This patch[1] is needed on linux-5.15.y because the panic[2] is also found on linux-5.15.y when debugging bug[3]. Back ported patch[4] is confirmed to fix the bug on linux-5.15.y in the latest test of page[3]. Maybe back port on more branches is needed per patch comments.
[1] 2d544ec923dbe5 ("ext4: remove deprecated noacl/nouser_xattr options")
[2] https://syzkaller.appspot.com/text?tag=CrashLog&x=171b1077880000
[3] https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0...
[4] https://syzkaller.appspot.com/text?tag=Patch&x=1766eb13880000
Regards, Jun
On Fri, Dec 16, 2022 at 11:41:16AM +0800, Jun Nie wrote:
This patch[1] is needed on linux-5.15.y because the panic[2] is also found on linux-5.15.y when debugging bug[3]. Back ported patch[4] is confirmed to fix the bug on linux-5.15.y in the latest test of page[3]. Maybe back port on more branches is needed per patch comments.
This is not a proper fix for the syzkaller report being reported here:
https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0...
It's true that the reproducer will no longer trigger, but that's just because the reproducer is just exiting early because it is passing in a mount option which is no longer being accepted. In fact, that mount option is completely unneeded and it's a failing of syzkaller that it doesn't adequately minimize the reproducer by trying to remove various random mount options that are not actually needed. For example, running the reproducer will trigger warnings like this:
EXT4-fs: Ignoring removed nobh option
If we modify the kernel to simply ignore nouser_xattr, then the reproducer will still trigger. So this is not the right patch to backport.
It's important that people who are trying to fix syzkaller bugs understand what is fundamentally going on, instead of using blunt force patches that simple paper over the issue. Please remember that syzkaller is supposed to help us improve the kernel, and it's not just about trying to reduce the count of open syzkaller reports for its own sake. (This is really much more of a quality of implementation issue, since this is not something that would really ever trigger in real life, nor is it really a security issue --- despite some people thinking that all syzkaller reports are actually security issues, and we must run around like chickens with their heads cut off and until they are all fixed.)
The real root cause of the problem is that the file system is getting mounted with these mount options:
nouser_xattr,acl,debug_want_extra_isize=0x0000000000000080,lazytime,nobh,quota
Of which nouser_attr, acl, nobh, and quota are completely pointless. It's also **super** unfortunate that the reproducer isn't written in C, but this horrible psuedo-assimply language:
memcpy( (void*)0x20000000, "\x6e\x6f\x75\x73\x65\x72\x5f\x78\x61\x74\x74\x72\x2c\x61\x63\x6c\x2c\x64" "\x65\x62\x75\x67\x5f\x77\x61\x6e\x74\x5f\x65\x78\x74\x72\x61\x5f\x69\x73" "\x69\x7a\x65\x3d\x30\x78\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30" "\x30\x30\x38\x30\x2c\x6c\x61\x7a\x79\x74\x69\x6d\x65\x2c\x6e\x6f\x62\x68" "\x2c\x71\x75\x6f\x74\x61\x2c\x00\x3d\x93\x09\x61\x36\x5d\x73\x58\x9c", 89);
... syz_mount_image(0x20000440, 0x20000480, 0x1e, 0x20000000, 2, 0x427, ^^^^^^^^^^ 0x200004c0);
(And again, this is stuff that I've complained to the syzkaller team for years and years and years as being fundamentally developer hostile and disrespects the time of upstream maintainers. ARGH!!!!)
Anyway..... So now let's look at the stack trace:
ext4_xattr_block_set+0x8f8/0x3820 fs/ext4/xattr.c:1971 ext4_xattr_move_to_block fs/ext4/xattr.c:2603 [inline] ext4_xattr_make_inode_space fs/ext4/xattr.c:2672 [inline] ext4_expand_extra_isize_ea+0x1591/0x1f30 fs/ext4/xattr.c:2764 __ext4_expand_extra_isize+0x29e/0x3d0 fs/ext4/inode.c:5826 ext4_try_to_expand_extra_isize fs/ext4/inode.c:5869 [inline] __ext4_mark_inode_dirty+0x4bf/0x7a0 fs/ext4/inode.c:5947 ext4_dirty_inode+0xbc/0x100 fs/ext4/inode.c:5979 __mark_inode_dirty+0x1f9/0x9d0 fs/fs-writeback.c:2431 mark_inode_dirty_sync include/linux/fs.h:2429 [inline] iput+0x155/0x7d0 fs/inode.c:1686 dentry_unlink_inode+0x349/0x430 fs/dcache.c:376 __dentry_kill+0x3e2/0x5d0 fs/dcache.c:582 shrink_dentry_list+0x379/0x4d0 fs/dcache.c:1176 shrink_dcache_parent+0xcd/0x350 do_one_tree fs/dcache.c:1657 [inline] shrink_dcache_for_umount+0x7c/0x1a0 fs/dcache.c:1674 generic_shutdown_super+0x69/0x2d0 fs/super.c:447 kill_block_super+0x80/0xe0 fs/super.c:1395
Because lazytime is enabled, after running the reproducer under strace, what happens is that inode #12 gets touched so its access time is modified, but because lazytime is enabled, we don't actually update the on-disk until we actually unmount the superblock. That's why generic_shutdown_super() is in the stack trace.
At that point, when we shrink the dentry cache, when we eject the inode from memory, iput() needs to update the on-disk inode with the updated atime. So far, so good. But then we call ext4_dirty_inode(), and then that interacts with the "debug_want_extra_isize-=128" mount option. So at this point, we try to expand inode's extra isize space, and in order to do that we have to move some extended attributes.
Unfortunately, how ext4 currently does this is a bit stupid, and it reads the contents of the ea_inode into memory, deletes the ea_inode and then creates a new ea_inode. That works, but it's horribly inefficient, and **that's*** what we should actually fix.
Unfortunately, because we try to create a new ea_inode, when ext4_xattr_block_set() calls the static function (which gets inlined) ext4_xattr_inode_create(), and at that point, the call to ext4_new_inode trips over the fact that the file system is being unmoutned, and sb->s_root has already been set to NULL.
So this is what actually goes *boom*:
ea_inode = ext4_new_inode(handle, inode->i_sb->s_root->d_inode, ^^^^^^^^^^^^^^^^^^^ NULL ptr, oops! S_IFREG | 0600, NULL, inode->i_ino + 1, owner, EXT4_EA_INODE_FL);
We can prove this is the issue by using the following debugging patch, which prevents the reproducer from triggering after prining the "fs being unmouinted" message:
--- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -2697,6 +2697,13 @@ int ext4_expand_extra_isize_ea(struct inode *inode, int new_extra_isize, int s_min_extra_isize = le16_to_cpu(sbi->s_es->s_min_extra_isize); int isize_diff; /* How much do we need to grow i_extra_isize */
+ pr_err("ext4_expand_extra_isize_ea ino %lu new_extra_isize %d curr %d\n", + inode->i_ino, new_extra_isize, EXT4_I(inode)->i_extra_isize); + if (inode->i_sb->s_root == NULL) { + pr_err("ext4_expand_extra_isize_ea: fs being unmounted\n"); + return -EINVAL; + } + retry: isize_diff = new_extra_isize - EXT4_I(inode)->i_extra_isize; if (EXT4_I(inode)->i_extra_isize >= new_extra_isize)
Fixing this the clean and proper way, which is by making ext4_xattr_move_to_block() more intelligent/efficient, is left as an exercise to the reader.
Cheers,
- Ted
P.S. Note that this fix is actually needed for the current upstream kernel; the reproducer will trigger in 6.1, although we need to either modify the reproducer to drop the completely pointless nouser_xattr mount option (which is a bit painful since the !@?! mount options is obfuscated by virtue of being in hex for no particular good reason) or by hacking the kernel to ignore that mount options, via a patch like this:
--- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1658,6 +1658,7 @@ static const struct fs_parameter_spec ext4_param_specs[] = { fsparam_flag ("oldalloc", Opt_removed), fsparam_flag ("orlov", Opt_removed), fsparam_flag ("user_xattr", Opt_user_xattr), + fsparam_flag ("nouser_xattr", Opt_removed), fsparam_flag ("acl", Opt_acl), fsparam_flag ("norecovery", Opt_noload), fsparam_flag ("noload", Opt_noload),
Here is a proper, minmized reproducer which reproduces on upstream, for someone who wants to try to work this bug.
On Fri, Dec 16, 2022 at 12:47:16AM -0500, Theodore Ts'o wrote:
Fixing this the clean and proper way, which is by making ext4_xattr_move_to_block() more intelligent/efficient, is left as an exercise to the reader.
For someone who wants to work the bug, here is a cleaner, properly minimzed, easier-for-humans-to-understand reproducer:
#!/bin/bash -vx # # This reproduces an ext4 bug caused by an unfortunate interaction # between lazytime updates happening when a file system is being # unmounted and expand_extra_isize # # Initially discovered via syzkaller: # https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0... #
img=/tmp/foo.img dir=/mnt file=$dir/file0
rm -f $img mke2fs -Fq -t ext4 -I 256 -O ea_inode -b 1024 $img 200k mount $img $dir v=$(dd if=/dev/zero bs=2000 count=1 2>/dev/null | tr '\0' =) touch $file attr -q -s test -V $v $file umount $dir mount -o debug_want_extra_isize=128,lazytime /tmp/foo.img $dir cat $file umount $dir
Theodore Ts'o tytso@mit.edu 于2022年12月16日周五 13:47写道：
On Fri, Dec 16, 2022 at 11:41:16AM +0800, Jun Nie wrote:
This patch[1] is needed on linux-5.15.y because the panic[2] is also found on linux-5.15.y when debugging bug[3]. Back ported patch[4] is confirmed to fix the bug on linux-5.15.y in the latest test of page[3]. Maybe back port on more branches is needed per patch comments.
This is not a proper fix for the syzkaller report being reported here:
https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0...
It's true that the reproducer will no longer trigger, but that's just because the reproducer is just exiting early because it is passing in a mount option which is no longer being accepted. In fact, that mount option is completely unneeded and it's a failing of syzkaller that it doesn't adequately minimize the reproducer by trying to remove various random mount options that are not actually needed. For example, running the reproducer will trigger warnings like this:
EXT4-fs: Ignoring removed nobh optionIf we modify the kernel to simply ignore nouser_xattr, then the reproducer will still trigger. So this is not the right patch to backport.
It's important that people who are trying to fix syzkaller bugs understand what is fundamentally going on, instead of using blunt force patches that simple paper over the issue. Please remember that syzkaller is supposed to help us improve the kernel, and it's not just about trying to reduce the count of open syzkaller reports for its own sake. (This is really much more of a quality of implementation issue, since this is not something that would really ever trigger in real life, nor is it really a security issue --- despite some people thinking that all syzkaller reports are actually security issues, and we must run around like chickens with their heads cut off and until they are all fixed.)
The real root cause of the problem is that the file system is getting mounted with these mount options:
nouser_xattr,acl,debug_want_extra_isize=0x0000000000000080,lazytime,nobh,quota
Of which nouser_attr, acl, nobh, and quota are completely pointless. It's also **super** unfortunate that the reproducer isn't written in C, but this horrible psuedo-assimply language:
memcpy( (void*)0x20000000, "\x6e\x6f\x75\x73\x65\x72\x5f\x78\x61\x74\x74\x72\x2c\x61\x63\x6c\x2c\x64" "\x65\x62\x75\x67\x5f\x77\x61\x6e\x74\x5f\x65\x78\x74\x72\x61\x5f\x69\x73" "\x69\x7a\x65\x3d\x30\x78\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30" "\x30\x30\x38\x30\x2c\x6c\x61\x7a\x79\x74\x69\x6d\x65\x2c\x6e\x6f\x62\x68" "\x2c\x71\x75\x6f\x74\x61\x2c\x00\x3d\x93\x09\x61\x36\x5d\x73\x58\x9c", 89);
...syz_mount_image(0x20000440, 0x20000480, 0x1e, 0x20000000, 2, 0x427, ^^^^^^^^^^ 0x200004c0);
(And again, this is stuff that I've complained to the syzkaller team for years and years and years as being fundamentally developer hostile and disrespects the time of upstream maintainers. ARGH!!!!)
Anyway..... So now let's look at the stack trace:
ext4_xattr_block_set+0x8f8/0x3820 fs/ext4/xattr.c:1971 ext4_xattr_move_to_block fs/ext4/xattr.c:2603 [inline] ext4_xattr_make_inode_space fs/ext4/xattr.c:2672 [inline] ext4_expand_extra_isize_ea+0x1591/0x1f30 fs/ext4/xattr.c:2764 __ext4_expand_extra_isize+0x29e/0x3d0 fs/ext4/inode.c:5826 ext4_try_to_expand_extra_isize fs/ext4/inode.c:5869 [inline] __ext4_mark_inode_dirty+0x4bf/0x7a0 fs/ext4/inode.c:5947 ext4_dirty_inode+0xbc/0x100 fs/ext4/inode.c:5979 __mark_inode_dirty+0x1f9/0x9d0 fs/fs-writeback.c:2431 mark_inode_dirty_sync include/linux/fs.h:2429 [inline] iput+0x155/0x7d0 fs/inode.c:1686 dentry_unlink_inode+0x349/0x430 fs/dcache.c:376 __dentry_kill+0x3e2/0x5d0 fs/dcache.c:582 shrink_dentry_list+0x379/0x4d0 fs/dcache.c:1176 shrink_dcache_parent+0xcd/0x350 do_one_tree fs/dcache.c:1657 [inline] shrink_dcache_for_umount+0x7c/0x1a0 fs/dcache.c:1674 generic_shutdown_super+0x69/0x2d0 fs/super.c:447 kill_block_super+0x80/0xe0 fs/super.c:1395
Because lazytime is enabled, after running the reproducer under strace, what happens is that inode #12 gets touched so its access time is modified, but because lazytime is enabled, we don't actually update the on-disk until we actually unmount the superblock. That's why generic_shutdown_super() is in the stack trace.
At that point, when we shrink the dentry cache, when we eject the inode from memory, iput() needs to update the on-disk inode with the updated atime. So far, so good. But then we call ext4_dirty_inode(), and then that interacts with the "debug_want_extra_isize-=128" mount option. So at this point, we try to expand inode's extra isize space, and in order to do that we have to move some extended attributes.
Unfortunately, how ext4 currently does this is a bit stupid, and it reads the contents of the ea_inode into memory, deletes the ea_inode and then creates a new ea_inode. That works, but it's horribly inefficient, and **that's*** what we should actually fix.
Hi Theodore,
Thanks for the detailed explanation here! I am new in syzkaller task force and not familiar with ext4 subsystem. So my reply here may be stupid as a new comer in ext4 side.
Do you mean we have a chance to expand ea_inode in place for some cases? If so, a new ea_inode with larger space should be created to hold expanded ea_inode data, thus data have to be copied and written out through memory in my mind. Or anything other than CPU/memory can utilized for this to avoid memory usage, such as DMA?
Unfortunately, because we try to create a new ea_inode, when ext4_xattr_block_set() calls the static function (which gets inlined) ext4_xattr_inode_create(), and at that point, the call to ext4_new_inode trips over the fact that the file system is being unmoutned, and sb->s_root has already been set to NULL.
per general understanding of a subsystem uninitialization, a flag shall be marked to reject further operation on the sub-system and flush the pending operation, then free the resource. In such a general method, current handling to create a new ea_inode should not crash even it is stupid. sb->s_root seems to be a key global resource in ext4 subsystem per my understanding, and should not be set as NULL until the last step of unmount operation.
Please help teach me what is wrong in my understanding.
So this is what actually goes *boom*:
ea_inode = ext4_new_inode(handle, inode->i_sb->s_root->d_inode, ^^^^^^^^^^^^^^^^^^^ NULL ptr, oops! S_IFREG | 0600, NULL, inode->i_ino + 1, owner, EXT4_EA_INODE_FL);We can prove this is the issue by using the following debugging patch, which prevents the reproducer from triggering after prining the "fs being unmouinted" message:
--- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -2697,6 +2697,13 @@ int ext4_expand_extra_isize_ea(struct inode *inode, int new_extra_isize, int s_min_extra_isize = le16_to_cpu(sbi->s_es->s_min_extra_isize); int isize_diff; /* How much do we need to grow i_extra_isize */
pr_err("ext4_expand_extra_isize_ea ino %lu new_extra_isize %d curr %d\n",inode->i_ino, new_extra_isize, EXT4_I(inode)->i_extra_isize);if (inode->i_sb->s_root == NULL) {pr_err("ext4_expand_extra_isize_ea: fs being unmounted\n");return -EINVAL;}retry: isize_diff = new_extra_isize - EXT4_I(inode)->i_extra_isize; if (EXT4_I(inode)->i_extra_isize >= new_extra_isize)
Fixing this the clean and proper way, which is by making ext4_xattr_move_to_block() more intelligent/efficient, is left as an exercise to the reader.
Cheers,
- TedP.S. Note that this fix is actually needed for the current upstream kernel; the reproducer will trigger in 6.1, although we need to either modify the reproducer to drop the completely pointless nouser_xattr mount option (which is a bit painful since the !@?! mount options is obfuscated by virtue of being in hex for no particular good reason) or by hacking the kernel to ignore that mount options, via a patch like this:
--- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1658,6 +1658,7 @@ static const struct fs_parameter_spec ext4_param_specs[] = { fsparam_flag ("oldalloc", Opt_removed), fsparam_flag ("orlov", Opt_removed), fsparam_flag ("user_xattr", Opt_user_xattr),
fsparam_flag ("nouser_xattr", Opt_removed), fsparam_flag ("acl", Opt_acl), fsparam_flag ("norecovery", Opt_noload), fsparam_flag ("noload", Opt_noload),
On Mon 19-12-22 17:23:18, Jun Nie wrote:
Theodore Ts'o tytso@mit.edu 于2022年12月16日周五 13:47写道：
On Fri, Dec 16, 2022 at 11:41:16AM +0800, Jun Nie wrote:
This patch[1] is needed on linux-5.15.y because the panic[2] is also found on linux-5.15.y when debugging bug[3]. Back ported patch[4] is confirmed to fix the bug on linux-5.15.y in the latest test of page[3]. Maybe back port on more branches is needed per patch comments.
This is not a proper fix for the syzkaller report being reported here:
https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0...
It's true that the reproducer will no longer trigger, but that's just because the reproducer is just exiting early because it is passing in a mount option which is no longer being accepted. In fact, that mount option is completely unneeded and it's a failing of syzkaller that it doesn't adequately minimize the reproducer by trying to remove various random mount options that are not actually needed. For example, running the reproducer will trigger warnings like this:
EXT4-fs: Ignoring removed nobh optionIf we modify the kernel to simply ignore nouser_xattr, then the reproducer will still trigger. So this is not the right patch to backport.
It's important that people who are trying to fix syzkaller bugs understand what is fundamentally going on, instead of using blunt force patches that simple paper over the issue. Please remember that syzkaller is supposed to help us improve the kernel, and it's not just about trying to reduce the count of open syzkaller reports for its own sake. (This is really much more of a quality of implementation issue, since this is not something that would really ever trigger in real life, nor is it really a security issue --- despite some people thinking that all syzkaller reports are actually security issues, and we must run around like chickens with their heads cut off and until they are all fixed.)
The real root cause of the problem is that the file system is getting mounted with these mount options:
nouser_xattr,acl,debug_want_extra_isize=0x0000000000000080,lazytime,nobh,quota
Of which nouser_attr, acl, nobh, and quota are completely pointless. It's also **super** unfortunate that the reproducer isn't written in C, but this horrible psuedo-assimply language:
memcpy( (void*)0x20000000, "\x6e\x6f\x75\x73\x65\x72\x5f\x78\x61\x74\x74\x72\x2c\x61\x63\x6c\x2c\x64" "\x65\x62\x75\x67\x5f\x77\x61\x6e\x74\x5f\x65\x78\x74\x72\x61\x5f\x69\x73" "\x69\x7a\x65\x3d\x30\x78\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30" "\x30\x30\x38\x30\x2c\x6c\x61\x7a\x79\x74\x69\x6d\x65\x2c\x6e\x6f\x62\x68" "\x2c\x71\x75\x6f\x74\x61\x2c\x00\x3d\x93\x09\x61\x36\x5d\x73\x58\x9c", 89);
...syz_mount_image(0x20000440, 0x20000480, 0x1e, 0x20000000, 2, 0x427, ^^^^^^^^^^ 0x200004c0);
(And again, this is stuff that I've complained to the syzkaller team for years and years and years as being fundamentally developer hostile and disrespects the time of upstream maintainers. ARGH!!!!)
Anyway..... So now let's look at the stack trace:
ext4_xattr_block_set+0x8f8/0x3820 fs/ext4/xattr.c:1971 ext4_xattr_move_to_block fs/ext4/xattr.c:2603 [inline] ext4_xattr_make_inode_space fs/ext4/xattr.c:2672 [inline] ext4_expand_extra_isize_ea+0x1591/0x1f30 fs/ext4/xattr.c:2764 __ext4_expand_extra_isize+0x29e/0x3d0 fs/ext4/inode.c:5826 ext4_try_to_expand_extra_isize fs/ext4/inode.c:5869 [inline] __ext4_mark_inode_dirty+0x4bf/0x7a0 fs/ext4/inode.c:5947 ext4_dirty_inode+0xbc/0x100 fs/ext4/inode.c:5979 __mark_inode_dirty+0x1f9/0x9d0 fs/fs-writeback.c:2431 mark_inode_dirty_sync include/linux/fs.h:2429 [inline] iput+0x155/0x7d0 fs/inode.c:1686 dentry_unlink_inode+0x349/0x430 fs/dcache.c:376 __dentry_kill+0x3e2/0x5d0 fs/dcache.c:582 shrink_dentry_list+0x379/0x4d0 fs/dcache.c:1176 shrink_dcache_parent+0xcd/0x350 do_one_tree fs/dcache.c:1657 [inline] shrink_dcache_for_umount+0x7c/0x1a0 fs/dcache.c:1674 generic_shutdown_super+0x69/0x2d0 fs/super.c:447 kill_block_super+0x80/0xe0 fs/super.c:1395
Because lazytime is enabled, after running the reproducer under strace, what happens is that inode #12 gets touched so its access time is modified, but because lazytime is enabled, we don't actually update the on-disk until we actually unmount the superblock. That's why generic_shutdown_super() is in the stack trace.
At that point, when we shrink the dentry cache, when we eject the inode from memory, iput() needs to update the on-disk inode with the updated atime. So far, so good. But then we call ext4_dirty_inode(), and then that interacts with the "debug_want_extra_isize-=128" mount option. So at this point, we try to expand inode's extra isize space, and in order to do that we have to move some extended attributes.
Unfortunately, how ext4 currently does this is a bit stupid, and it reads the contents of the ea_inode into memory, deletes the ea_inode and then creates a new ea_inode. That works, but it's horribly inefficient, and **that's*** what we should actually fix.
Hi Theodore,
Thanks for the detailed explanation here! I am new in syzkaller task force and not familiar with ext4 subsystem. So my reply here may be stupid as a new comer in ext4 side.
Do you mean we have a chance to expand ea_inode in place for some cases? If so, a new ea_inode with larger space should be created to hold expanded ea_inode data, thus data have to be copied and written out through memory in my mind. Or anything other than CPU/memory can utilized for this to avoid memory usage, such as DMA?
There are two things you seem to be mixing together so let me explain in a bit more detail:
There is normal inode which has some extented attributes. In the filesystem image created by syzbot, *headers* of these extended attributes are stored in the inode. The actual extended attribute content is stored in other inodes - so called ea_inodes. The mount option "debug_want_extra_isize" asks the kernel to make more space in the inode itself (say for higher precision time stamps) so we have now less space for extended attribute headers and we need to move them to an external block. How we currently do it is that we read the whole extended attribute in memory, delete the extended attribute, and then pretend the user called setxattr(2) to store the extended atribute header in the freshly allocated block and the attribute data itself in the newly allocated ea_inode. This works but it is inefficient as we could have just moved the extended attribute header from the inode into the newly allocated block, keep the ea_inode and be done with it...
Unfortunately, because we try to create a new ea_inode, when ext4_xattr_block_set() calls the static function (which gets inlined) ext4_xattr_inode_create(), and at that point, the call to ext4_new_inode trips over the fact that the file system is being unmoutned, and sb->s_root has already been set to NULL.
per general understanding of a subsystem uninitialization, a flag shall be marked to reject further operation on the sub-system and flush the pending operation, then free the resource. In such a general method, current handling to create a new ea_inode should not crash even it is stupid. sb->s_root seems to be a key global resource in ext4 subsystem per my understanding, and should not be set as NULL until the last step of unmount operation.
Please help teach me what is wrong in my understanding.
Well, sb->s_root is important but actually not that much because it just speaks about how the filesystem is attached to the overall directory hierarchy. In particular while sb->s_root is set, it pins corresponding inode in memory so we cannot fully cleanup the filesystem. So I think clearing sb->s_root during unmount process is fine where it is. It is the ea_inode handling that should be tweaked to avoid issues.
Honza
On Mon, Dec 19, 2022 at 05:23:18PM +0800, Jun Nie wrote:
Do you mean we have a chance to expand ea_inode in place for some cases? If so, a new ea_inode with larger space should be created to hold expanded ea_inode data, thus data have to be copied and written out through memory in my mind. Or anything other than CPU/memory can utilized for this to avoid memory usage, such as DMA?
There are two inodes in question here. The first is the base inode, which in this case is /file0. The second is the ea_inode which stores the value of one of the extended attributes. In the syzkaller fuzzed file system, there is an ea_inode field which is already created; it contains a value which is too large to fit in the inode or the extended attribute block; but that's OK, because we can put it in a ea_inode. Unfortunately, we are unnecessarily created and deleting the ea_inode (which contains the xattr *value*) when we move the xattr from in-inode storage to the external xattr block.
Extended attributes can be stored either in the on-disk inode, or in an extended attribute block. The storage in the on-disk inode is limited, but extended attributes stored don't require a random access 4k read as in the case of the extended attribute block. So we try to store extended attributes in the inode if possible --- especially the ones which might be accessed frequently, such as a POSIX ACL or a SELinux security id.
The ext4 inode is composed of two portions. The base 128 byte inode, which is always present, and which is what was originally used in ext2. And the "extra inode fields", which are these fields as currently defined at the end of struct ext4_inode:
__le16 i_extra_isize; __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ __le32 i_crtime; /* File Creation time */ __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ __le32 i_version_hi; /* high 32 bits for 64-bit version */ __le32 i_projid; /* Project ID */
The i_extra_isize is the field that is always present for inodes larger than 128 bytes and for which the ext4 file system feature EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE is enabled. The i_extra_isize field tells us how many of the fields beyond the first 128 are present. These fields are necessary for various "advanced" (newer than ext2) ext4 features, including metadata checksums, support for dates beyond 2038 and sub-second timestamp granularity, file creation time, 64-bit i_version, and the project id for project quotas. Everything beyond i_extra_isize is used for in-inode extended attributes.
Now, what if we need to add extra space for new ext4 features? Well, ext4 has a way of expanding the extra inode fields, and one of the ways to trigger this is via the debugging mount option, debug_want_extra_isize. In this particular syzbot reproducer, the mount option, "debug_want_extra_isize=128" sets the i_extra_isize field to maximum allowable size for a 256 byte inode size, and this means that all extended attributes should be ejected out from in-inode storage to the external extended attribute block. We do this on a best efforts bases, when a modified inode is written back to the disk.
The lazytime mount option delays inode updates until the very last minute. The reason for this is to avoid multiple writes to the inode table blocks. This improves performance by reducing random 4k writes, and for flash based storage, reducing flash wearout for flash-based storage. For hard drives (HDD's), it reducing random 4k writes reduces the need to perform Adjacent Track Interference (ATI) mitigations. ATI mitigations can significantly increase the 99.9 percentile tail latency on file system operations, and decreasing tail latency can be worth $$$ for some use cases[1].
[1] https://research.google/pubs/pub44830
The downside of using lazytime updates is that on a crash, the inode timestamps might not get updated --- but very often, this is not a big deal. And normally, when some other inode in the same inode table block is updated, we take that opportunity to update all of the timestamps that were deferred. Or, in the worst case, this will get delayed until the file system is unounted.
Now, back to the extra space expansion. Eexpanding to allow extra inode fields to be used in the future is a "nice to have" sort of thing. It can fail for a number of reasons, including there not being enough space in the extended attribute block to evict the extended attributes in the inode; or if the file system is full, we might not be able to allocate an external block for the extended attribute block in the first place.
So it's OK for us to simply pass on making space for the extra inode fields if it turns out we happen to be in the process of unmounting file system. However, that doesn't fix the performance problem of unnecessarily deleting and creating the ea_inode when moving the xattr from the inode to the exernal xattr block. So fixing that performance issue is the ideal solution. Simply passing on the extra_isize expansion is the second best issue. Backporting an unrelated fix[2] which papers over the problem by disallowing the mount option nouser_xattr is the worst option, since it doesn't actually fix the underlying file system bug.
[2] commit 2d544ec923dbe5 ("ext4: remove deprecated noacl/nouser_xattr options")
Backporting [2] will shut up the syzbot reproducer, yes. But that's because the syzbot reproducer was inadequately minimized. *This* reproducer, which is a easier for a human to understand and which is appropriately minimized will trigger exact same issue, with or without
#!/bin/bash -vx # # This reproduces an ext4 bug caused by an unfortunate interaction # between lazytime updates happening when a file system is being # unmounted and expand_extra_isize # # Initially discovered via syzkaller: # https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0... # img=/tmp/foo.img dir=/mnt file=$dir/file0
rm -f $img mke2fs -Fq -t ext4 -I 256 -O ea_inode -b 1024 $img 200k mount $img $dir v=$(dd if=/dev/zero bs=2000 count=1 2>/dev/null | tr '\0' =) touch $file attr -q -s test -V $v $file umount $dir mount -o debug_want_extra_isize=128,lazytime /tmp/foo.img $dir cat $file umount $dir
This is why your proposal to backport commit 2d544ec923dbe5 is not the right answer.
per general understanding of a subsystem uninitialization, a flag shall be marked to reject further operation on the sub-system and flush the pending operation, then free the resource. In such a general method, current handling to create a new ea_inode should not crash even it is stupid. sb->s_root seems to be a key global resource in ext4 subsystem per my understanding, and should not be set as NULL until the last step of unmount operation.
That's true in general. And yes, simply bypassing the extra_isize expansion when the file system is being unmounted is certainly better that backporting the unrelated commit[2]. But the true correct fix is to optimize how we migrate the xattr from the in-inode storage to the external xattr block.
This is also at *best* P2 bug, since (a) it's not real security issue; just a null pointer derference, and there is no way this could be leveraged into any kind of denial of service or privilege escalation attack, and (b) it requires root access, and use of a debugging option to enable a code path which is in practice never used in production. It is a syzkaller report, and unfortunately, there seems to be this assumption that all syzkaller issues are P0 or P1 issues that must be remediated right away. Which is not the case in this instance. It's a real bug, and so it should be fixed; but it's not a high priority bug.
That being said, if you'd like ot become more experienced in a portion of ext4 internals, I'd certainly invite you to try to understand how ext4 extended attributes are managed, and try your hand at fixing this bug.
Best regards,
- Ted
On Mon, Dec 19, 2022 at 11:41:11AM -0500, Theodore Ts'o wrote:
On Mon, Dec 19, 2022 at 05:23:18PM +0800, Jun Nie wrote:
Do you mean we have a chance to expand ea_inode in place for some cases? If so, a new ea_inode with larger space should be created to hold expanded ea_inode data, thus data have to be copied and written out through memory in my mind. Or anything other than CPU/memory can utilized for this to avoid memory usage, such as DMA?
There are two inodes in question here. The first is the base inode, which in this case is /file0. The second is the ea_inode which stores the value of one of the extended attributes. In the syzkaller fuzzed file system, there is an ea_inode field which is already created; it contains a value which is too large to fit in the inode or the extended attribute block; but that's OK, because we can put it in a ea_inode. Unfortunately, we are unnecessarily created and deleting the ea_inode (which contains the xattr *value*) when we move the xattr from in-inode storage to the external xattr block.
Extended attributes can be stored either in the on-disk inode, or in an extended attribute block. The storage in the on-disk inode is limited, but extended attributes stored don't require a random access 4k read as in the case of the extended attribute block. So we try to store extended attributes in the inode if possible --- especially the ones which might be accessed frequently, such as a POSIX ACL or a SELinux security id.
The ext4 inode is composed of two portions. The base 128 byte inode, which is always present, and which is what was originally used in ext2. And the "extra inode fields", which are these fields as currently defined at the end of struct ext4_inode:
__le16 i_extra_isize; __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ __le32 i_crtime; /* File Creation time */ __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ __le32 i_version_hi; /* high 32 bits for 64-bit version */ __le32 i_projid; /* Project ID */
The i_extra_isize is the field that is always present for inodes larger than 128 bytes and for which the ext4 file system feature EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE is enabled. The i_extra_isize field tells us how many of the fields beyond the first 128 are present. These fields are necessary for various "advanced" (newer than ext2) ext4 features, including metadata checksums, support for dates beyond 2038 and sub-second timestamp granularity, file creation time, 64-bit i_version, and the project id for project quotas. Everything beyond i_extra_isize is used for in-inode extended attributes.
Now, what if we need to add extra space for new ext4 features? Well, ext4 has a way of expanding the extra inode fields, and one of the ways to trigger this is via the debugging mount option, debug_want_extra_isize. In this particular syzbot reproducer, the mount option, "debug_want_extra_isize=128" sets the i_extra_isize field to maximum allowable size for a 256 byte inode size, and this means that all extended attributes should be ejected out from in-inode storage to the external extended attribute block. We do this on a best efforts bases, when a modified inode is written back to the disk.
The lazytime mount option delays inode updates until the very last minute. The reason for this is to avoid multiple writes to the inode table blocks. This improves performance by reducing random 4k writes, and for flash based storage, reducing flash wearout for flash-based storage. For hard drives (HDD's), it reducing random 4k writes reduces the need to perform Adjacent Track Interference (ATI) mitigations. ATI mitigations can significantly increase the 99.9 percentile tail latency on file system operations, and decreasing tail latency can be worth $$$ for some use cases[1].
[1] https://research.google/pubs/pub44830
The downside of using lazytime updates is that on a crash, the inode timestamps might not get updated --- but very often, this is not a big deal. And normally, when some other inode in the same inode table block is updated, we take that opportunity to update all of the timestamps that were deferred. Or, in the worst case, this will get delayed until the file system is unounted.
Now, back to the extra space expansion. Eexpanding to allow extra inode fields to be used in the future is a "nice to have" sort of thing. It can fail for a number of reasons, including there not being enough space in the extended attribute block to evict the extended attributes in the inode; or if the file system is full, we might not be able to allocate an external block for the extended attribute block in the first place.
So it's OK for us to simply pass on making space for the extra inode fields if it turns out we happen to be in the process of unmounting file system. However, that doesn't fix the performance problem of unnecessarily deleting and creating the ea_inode when moving the xattr from the inode to the exernal xattr block. So fixing that performance issue is the ideal solution. Simply passing on the extra_isize expansion is the second best issue. Backporting an unrelated fix[2] which papers over the problem by disallowing the mount option nouser_xattr is the worst option, since it doesn't actually fix the underlying file system bug.
[2] commit 2d544ec923dbe5 ("ext4: remove deprecated noacl/nouser_xattr options")
Backporting [2] will shut up the syzbot reproducer, yes. But that's because the syzbot reproducer was inadequately minimized. *This* reproducer, which is a easier for a human to understand and which is appropriately minimized will trigger exact same issue, with or without
#!/bin/bash -vx # # This reproduces an ext4 bug caused by an unfortunate interaction # between lazytime updates happening when a file system is being # unmounted and expand_extra_isize # # Initially discovered via syzkaller: # https://syzkaller.appspot.com/bug?id=3613786cb88c93aa1c6a279b1df6a7b201347d0... # img=/tmp/foo.img dir=/mnt file=$dir/file0
rm -f $img mke2fs -Fq -t ext4 -I 256 -O ea_inode -b 1024 $img 200k mount $img $dir v=$(dd if=/dev/zero bs=2000 count=1 2>/dev/null | tr '\0' =) touch $file attr -q -s test -V $v $file umount $dir mount -o debug_want_extra_isize=128,lazytime /tmp/foo.img $dir cat $file umount $dir
This is why your proposal to backport commit 2d544ec923dbe5 is not the right answer.
per general understanding of a subsystem uninitialization, a flag shall be marked to reject further operation on the sub-system and flush the pending operation, then free the resource. In such a general method, current handling to create a new ea_inode should not crash even it is stupid. sb->s_root seems to be a key global resource in ext4 subsystem per my understanding, and should not be set as NULL until the last step of unmount operation.
That's true in general. And yes, simply bypassing the extra_isize expansion when the file system is being unmounted is certainly better that backporting the unrelated commit[2]. But the true correct fix is to optimize how we migrate the xattr from the in-inode storage to the external xattr block.
This is also at *best* P2 bug, since (a) it's not real security issue; just a null pointer derference, and there is no way this could be leveraged into any kind of denial of service or privilege escalation attack, and (b) it requires root access, and use of a debugging option to enable a code path which is in practice never used in production. It is a syzkaller report, and unfortunately, there seems to be this assumption that all syzkaller issues are P0 or P1 issues that must be remediated right away. Which is not the case in this instance. It's a real bug, and so it should be fixed; but it's not a high priority bug.
That being said, if you'd like ot become more experienced in a portion of ext4 internals, I'd certainly invite you to try to understand how ext4 extended attributes are managed, and try your hand at fixing this bug.
Best regards,
- Ted
Thanks for the elabration of the logic here. I guess below change is similiar with what we are expecting. There are 2 question in the change I do not have anwser yet.
But for the crash with NULL sb->s_root, below change does not impact anything because all functions are still called. So I guess the protection with rejecting further request during umount is still needed. Or I still missed something?
fs/ext4/xattr.c | 26 ++++++++++++++++---------- 1 file changed, 16 insertions(+), 10 deletions(-)
diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index 7decaaf27e82..546808dbbdd6 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -2551,9 +2551,8 @@ static int ext4_xattr_move_to_block(handle_t *handle, struct inode *inode,
is = kzalloc(sizeof(struct ext4_xattr_ibody_find), GFP_NOFS); bs = kzalloc(sizeof(struct ext4_xattr_block_find), GFP_NOFS); - buffer = kvmalloc(value_size, GFP_NOFS); b_entry_name = kmalloc(entry->e_name_len + 1, GFP_NOFS); - if (!is || !bs || !buffer || !b_entry_name) { + if (!is || !bs || !b_entry_name) { error = -ENOMEM; goto out; } @@ -2565,14 +2564,21 @@ static int ext4_xattr_move_to_block(handle_t *handle, struct inode *inode,
/* Save the entry name and the entry value */ if (entry->e_value_inum) { + buffer = kvmalloc(value_size, GFP_NOFS); + if (!buffer) { + error = -ENOMEM; + goto out; + } + error = ext4_xattr_inode_get(inode, entry, buffer, value_size); if (error) goto out; } else { size_t value_offs = le16_to_cpu(entry->e_value_offs); - memcpy(buffer, (void *)IFIRST(header) + value_offs, value_size); + buffer = (void *)IFIRST(header) + value_offs; }
+ /* Can we reuse entry->e_name with assumption of \0 for all e_name? */ memcpy(b_entry_name, entry->e_name, entry->e_name_len); b_entry_name[entry->e_name_len] = '\0'; i.name = b_entry_name; @@ -2585,11 +2591,6 @@ static int ext4_xattr_move_to_block(handle_t *handle, struct inode *inode, if (error) goto out;
- /* Remove the chosen entry from the inode */ - error = ext4_xattr_ibody_set(handle, inode, &i, is); - if (error) - goto out; - i.value = buffer; i.value_len = value_size; error = ext4_xattr_block_find(inode, &i, bs); @@ -2597,13 +2598,18 @@ static int ext4_xattr_move_to_block(handle_t *handle, struct inode *inode, goto out;
/* Add entry which was removed from the inode into the block */ + /* Can this function remove in inode xattr automatically? */ error = ext4_xattr_block_set(handle, inode, &i, bs); if (error) goto out; - error = 0; + + /* Remove the chosen entry from the inode */ + error = ext4_xattr_ibody_set(handle, inode, &i, is); + out: kfree(b_entry_name); - kvfree(buffer); + if (entry->e_value_inum && buffer) + kvfree(buffer); if (is) brelse(is->iloc.bh); if (bs)
linux-stable-mirror@lists.linaro.org
-
Jan Kara -
Jun Nie -
Theodore Ts'o