FAILED: patch "[PATCH] ext4: fix rec_len verify error" failed to apply to 4.19-stable tree
The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to stable@vger.kernel.org.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-4.19.y git checkout FETCH_HEAD git cherry-pick -x 7fda67e8c3ab6069f75888f67958a6d30454a9f6 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to 'stable@vger.kernel.org' --in-reply-to '2023092055-disband-unveiling-f6cc@gregkh' --subject-prefix 'PATCH 4.19.y' HEAD^..
Possible dependencies:
7fda67e8c3ab ("ext4: fix rec_len verify error") 46c116b920eb ("ext4: verify dir block before splitting it") f036adb39976 ("ext4: rename "dirent_csum" functions to use "dirblock"") b886ee3e778e ("ext4: Support case-insensitive file name lookups")
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 7fda67e8c3ab6069f75888f67958a6d30454a9f6 Mon Sep 17 00:00:00 2001 From: Shida Zhang zhangshida@kylinos.cn Date: Thu, 3 Aug 2023 14:09:38 +0800 Subject: [PATCH] ext4: fix rec_len verify error
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index c0f0b4e2413b..c1ceccab05f5 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -343,17 +343,17 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, struct buffer_head *bh) { struct ext4_dir_entry_tail *t; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
#ifdef PARANOID struct ext4_dir_entry *d, *top;
d = (struct ext4_dir_entry *)bh->b_data; top = (struct ext4_dir_entry *)(bh->b_data + - (EXT4_BLOCK_SIZE(inode->i_sb) - - sizeof(struct ext4_dir_entry_tail))); - while (d < top && d->rec_len) + (blocksize - sizeof(struct ext4_dir_entry_tail))); + while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize)) d = (struct ext4_dir_entry *)(((void *)d) + - le16_to_cpu(d->rec_len)); + ext4_rec_len_from_disk(d->rec_len, blocksize));
if (d != top) return NULL; @@ -364,7 +364,8 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, #endif
if (t->det_reserved_zero1 || - le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) || + (ext4_rec_len_from_disk(t->det_rec_len, blocksize) != + sizeof(struct ext4_dir_entry_tail)) || t->det_reserved_zero2 || t->det_reserved_ft != EXT4_FT_DIR_CSUM) return NULL; @@ -445,13 +446,14 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode, struct ext4_dir_entry *dp; struct dx_root_info *root; int count_offset; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb); + unsigned int rlen = ext4_rec_len_from_disk(dirent->rec_len, blocksize);
- if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb)) + if (rlen == blocksize) count_offset = 8; - else if (le16_to_cpu(dirent->rec_len) == 12) { + else if (rlen == 12) { dp = (struct ext4_dir_entry *)(((void *)dirent) + 12); - if (le16_to_cpu(dp->rec_len) != - EXT4_BLOCK_SIZE(inode->i_sb) - 12) + if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12) return NULL; root = (struct dx_root_info *)(((void *)dp + 12)); if (root->reserved_zero || @@ -1315,6 +1317,7 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, unsigned int buflen = bh->b_size; char *base = bh->b_data; struct dx_hash_info h = *hinfo; + int blocksize = EXT4_BLOCK_SIZE(dir->i_sb);
if (ext4_has_metadata_csum(dir->i_sb)) buflen -= sizeof(struct ext4_dir_entry_tail); @@ -1335,11 +1338,12 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, map_tail--; map_tail->hash = h.hash; map_tail->offs = ((char *) de - base)>>2; - map_tail->size = le16_to_cpu(de->rec_len); + map_tail->size = ext4_rec_len_from_disk(de->rec_len, + blocksize); count++; cond_resched(); } - de = ext4_next_entry(de, dir->i_sb->s_blocksize); + de = ext4_next_entry(de, blocksize); } return count; }
From: Shida Zhang zhangshida@kylinos.cn
[ Upstream commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6 ]
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Shida Zhang zhangshida@kylinos.cn --- fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index db9bba3473b5..93d392576c12 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -322,17 +322,17 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, struct ext4_dir_entry *de) { struct ext4_dir_entry_tail *t; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
#ifdef PARANOID struct ext4_dir_entry *d, *top;
d = de; top = (struct ext4_dir_entry *)(((void *)de) + - (EXT4_BLOCK_SIZE(inode->i_sb) - - sizeof(struct ext4_dir_entry_tail))); - while (d < top && d->rec_len) + (blocksize - sizeof(struct ext4_dir_entry_tail))); + while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize)) d = (struct ext4_dir_entry *)(((void *)d) + - le16_to_cpu(d->rec_len)); + ext4_rec_len_from_disk(d->rec_len, blocksize));
if (d != top) return NULL; @@ -343,7 +343,8 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, #endif
if (t->det_reserved_zero1 || - le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) || + (ext4_rec_len_from_disk(t->det_rec_len, blocksize) != + sizeof(struct ext4_dir_entry_tail)) || t->det_reserved_zero2 || t->det_reserved_ft != EXT4_FT_DIR_CSUM) return NULL; @@ -425,13 +426,14 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode, struct ext4_dir_entry *dp; struct dx_root_info *root; int count_offset; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb); + unsigned int rlen = ext4_rec_len_from_disk(dirent->rec_len, blocksize);
- if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb)) + if (rlen == blocksize) count_offset = 8; - else if (le16_to_cpu(dirent->rec_len) == 12) { + else if (rlen == 12) { dp = (struct ext4_dir_entry *)(((void *)dirent) + 12); - if (le16_to_cpu(dp->rec_len) != - EXT4_BLOCK_SIZE(inode->i_sb) - 12) + if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12) return NULL; root = (struct dx_root_info *)(((void *)dp + 12)); if (root->reserved_zero || @@ -1244,6 +1246,7 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, unsigned int buflen = bh->b_size; char *base = bh->b_data; struct dx_hash_info h = *hinfo; + int blocksize = EXT4_BLOCK_SIZE(dir->i_sb);
if (ext4_has_metadata_csum(dir->i_sb)) buflen -= sizeof(struct ext4_dir_entry_tail); @@ -1257,11 +1260,12 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, map_tail--; map_tail->hash = h.hash; map_tail->offs = ((char *) de - base)>>2; - map_tail->size = le16_to_cpu(de->rec_len); + map_tail->size = ext4_rec_len_from_disk(de->rec_len, + blocksize); count++; cond_resched(); } - de = ext4_next_entry(de, dir->i_sb->s_blocksize); + de = ext4_next_entry(de, blocksize); } return count; }
On Fri, Sep 22, 2023 at 10:54:58AM +0800, zhangshida wrote:
From: Shida Zhang zhangshida@kylinos.cn
[ Upstream commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6 ]
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Shida Zhang zhangshida@kylinos.cn
fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
What stable kernel tree(s) are you asking for this to be applied to?
thanks,
greg k-h
Greg KH gregkh@linuxfoundation.org 于2023年9月22日周五 17:27写道：
On Fri, Sep 22, 2023 at 10:54:58AM +0800, zhangshida wrote:
From: Shida Zhang zhangshida@kylinos.cn
[ Upstream commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6 ]
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Shida Zhang zhangshida@kylinos.cn
fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
What stable kernel tree(s) are you asking for this to be applied to?
thanks,
greg k-h
This one is intended for 4.14.y. And the other one from zhangshida@kylinos.cn is for 4.19.y.
Apologies for this confusion. It appears that the '--subject-prefix' option of the 'git send-email' command does not work with the local patch file.
BTW, I forgot to CC my Gmail address in another thread, so I reply here only.
Best regards, Shida
On Sat, Sep 23, 2023 at 05:41:19PM +0800, Stephen Zhang wrote:
Apologies for this confusion. It appears that the '--subject-prefix' option of the 'git send-email' command does not work with the local patch file.
Stephen,
The --subject-prefix option applies to "git format-patch" when the local patch file is generated, as opposed to "git send-email". So my general workflow is to run "rm -rf /tmp/p ; git format-patch -o /tmp/p ..." and then examine the files in /tmp/p, and then if they look good, run "git send-email /tmp/p/*".
I suspect it will be easier for Greg if you were to simply regenerate the patches with the proper subject prefix, and then resend them, since he has automation tools that can handle parsing the subject line, which scripts can do much more easily than to disentangling the "In-Reply-To" header to identify e-mail chains, and then parsing human/natural language to figure out which git tree the patches should be applied to. :-)
Cheers,
- Ted
On Sat, Sep 23, 2023 at 10:47:29AM -0400, Theodore Ts'o wrote:
On Sat, Sep 23, 2023 at 05:41:19PM +0800, Stephen Zhang wrote:
Apologies for this confusion. It appears that the '--subject-prefix' option of the 'git send-email' command does not work with the local patch file.
Stephen,
The --subject-prefix option applies to "git format-patch" when the local patch file is generated, as opposed to "git send-email". So my general workflow is to run "rm -rf /tmp/p ; git format-patch -o /tmp/p ..." and then examine the files in /tmp/p, and then if they look good, run "git send-email /tmp/p/*".
I suspect it will be easier for Greg if you were to simply regenerate the patches with the proper subject prefix, and then resend them, since he has automation tools that can handle parsing the subject line, which scripts can do much more easily than to disentangling the "In-Reply-To" header to identify e-mail chains, and then parsing human/natural language to figure out which git tree the patches should be applied to. :-)
Just a hint, I can usually NOT see the In-Reply-To headers at all, as the email it was in response to is long gone from my system.
Which is why, the description of how to fix up a patch for the stable tree that has failed to apply there, tells you how to do this in a way that will show the proper version number.
If only anyone would actually read the emails we sent with the helpful text, here's one example for a DRM patch that failed to apply:
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.1.y git checkout FETCH_HEAD git cherry-pick -x ec5fa9fcdeca69edf7dab5ca3b2e0ceb1c08fe9a # <resolve conflicts, build, test, etc.> git commit -s git send-email --to 'stable@vger.kernel.org' --in-reply-to '2023092029-banter-truth-cf72@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^..
thanks,
greg k-h
From: Shida Zhang zhangshida@kylinos.cn
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu (cherry picked from commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6) Signed-off-by: Shida Zhang zhangshida@kylinos.cn --- fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index db9bba3473b5..93d392576c12 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -322,17 +322,17 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, struct ext4_dir_entry *de) { struct ext4_dir_entry_tail *t; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
#ifdef PARANOID struct ext4_dir_entry *d, *top;
d = de; top = (struct ext4_dir_entry *)(((void *)de) + - (EXT4_BLOCK_SIZE(inode->i_sb) - - sizeof(struct ext4_dir_entry_tail))); - while (d < top && d->rec_len) + (blocksize - sizeof(struct ext4_dir_entry_tail))); + while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize)) d = (struct ext4_dir_entry *)(((void *)d) + - le16_to_cpu(d->rec_len)); + ext4_rec_len_from_disk(d->rec_len, blocksize));
if (d != top) return NULL; @@ -343,7 +343,8 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, #endif
if (t->det_reserved_zero1 || - le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) || + (ext4_rec_len_from_disk(t->det_rec_len, blocksize) != + sizeof(struct ext4_dir_entry_tail)) || t->det_reserved_zero2 || t->det_reserved_ft != EXT4_FT_DIR_CSUM) return NULL; @@ -425,13 +426,14 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode, struct ext4_dir_entry *dp; struct dx_root_info *root; int count_offset; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb); + unsigned int rlen = ext4_rec_len_from_disk(dirent->rec_len, blocksize);
- if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb)) + if (rlen == blocksize) count_offset = 8; - else if (le16_to_cpu(dirent->rec_len) == 12) { + else if (rlen == 12) { dp = (struct ext4_dir_entry *)(((void *)dirent) + 12); - if (le16_to_cpu(dp->rec_len) != - EXT4_BLOCK_SIZE(inode->i_sb) - 12) + if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12) return NULL; root = (struct dx_root_info *)(((void *)dp + 12)); if (root->reserved_zero || @@ -1244,6 +1246,7 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, unsigned int buflen = bh->b_size; char *base = bh->b_data; struct dx_hash_info h = *hinfo; + int blocksize = EXT4_BLOCK_SIZE(dir->i_sb);
if (ext4_has_metadata_csum(dir->i_sb)) buflen -= sizeof(struct ext4_dir_entry_tail); @@ -1257,11 +1260,12 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, map_tail--; map_tail->hash = h.hash; map_tail->offs = ((char *) de - base)>>2; - map_tail->size = le16_to_cpu(de->rec_len); + map_tail->size = ext4_rec_len_from_disk(de->rec_len, + blocksize); count++; cond_resched(); } - de = ext4_next_entry(de, dir->i_sb->s_blocksize); + de = ext4_next_entry(de, blocksize); } return count; }
zhangshida starzhangzsd@gmail.com 于2023年9月24日周日 10:27写道：
From: Shida Zhang zhangshida@kylinos.cn
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu (cherry picked from commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6)
Oh no, I forgot to change that line to [Upstream commit ...] T_T Will resend.
Best regards, Shida.
Signed-off-by: Shida Zhang zhangshida@kylinos.cn
fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index db9bba3473b5..93d392576c12 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -322,17 +322,17 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, struct ext4_dir_entry *de) { struct ext4_dir_entry_tail *t;
int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);#ifdef PARANOID struct ext4_dir_entry *d, *top;
d = de; top = (struct ext4_dir_entry *)(((void *)de) +
(EXT4_BLOCK_SIZE(inode->i_sb) -sizeof(struct ext4_dir_entry_tail)));while (d < top && d->rec_len)
(blocksize - sizeof(struct ext4_dir_entry_tail)));while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize)) d = (struct ext4_dir_entry *)(((void *)d) +
le16_to_cpu(d->rec_len));
ext4_rec_len_from_disk(d->rec_len, blocksize)); if (d != top) return NULL;@@ -343,7 +343,8 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, #endif
if (t->det_reserved_zero1 ||
le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) ||
(ext4_rec_len_from_disk(t->det_rec_len, blocksize) !=sizeof(struct ext4_dir_entry_tail)) || t->det_reserved_zero2 || t->det_reserved_ft != EXT4_FT_DIR_CSUM) return NULL;@@ -425,13 +426,14 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode, struct ext4_dir_entry *dp; struct dx_root_info *root; int count_offset;
int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);unsigned int rlen = ext4_rec_len_from_disk(dirent->rec_len, blocksize);
if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb))
if (rlen == blocksize) count_offset = 8;
else if (le16_to_cpu(dirent->rec_len) == 12) {
else if (rlen == 12) { dp = (struct ext4_dir_entry *)(((void *)dirent) + 12);
if (le16_to_cpu(dp->rec_len) !=EXT4_BLOCK_SIZE(inode->i_sb) - 12)
if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12) return NULL; root = (struct dx_root_info *)(((void *)dp + 12)); if (root->reserved_zero ||@@ -1244,6 +1246,7 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, unsigned int buflen = bh->b_size; char *base = bh->b_data; struct dx_hash_info h = *hinfo;
int blocksize = EXT4_BLOCK_SIZE(dir->i_sb); if (ext4_has_metadata_csum(dir->i_sb)) buflen -= sizeof(struct ext4_dir_entry_tail);@@ -1257,11 +1260,12 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, map_tail--; map_tail->hash = h.hash; map_tail->offs = ((char *) de - base)>>2;
map_tail->size = le16_to_cpu(de->rec_len);
map_tail->size = ext4_rec_len_from_disk(de->rec_len,blocksize); count++; cond_resched(); }
de = ext4_next_entry(de, dir->i_sb->s_blocksize);
de = ext4_next_entry(de, blocksize); } return count;}
2.27.0
From: Shida Zhang zhangshida@kylinos.cn
[ Upstream commit 7fda67e8c3ab6069f75888f67958a6d30454a9f6 ]
With the configuration PAGE_SIZE 64k and filesystem blocksize 64k, a problem occurred when more than 13 million files were directly created under a directory:
EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt? Run e2fsck -D. EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
When enough files are created, the fake_dirent->reclen will be 0xffff. it doesn't equal to the blocksize 65536, i.e. 0x10000.
But it is not the same condition when blocksize equals to 4k. when enough files are created, the fake_dirent->reclen will be 0x1000. it equals to the blocksize 4k, i.e. 0x1000.
The problem seems to be related to the limitation of the 16-bit field when the blocksize is set to 64k. To address this, helpers like ext4_rec_len_{from,to}_disk has already been introduced to complete the conversion between the encoded and the plain form of rec_len.
So fix this one by using the helper, and all the other in this file too.
Cc: stable@kernel.org Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes") Suggested-by: Andreas Dilger adilger@dilger.ca Suggested-by: Darrick J. Wong djwong@kernel.org Signed-off-by: Shida Zhang zhangshida@kylinos.cn Reviewed-by: Andreas Dilger adilger@dilger.ca Reviewed-by: Darrick J. Wong djwong@kernel.org Link: https://lore.kernel.org/r/20230803060938.1929759-1-zhangshida@kylinos.cn Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Shida Zhang zhangshida@kylinos.cn --- fs/ext4/namei.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index db9bba3473b5..93d392576c12 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -322,17 +322,17 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, struct ext4_dir_entry *de) { struct ext4_dir_entry_tail *t; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
#ifdef PARANOID struct ext4_dir_entry *d, *top;
d = de; top = (struct ext4_dir_entry *)(((void *)de) + - (EXT4_BLOCK_SIZE(inode->i_sb) - - sizeof(struct ext4_dir_entry_tail))); - while (d < top && d->rec_len) + (blocksize - sizeof(struct ext4_dir_entry_tail))); + while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize)) d = (struct ext4_dir_entry *)(((void *)d) + - le16_to_cpu(d->rec_len)); + ext4_rec_len_from_disk(d->rec_len, blocksize));
if (d != top) return NULL; @@ -343,7 +343,8 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode, #endif
if (t->det_reserved_zero1 || - le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) || + (ext4_rec_len_from_disk(t->det_rec_len, blocksize) != + sizeof(struct ext4_dir_entry_tail)) || t->det_reserved_zero2 || t->det_reserved_ft != EXT4_FT_DIR_CSUM) return NULL; @@ -425,13 +426,14 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode, struct ext4_dir_entry *dp; struct dx_root_info *root; int count_offset; + int blocksize = EXT4_BLOCK_SIZE(inode->i_sb); + unsigned int rlen = ext4_rec_len_from_disk(dirent->rec_len, blocksize);
- if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb)) + if (rlen == blocksize) count_offset = 8; - else if (le16_to_cpu(dirent->rec_len) == 12) { + else if (rlen == 12) { dp = (struct ext4_dir_entry *)(((void *)dirent) + 12); - if (le16_to_cpu(dp->rec_len) != - EXT4_BLOCK_SIZE(inode->i_sb) - 12) + if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12) return NULL; root = (struct dx_root_info *)(((void *)dp + 12)); if (root->reserved_zero || @@ -1244,6 +1246,7 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, unsigned int buflen = bh->b_size; char *base = bh->b_data; struct dx_hash_info h = *hinfo; + int blocksize = EXT4_BLOCK_SIZE(dir->i_sb);
if (ext4_has_metadata_csum(dir->i_sb)) buflen -= sizeof(struct ext4_dir_entry_tail); @@ -1257,11 +1260,12 @@ static int dx_make_map(struct inode *dir, struct buffer_head *bh, map_tail--; map_tail->hash = h.hash; map_tail->offs = ((char *) de - base)>>2; - map_tail->size = le16_to_cpu(de->rec_len); + map_tail->size = ext4_rec_len_from_disk(de->rec_len, + blocksize); count++; cond_resched(); } - de = ext4_next_entry(de, dir->i_sb->s_blocksize); + de = ext4_next_entry(de, blocksize); } return count; }
linux-stable-mirror@lists.linaro.org
-
Greg KH -
gregkh＠linuxfoundation.org
-
Stephen Zhang -
Theodore Ts'o -
zhangshida