data);
if (!result_version) if (!result_version)
{ {
result_version = dirty_it->first.version; result_version = dirty_it->first.version;
@ -415,7 +428,8 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *read_op)
if (!IS_JOURNAL(dirty.state)) if (!IS_JOURNAL(dirty.state))
{ {
// Read from data disk, possibly checking checksums // Read from data disk, possibly checking checksums
if (!fulfill_clean_read(read_op, fulfilled, bmp_ptr, dirty.offset, dirty.offset+dirty.len, dirty.location, dirty_it->first.version)) if (!fulfill_clean_read(read_op, fulfilled, bmp_ptr, dyn_data,
dirty.offset, dirty.offset+dirty.len, dirty.location, dirty_it->first.version))
{ {
goto undo_read; goto undo_read;
} }
@ -425,7 +439,7 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *read_op)
// Copy from memory or read from journal, possibly checking checksums // Copy from memory or read from journal, possibly checking checksums
if (!fulfill_read(read_op, fulfilled, dirty.offset, dirty.offset + dirty.len, if (!fulfill_read(read_op, fulfilled, dirty.offset, dirty.offset + dirty.len,
dirty.state, dirty_it->first.version, dirty.location, dirty.journal_sector+1, dirty.state, dirty_it->first.version, dirty.location, dirty.journal_sector+1,
journal.inmemory ? NULL : bmp_ptr+dsk.clean_entry_bitmap_size)) journal.inmemory ? NULL : bmp_ptr+dsk.clean_entry_bitmap_size, dyn_data))
{ {
goto undo_read; goto undo_read;
} }
@ -451,7 +465,8 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *read_op)
} }
if (fulfilled < read_op->len) if (fulfilled < read_op->len)
{ {
if (!fulfill_clean_read(read_op, fulfilled, NULL, 0, dsk.data_block_size, clean_it->second.location, clean_it->second.version)) if (!fulfill_clean_read(read_op, fulfilled, NULL, NULL, 0, dsk.data_block_size,
clean_it->second.location, clean_it->second.version))
{ {
goto undo_read; goto undo_read;
} }
@ -501,6 +516,11 @@ undo_read:
free(vec.buf); free(vec.buf);
vec.buf = NULL; vec.buf = NULL;
} }
if (vec.dyn_data && --(*vec.dyn_data) == 0) // refcount
{
free(vec.dyn_data);
vec.dyn_data = NULL;
}
} }
} }
rv.clear(); rv.clear();
@ -509,7 +529,7 @@ undo_read:
int blockstore_impl_t::pad_journal_read(std::vector<copy_buffer_t> & rv, copy_buffer_t & cp, int blockstore_impl_t::pad_journal_read(std::vector<copy_buffer_t> & rv, copy_buffer_t & cp,
// FIXME Passing dirty_entry& would be nicer // FIXME Passing dirty_entry& would be nicer
uint64_t dirty_offset, uint64_t dirty_end, uint64_t dirty_loc, uint8_t *csum_ptr, uint64_t dirty_offset, uint64_t dirty_end, uint64_t dirty_loc, uint8_t *csum_ptr, int *dyn_data,
uint64_t offset, uint64_t submit_len, uint64_t & blk_begin, uint64_t & blk_end, uint8_t* & blk_buf) uint64_t offset, uint64_t submit_len, uint64_t & blk_begin, uint64_t & blk_end, uint8_t* & blk_buf)
{ {
if (offset % dsk.csum_block_size || submit_len % dsk.csum_block_size) if (offset % dsk.csum_block_size || submit_len % dsk.csum_block_size)
@ -543,7 +563,12 @@ int blockstore_impl_t::pad_journal_read(std::vector<copy_buffer_t> & rv, copy_bu
.buf = blk_buf, .buf = blk_buf,
.csum_buf = (csum_ptr + (blk_begin/dsk.csum_block_size - .csum_buf = (csum_ptr + (blk_begin/dsk.csum_block_size -
dirty_offset/dsk.csum_block_size) * (dsk.data_csum_type & 0xFF)), dirty_offset/dsk.csum_block_size) * (dsk.data_csum_type & 0xFF)),
.dyn_data = dyn_data,
}); });
if (dyn_data)
{
(*dyn_data)++;
}
return 1; return 1;
} }
} }
@ -552,7 +577,7 @@ int blockstore_impl_t::pad_journal_read(std::vector<copy_buffer_t> & rv, copy_bu
} }
bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t & fulfilled, bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t & fulfilled,
uint8_t *clean_entry_bitmap, uint32_t item_start, uint32_t item_end, uint64_t clean_loc, uint64_t clean_ver) uint8_t *clean_entry_bitmap, int *dyn_data, uint32_t item_start, uint32_t item_end, uint64_t clean_loc, uint64_t clean_ver)
{ {
bool from_journal = clean_entry_bitmap != NULL; bool from_journal = clean_entry_bitmap != NULL;
if (!clean_entry_bitmap) if (!clean_entry_bitmap)
@ -564,7 +589,7 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
if (dsk.csum_block_size > dsk.bitmap_granularity) if (dsk.csum_block_size > dsk.bitmap_granularity)
{ {
auto & rv = PRIV(read_op)->read_vec; auto & rv = PRIV(read_op)->read_vec;
int req = fill_partial_checksum_blocks(rv, fulfilled, clean_entry_bitmap, from_journal, int req = fill_partial_checksum_blocks(rv, fulfilled, clean_entry_bitmap, dyn_data, from_journal,
(uint8_t*)read_op->buf, read_op->offset, read_op->offset+read_op->len); (uint8_t*)read_op->buf, read_op->offset, read_op->offset+read_op->len);
for (int i = req; i > 0; i--) for (int i = req; i > 0; i--)
{ {
@ -581,19 +606,19 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
uint8_t *csum = !dsk.csum_block_size ? 0 : (clean_entry_bitmap + dsk.clean_entry_bitmap_size + uint8_t *csum = !dsk.csum_block_size ? 0 : (clean_entry_bitmap + dsk.clean_entry_bitmap_size +
item_start/dsk.csum_block_size*(dsk.data_csum_type & 0xFF)); item_start/dsk.csum_block_size*(dsk.data_csum_type & 0xFF));
if (!fulfill_read(read_op, fulfilled, item_start, item_end, if (!fulfill_read(read_op, fulfilled, item_start, item_end,
(BS_ST_BIG_WRITE | BS_ST_STABLE), 0, clean_loc + item_start, 0, csum)) (BS_ST_BIG_WRITE | BS_ST_STABLE), 0, clean_loc + item_start, 0, csum, dyn_data))
{ {
return false; return false;
} }
if (item_start > 0 && fulfilled < read_op->len) if (item_start > 0 && fulfilled < read_op->len)
{ {
// fill with zeroes // fill with zeroes
assert(fulfill_read(read_op, fulfilled, 0, item_start, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL)); assert(fulfill_read(read_op, fulfilled, 0, item_start, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL, NULL));
} }
if (item_end < dsk.data_block_size && fulfilled < read_op->len) if (item_end < dsk.data_block_size && fulfilled < read_op->len)
{ {
// fill with zeroes // fill with zeroes
assert(fulfill_read(read_op, fulfilled, item_end, dsk.data_block_size, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL)); assert(fulfill_read(read_op, fulfilled, item_end, dsk.data_block_size, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL, NULL));
} }
} }
else else
@ -609,7 +634,7 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
{ {
// fill with zeroes // fill with zeroes
assert(fulfill_read(read_op, fulfilled, bmp_start * dsk.bitmap_granularity, assert(fulfill_read(read_op, fulfilled, bmp_start * dsk.bitmap_granularity,
bmp_end * dsk.bitmap_granularity, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL)); bmp_end * dsk.bitmap_granularity, (BS_ST_DELETE | BS_ST_STABLE), 0, 0, 0, NULL, NULL));
} }
bmp_start = bmp_end; bmp_start = bmp_end;
while (clean_entry_bitmap[bmp_end >> 3] & (1 << (bmp_end & 0x7)) && bmp_end < bmp_size) while (clean_entry_bitmap[bmp_end >> 3] & (1 << (bmp_end & 0x7)) && bmp_end < bmp_size)
@ -622,7 +647,7 @@ bool blockstore_impl_t::fulfill_clean_read(blockstore_op_t *read_op, uint64_t &
bmp_start*dsk.bitmap_granularity/dsk.csum_block_size*(dsk.data_csum_type & 0xFF)); bmp_start*dsk.bitmap_granularity/dsk.csum_block_size*(dsk.data_csum_type & 0xFF));
if (!fulfill_read(read_op, fulfilled, bmp_start * dsk.bitmap_granularity, if (!fulfill_read(read_op, fulfilled, bmp_start * dsk.bitmap_granularity,
bmp_end * dsk.bitmap_granularity, (BS_ST_BIG_WRITE | BS_ST_STABLE), 0, bmp_end * dsk.bitmap_granularity, (BS_ST_BIG_WRITE | BS_ST_STABLE), 0,
clean_loc + bmp_start * dsk.bitmap_granularity, 0, csum)) clean_loc + bmp_start * dsk.bitmap_granularity, 0, csum, dyn_data))
{ {
return false; return false;
} }
@ -800,6 +825,11 @@ void blockstore_impl_t::handle_read_event(ring_data_t *data, blockstore_op_t *op
} }
free(rv[i].buf); free(rv[i].buf);
rv[i].buf = NULL; rv[i].buf = NULL;
if (rv[i].dyn_data && --(*rv[i].dyn_data) == 0) // refcount
{
free(rv[i].dyn_data);
rv[i].dyn_data = NULL;
}
} }
} }
else else
@ -825,6 +855,11 @@ void blockstore_impl_t::handle_read_event(ring_data_t *data, blockstore_op_t *op
} }
} }
} }
if (vec.dyn_data && --(*vec.dyn_data) == 0) // refcount
{
free(vec.dyn_data);
vec.dyn_data = NULL;
}
} }
} }
} }
@ -917,7 +952,8 @@ int blockstore_impl_t::read_bitmap(object_id oid, uint64_t target_version, void
if (bitmap) if (bitmap)
{ {
size_t dyn_size = dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len); size_t dyn_size = dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len);
void *dyn_ptr = (dyn_size > sizeof(void*) ? dirty_it->second.dyn_data : &dirty_it->second.dyn_data); void *dyn_ptr = (dyn_size > sizeof(void*)
? (uint8_t*)dirty_it->second.dyn_data + sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data);
memcpy(bitmap, dyn_ptr, dsk.clean_entry_bitmap_size); memcpy(bitmap, dyn_ptr, dsk.clean_entry_bitmap_size);
} }
return 0; return 0;

3
src/blockstore_rollback.cpp
View File

@ -244,7 +244,8 @@ void blockstore_impl_t::free_dirty_dyn_data(dirty_entry & e)
size_t dyn_size = dsk.dirty_dyn_size(e.offset, e.len); size_t dyn_size = dsk.dirty_dyn_size(e.offset, e.len);
if (dyn_size > sizeof(void*) && if (dyn_size > sizeof(void*) &&
(!journal.inmemory || e.dyn_data < journal.buffer || (!journal.inmemory || e.dyn_data < journal.buffer ||
e.dyn_data >= (uint8_t*)journal.buffer + journal.len)) e.dyn_data >= (uint8_t*)journal.buffer + journal.len) &&
--*((int*)e.dyn_data) == 0) // refcount
{ {
// dyn_data contains the bitmap and checksums // dyn_data contains the bitmap and checksums
// free it if it doesn't refer to the in-memory journal // free it if it doesn't refer to the in-memory journal

3
src/blockstore_sync.cpp
View File

@ -132,7 +132,8 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
je->offset = dirty_entry.offset; je->offset = dirty_entry.offset;
je->len = dirty_entry.len; je->len = dirty_entry.len;
je->location = dirty_entry.location; je->location = dirty_entry.location;
memcpy((void*)(je+1), (dyn_size > sizeof(void*) ? dirty_entry.dyn_data : &dirty_entry.dyn_data), dyn_size); memcpy((void*)(je+1), (dyn_size > sizeof(void*)
? (uint8_t*)dirty_entry.dyn_data+sizeof(int) : (uint8_t*)&dirty_entry.dyn_data), dyn_size);
je->crc32 = je_crc32((journal_entry*)je); je->crc32 = je_crc32((journal_entry*)je);
journal.crc32_last = je->crc32; journal.crc32_last = je->crc32;
it++; it++;

17
src/blockstore_write.cpp
View File

@ -16,9 +16,12 @@ bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
size_t dyn_size = dsk.dirty_dyn_size(op->offset, op->len); size_t dyn_size = dsk.dirty_dyn_size(op->offset, op->len);
if (!is_del && dyn_size > sizeof(void*)) if (!is_del && dyn_size > sizeof(void*))
{ {
dyn = calloc_or_die(1, dyn_size); // FIXME: Working with `dyn_data` has to be refactored somehow but I first have to decide how :)
// +sizeof(int) = refcount
dyn = calloc_or_die(1, dyn_size+sizeof(int));
*((int*)dyn) = 1;
} }
uint8_t *dyn_ptr = (uint8_t*)(dyn_size > sizeof(void*) ? dyn : &dyn); uint8_t *dyn_ptr = (uint8_t*)(dyn_size > sizeof(void*) ? dyn+sizeof(int) : &dyn);
uint64_t version = 1; uint64_t version = 1;
if (dirty_db.size() > 0) if (dirty_db.size() > 0)
{ {
@ -40,7 +43,7 @@ bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
if (!is_del && !deleted) if (!is_del && !deleted)
{ {
void *dyn_from = dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len) > sizeof(void*) void *dyn_from = dsk.dirty_dyn_size(dirty_it->second.offset, dirty_it->second.len) > sizeof(void*)
? dirty_it->second.dyn_data : &dirty_it->second.dyn_data; ? (uint8_t*)dirty_it->second.dyn_data + sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data;
memcpy(dyn_ptr, dyn_from, dsk.clean_entry_bitmap_size); memcpy(dyn_ptr, dyn_from, dsk.clean_entry_bitmap_size);
} }
} }
@ -55,7 +58,7 @@ bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
if (!is_del) if (!is_del)
{ {
void *bmp_ptr = get_clean_entry_bitmap(clean_it->second.location, dsk.clean_entry_bitmap_size); void *bmp_ptr = get_clean_entry_bitmap(clean_it->second.location, dsk.clean_entry_bitmap_size);
memcpy((dyn_size > sizeof(void*) ? dyn : &dyn), bmp_ptr, dsk.clean_entry_bitmap_size); memcpy(dyn_ptr, bmp_ptr, dsk.clean_entry_bitmap_size);
} }
} }
else else
@ -461,7 +464,8 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
je->len = op->len; je->len = op->len;
je->data_offset = journal.next_free; je->data_offset = journal.next_free;
je->crc32_data = dsk.csum_block_size ? 0 : crc32c(0, op->buf, op->len); je->crc32_data = dsk.csum_block_size ? 0 : crc32c(0, op->buf, op->len);
memcpy((void*)(je+1), (dyn_size > sizeof(void*) ? dirty_it->second.dyn_data : &dirty_it->second.dyn_data), dyn_size); memcpy((void*)(je+1), (dyn_size > sizeof(void*)
? (uint8_t*)dirty_it->second.dyn_data+sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data), dyn_size);
je->crc32 = je_crc32((journal_entry*)je); je->crc32 = je_crc32((journal_entry*)je);
journal.crc32_last = je->crc32; journal.crc32_last = je->crc32;
if (immediate_commit != IMMEDIATE_NONE) if (immediate_commit != IMMEDIATE_NONE)
@ -556,7 +560,8 @@ resume_2:
je->offset = op->offset; je->offset = op->offset;
je->len = op->len; je->len = op->len;
je->location = dirty_it->second.location; je->location = dirty_it->second.location;
memcpy((void*)(je+1), (dyn_size > sizeof(void*) ? dirty_it->second.dyn_data : &dirty_it->second.dyn_data), dyn_size); memcpy((void*)(je+1), (dyn_size > sizeof(void*)
? (uint8_t*)dirty_it->second.dyn_data+sizeof(int) : (uint8_t*)&dirty_it->second.dyn_data), dyn_size);
je->crc32 = je_crc32((journal_entry*)je); je->crc32 = je_crc32((journal_entry*)je);
journal.crc32_last = je->crc32; journal.crc32_last = je->crc32;
prepare_journal_sector_write(journal.cur_sector, op); prepare_journal_sector_write(journal.cur_sector, op);