btrfs-progs: tune: fix the leftover csum change item and add a test case for it
cover btrfs-progs: tune: fix the leftover csum change item and add a test case for it
David's cyclic csum change tests exposed a bug even with the latest csum change code, that I'm a total idiot and forgot to delete the csum change item.
This results a bug that if we have run multiple csum changes to the same target csum type (e.g. CRC32C->SHA256->CRC32C->SHA256), the second conversion to the same type would fail due to the left over csum change items.
This series would do the fix and add a test case to cover this bug.
Qu Wenruo (2):
btrfs-progs: tune: delete the csum change item after converting the fs
btrfs-progs: tests/misc: add a test case for csum change
.../058-btrfstune-csum-change/test.sh | 26 ++++++++++++ tune/change-csum.c | 42 +++++++++++++++++-- 2 files changed, 64 insertions(+), 4 deletions(-) create mode 100755 tests/misc-tests/058-btrfstune-csum-change/test.sh
Signature · 1 line
2.40.1
PATCH 2/2 btrfs-progs: tests/misc: add a test case for csum change
The test case would:
- Populate the fs
- Run csum change for all the target csum type
- Run above csum change again
This should more or less cover the common paths, and detect any conflicting csum change item.
--- .../058-btrfstune-csum-change/test.sh | 26 +++++++++++++++++++ 1 file changed, 26 insertions(+) create mode 100755 tests/misc-tests/058-btrfstune-csum-change/test.sh
tests/misc-tests/058-btrfstune-csum-change/test.sh new file +26 -0
@@ -0,0 +1,26 @@ +#!/bin/bash +# Test btrfstune --csum option + +source "$TEST_TOP/common" || exit +source "$TEST_TOP/common.convert" || exit + +check_prereq mkfs.btrfs +check_prereq btrfstune +check_prereq btrfs + +setup_root_helper +prepare_test_dev + +run_check_mkfs_test_dev --csum crc32c +run_check_mount_test_dev +populate_fs +run_check_umount_test_dev + +for hash in "blake2" "crc32c" "xxhash" "sha256"; do + run_check $SUDO_HELPER "$TOP/btrfstune" --csum $hash "$TEST_DEV" + run_check $SUDO_HELPER "$TOP/btrfs" check --check-data-csum "$TEST_DEV" +done +for hash in "blake2" "crc32c" "xxhash" "sha256"; do + run_check $SUDO_HELPER "$TOP/btrfstune" --csum $hash "$TEST_DEV" + run_check $SUDO_HELPER "$TOP/btrfs" check --check-data-csum "$TEST_DEV" +done
Signature · 1 line
2.40.1
PATCH 1/2 btrfs-progs: tune: delete the csum change item after converting the fs
In reply to Qu Wenruo[BUG]
Doing the following csum change in a row, it would fail:
# mkfs.btrfs -f --csum crc32c $dev # btrfstune --csum sha256 $dev # btrfstune --csum crc32c $dev # btrfstune --csum sha256 $dev WARNING: Experimental build with unstable or unfinished features WARNING: Switching checksums is experimental, do not use for valuable data!
Proceed to switch checksums
ERROR: failed to insert csum change item: File exists
ERROR: failed to generate new data csums: File exists
WARNING: reserved space leaked, flag=0x4 bytes_reserved=16384 extent buffer leak: start 30572544 len 16384
extent buffer leak: start 30441472 len 16384
WARNING: dirty eb leak (aborted trans): start 30441472 len 16384
[CAUSE]
During every csum change operation, btrfstune would insert an temporaray csum change item into root tree.
But unfortunately after the conversion btrfstune doesn't properly delete the csum change item, result the following items in the root tree:
item 10 key (CSUM_CHANGE TEMPORARY_ITEM 0) itemoff 13423 itemsize 0
temporary item objectid CSUM_CHANGE offset 0 target csum type crc32c (0)
item 11 key (CSUM_CHANGE TEMPORARY_ITEM 2) itemoff 13423 itemsize 0
temporary item objectid CSUM_CHANGE offset 2 target csum type sha256 (2)
Thus at the last conversion try to go back to SHA256, we failed to insert the same item, and caused the above error.
[FIX]
After finishing the metadata csum conversion, do a proper removal of the csum item.
--- tune/change-csum.c | 42 ++++++++++++++++++++++++++++++++++++++---- 1 file changed, 38 insertions(+), 4 deletions(-)
tune/change-csum.c +38 -4
@@ -583,10 +583,13 @@ out:btrfs_release_path (&path);/* -* Finish the change by clearing the csum change flag and update the superblock -* csum type. +* Finish the change by clearing the csum change flag, update the superblock +* csum type, and delete the csum change item in the fs with new csum type. */ if (ret ==0 ) { +struct btrfs_root *tree_root = fs_info->tree_root; +struct btrfs_trans_handle *trans; +u64 super_flags =btrfs_super_flags (fs_info->super_copy);btrfs_set_super_csum_type (fs_info->super_copy, new_csum_type);
@@ -596,11 +599,42 @@ out: fs_info->csum_type = new_csum_type; fs_info->csum_size =btrfs_csum_type_size (new_csum_type); + fs_info->skip_csum_check =0 ; - ret =write_all_supers (fs_info); + trans =btrfs_start_transaction (tree_root,1 ); +if (IS_ERR (trans)) { + ret =PTR_ERR (trans); + errno = -ret; +error ("failed to start new transaction with new csum type: %m" ); +return ret; + } + key.objectid =BTRFS_CSUM_CHANGE_OBJECTID ; + key.type =BTRFS_TEMPORARY_ITEM_KEY ; + key.offset = new_csum_type; + + ret =btrfs_search_slot (trans, tree_root, &key, &path,-1 ,1 ); +if (ret >0 ) + ret = -ENOENT ;if (ret <0 ) { errno = -ret; -error ("failed to write super blocks: %m" ); +error ("failed to locate the csum change item: %m" ); +btrfs_release_path (&path); +btrfs_abort_transaction (trans, ret); +return ret; + } + ret =btrfs_del_item (trans, tree_root, &path); +if (ret <0 ) { + errno = -ret; +error ("failed to delete the csum change item: %m" ); +btrfs_release_path (&path); +btrfs_abort_transaction (trans, ret); +return ret; + } +btrfs_release_path (&path); + ret =btrfs_commit_transaction (trans, tree_root); +if (ret <0 ) { + errno = -ret; +error ("failed to finalize the csum change: %m" ); } }return ret;
Signature · 1 line
2.40.1
Quoted 8 lines · Qu Wenruo
David's cyclic csum change tests exposed a bug even with the latest csum change code, that I'm a total idiot and forgot to delete the csum change item.
This results a bug that if we have run multiple csum changes to the same target csum type (e.g.
CRC32C->SHA256->CRC32C->SHA256), the second conversion to the same type would fail due to the left over csum change items.
Quoted 5 lines
This series would do the fix and add a test case to cover this bug.
Qu Wenruo (2):
btrfs-progs: tune: delete the csum change item after converting the fs
btrfs-progs: tests/misc: add a test case for csum change
Added to devel thanks, no more errors reported.
Quoted 9 lines · David Sterba
Quoted 8 lines · Qu Wenruo
David's cyclic csum change tests exposed a bug even with the latest csum change code, that I'm a total idiot and forgot to delete the csum change item.
This results a bug that if we have run multiple csum changes to the same target csum type (e.g.
CRC32C->SHA256->CRC32C->SHA256), the second conversion to the same type would fail due to the left over csum change items.
Quoted 7 lines
Quoted 5 lines
This series would do the fix and add a test case to cover this bug.
Qu Wenruo (2):
btrfs-progs: tune: delete the csum change item after converting the fs
btrfs-progs: tests/misc: add a test case for csum changeAdded to devel thanks, no more errors reported.
Finally I can work on the resume part (which is not much compared to the main workflow).
Thanks,
Qu