| Message ID | 20220829101246.343507-1-alexander.ivanov@virtuozzo.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | parallels: Refactor the code of images checks and fix a bug | expand |
On 29.08.2022 12:12, Alexander Ivanov wrote: > Fix image inflation when offset in BAT is out of image. > > Replace whole BAT syncing by flushing only dirty blocks. > > Move all the checks outside the main check function in > separate functions > > Use WITH_QEMU_LOCK_GUARD for simplier code. > > v7: > 1,2: Fix string lengths in the commit messages. > 3: Fix a typo in the commit message. > > v6: > 1: Move the error check inside the loop. Move file size getting > to the function beginning. Skip out-of-image offsets. > 2: A new patch - don't let high_off be more than the end of the last cluster. > 3: Set data_end without any condition. > 7: Move data_end setting to parallels_check_outside_image(). > 8: Remove s->data_end setting from parallels_check_leak(). > Fix 'i' type. > > v5: > 2: Change the way of data_end fixing. > 6,7: Move data_end check to parallels_check_leak(). > > v4: > 1: Move s->data_end fix to parallels_co_check(). Split the check > in parallels_open() and the fix in parallels_co_check() to two patches. > 2: A new patch - a part of the patch 1. > Add a fix for data_end to parallels_co_check(). > 3: Move offset convertation to parallels_set_bat_entry(). > 4: Fix 'ret' rewriting by bdrv_co_flush() results. > 7: Keep 'i' as uint32_t. > > v3: > > 1-8: Fix commit message. > > v2: > > 2: A new patch - a part of the splitted patch 2. > 3: Patch order was changed so the replacement is done in parallels_co_check. > Now we use a helper to set BAT entry and mark the block dirty. > 4: Revert the condition with s->header_unclean. > 5: Move unrelated helper parallels_set_bat_entry creation to a separate patch. > 7: Move fragmentation counting code to this function too. > 8: Fix an incorrect usage of WITH_QEMU_LOCK_GUARD. > > > Alexander Ivanov (10): > parallels: Out of image offset in BAT leads to image inflation > parallels: Fix high_off calculation in parallels_co_check() > parallels: Fix data_end after out-of-image check > parallels: create parallels_set_bat_entry_helper() to assign BAT value > parallels: Use generic infrastructure for BAT writing in > parallels_co_check() > parallels: Move check of unclean image to a separate function > parallels: Move check of cluster outside image to a separate function > parallels: Move check of leaks to a separate function > parallels: Move statistic collection to a separate function > parallels: Replace qemu_co_mutex_lock by WITH_QEMU_LOCK_GUARD > > block/parallels.c | 193 +++++++++++++++++++++++++++++++++------------- > 1 file changed, 138 insertions(+), 55 deletions(-) > queued, thanks
Fix image inflation when offset in BAT is out of image. Replace whole BAT syncing by flushing only dirty blocks. Move all the checks outside the main check function in separate functions Use WITH_QEMU_LOCK_GUARD for simplier code. v7: 1,2: Fix string lengths in the commit messages. 3: Fix a typo in the commit message. v6: 1: Move the error check inside the loop. Move file size getting to the function beginning. Skip out-of-image offsets. 2: A new patch - don't let high_off be more than the end of the last cluster. 3: Set data_end without any condition. 7: Move data_end setting to parallels_check_outside_image(). 8: Remove s->data_end setting from parallels_check_leak(). Fix 'i' type. v5: 2: Change the way of data_end fixing. 6,7: Move data_end check to parallels_check_leak(). v4: 1: Move s->data_end fix to parallels_co_check(). Split the check in parallels_open() and the fix in parallels_co_check() to two patches. 2: A new patch - a part of the patch 1. Add a fix for data_end to parallels_co_check(). 3: Move offset convertation to parallels_set_bat_entry(). 4: Fix 'ret' rewriting by bdrv_co_flush() results. 7: Keep 'i' as uint32_t. v3: 1-8: Fix commit message. v2: 2: A new patch - a part of the splitted patch 2. 3: Patch order was changed so the replacement is done in parallels_co_check. Now we use a helper to set BAT entry and mark the block dirty. 4: Revert the condition with s->header_unclean. 5: Move unrelated helper parallels_set_bat_entry creation to a separate patch. 7: Move fragmentation counting code to this function too. 8: Fix an incorrect usage of WITH_QEMU_LOCK_GUARD. Alexander Ivanov (10): parallels: Out of image offset in BAT leads to image inflation parallels: Fix high_off calculation in parallels_co_check() parallels: Fix data_end after out-of-image check parallels: create parallels_set_bat_entry_helper() to assign BAT value parallels: Use generic infrastructure for BAT writing in parallels_co_check() parallels: Move check of unclean image to a separate function parallels: Move check of cluster outside image to a separate function parallels: Move check of leaks to a separate function parallels: Move statistic collection to a separate function parallels: Replace qemu_co_mutex_lock by WITH_QEMU_LOCK_GUARD block/parallels.c | 193 +++++++++++++++++++++++++++++++++------------- 1 file changed, 138 insertions(+), 55 deletions(-)