[syzbot] [hfs?] general protection fault in hfsplus_cat_write_inode

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: 26ff969926a0 Merge tag 'rust-7.1' of git://git.kernel.org/..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=123510ce580000
kernel config: https://syzkaller.appspot.com/x/.config?x=bcb138057684c76a
dashboard link: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
compiler: Debian clang version 21.1.8 (++20251221033036+2078da43e25a-1~exp1~20251221153213.50), Debian LLD 21.1.8
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=163510ce580000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=112bdb02580000

Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/8710bc1acc42/disk-26ff9699.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/f03e7f936c8a/vmlinux-26ff9699.xz
kernel image: https://storage.googleapis.com/syzbot-assets/ba58fa2f951e/bzImage-26ff9699.xz
mounted in repro: https://storage.googleapis.com/syzbot-assets/08271f11c795/mount_0.gz

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com

loop0: detected capacity change from 0 to 1024
hfsplus: xattr search failed
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] SMP KASAN PTI
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 0 UID: 0 PID: 6072 Comm: syz.0.17 Not tainted syzkaller #0 PREEMPT_{RT,(full)}
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/18/2026
RIP: 0010:hfsplus_cat_write_inode+0xbe/0x8f0 fs/hfsplus/inode.c:637
Code: f3 f3 4a 89 44 33 51 4a 89 44 33 59 66 42 c7 44 33 61 f3 f3 42 c6 44 33 63 f3 e8 2d 77 23 ff 4d 8d 7d 28 4c 89 f8 48 c1 e8 03 <42> 80 3c 30 00 74 08 4c 89 ff e8 73 9b 89 ff 41 be 98 07 00 00 4d
RSP: 0018:ffffc900040f7880 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff9200081ef14 RCX: ffff888028725b80
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: ffffc900040f7c10 R08: 0000000000000004 R09: 0000000000000004
R10: fffff5200081ef24 R11: fffffbfff1ed4f17 R12: 1ffff9200081ef8c
R13: 0000000000000000 R14: dffffc0000000000 R15: 0000000000000028
FS: 0000555586e4a500(0000) GS:ffff888126332000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f427032f000 CR3: 0000000036c06000 CR4: 00000000003526f0
Call Trace:
<TASK>
hfsplus_unlink+0x57a/0x930 fs/hfsplus/dir.c:435
vfs_unlink+0x272/0x6d0 fs/namei.c:5476
filename_unlinkat+0x3d3/0x610 fs/namei.c:5546
__do_sys_unlink fs/namei.c:5581 [inline]
__se_sys_unlink+0x2e/0x140 fs/namei.c:5578
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0x14d/0xf80 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f464eabc819
Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 e8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007ffd69795788 EFLAGS: 00000246 ORIG_RAX: 0000000000000057
RAX: ffffffffffffffda RBX: 00007f464ed35fa0 RCX: 00007f464eabc819
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000200000000200
RBP: 00007f464eb52c91 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007f464ed35fac R14: 00007f464ed35fa0 R15: 00007f464ed35fa0
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:hfsplus_cat_write_inode+0xbe/0x8f0 fs/hfsplus/inode.c:637
Code: f3 f3 4a 89 44 33 51 4a 89 44 33 59 66 42 c7 44 33 61 f3 f3 42 c6 44 33 63 f3 e8 2d 77 23 ff 4d 8d 7d 28 4c 89 f8 48 c1 e8 03 <42> 80 3c 30 00 74 08 4c 89 ff e8 73 9b 89 ff 41 be 98 07 00 00 4d
RSP: 0018:ffffc900040f7880 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff9200081ef14 RCX: ffff888028725b80
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: ffffc900040f7c10 R08: 0000000000000004 R09: 0000000000000004
R10: fffff5200081ef24 R11: fffffbfff1ed4f17 R12: 1ffff9200081ef8c
R13: 0000000000000000 R14: dffffc0000000000 R15: 0000000000000028
FS: 0000555586e4a500(0000) GS:ffff888126332000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f427032f000 CR3: 0000000036c06000 CR4: 00000000003526f0
----------------
Code disassembly (best guess):
0: f3 f3 4a 89 44 33 51 repz xrelease mov %rax,0x51(%rbx,%r14,1)
7: 4a 89 44 33 59 mov %rax,0x59(%rbx,%r14,1)
c: 66 42 c7 44 33 61 f3 movw $0xf3f3,0x61(%rbx,%r14,1)
13: f3
14: 42 c6 44 33 63 f3 movb $0xf3,0x63(%rbx,%r14,1)
1a: e8 2d 77 23 ff call 0xff23774c
1f: 4d 8d 7d 28 lea 0x28(%r13),%r15
23: 4c 89 f8 mov %r15,%rax
26: 48 c1 e8 03 shr $0x3,%rax
* 2a: 42 80 3c 30 00 cmpb $0x0,(%rax,%r14,1) <-- trapping instruction
2f: 74 08 je 0x39
31: 4c 89 ff mov %r15,%rdi
34: e8 73 9b 89 ff call 0xff899bac
39: 41 be 98 07 00 00 mov $0x798,%r14d
3f: 4d rex.WRB


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

If the report is already addressed, let syzbot know by replying with:
#syz fix: exact-commit-title

If you want syzbot to run the reproducer, reply with:
#syz test: git://repo/address.git branch-or-commit-hash
If you attach or paste a git patch, syzbot will apply it before testing.

If you want to overwrite report's subsystems, reply with:
#syz set subsystems: new-subsystem
(See the list of subsystem names on the web dashboard)

If the report is a duplicate of another one, reply with:
#syz dup: exact-subject-of-another-report

If you want to undo deduplication, reply with:
#syz undup

[syzbot]

For archival purposes, forwarding an incoming command email to
linux-...@vger.kernel.org, syzkall...@googlegroups.com.

***

Subject: [PATCH] hfsplus: fix ignored error return in hfsplus_delete_cat
Author: karti...@gmail.com

#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master

hfsplus_delete_cat() calls hfsplus_delete_all_attrs() to remove all
extended attributes associated with a catalog entry, but silently
discards its return value. When the xattr deletion fails due to
filesystem corruption (as in a crafted image reported by syzbot),
hfsplus_delete_cat() returns 0 (success) to hfsplus_unlink().

hfsplus_unlink() then proceeds to call hfsplus_cat_write_inode() on
the inode in a corrupt, half-deleted state. Inside that function, if
HFSPLUS_IS_RSRC() is true but rsrc_inode was never properly set,
main_inode is assigned NULL. The subsequent dereference of
main_inode->i_nlink triggers a general protection fault, caught by
KASAN as a null-ptr-deref at offset 0x28.

Fix this by capturing the return value of hfsplus_delete_all_attrs()
and propagating genuine errors back to the caller. -ENOENT is
excluded since it signals normal loop termination (no more xattrs
left to delete) and is not an error condition.

Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
Signed-off-by: Deepanshu Kartikey <karti...@gmail.com>
---
fs/hfsplus/catalog.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/fs/hfsplus/catalog.c b/fs/hfsplus/catalog.c
index 02c1eee4a4b8..adbaeabc06ab 100644
--- a/fs/hfsplus/catalog.c
+++ b/fs/hfsplus/catalog.c
@@ -421,8 +421,14 @@ int hfsplus_delete_cat(u32 cnid, struct inode *dir, const struct qstr *str)
hfsplus_mark_inode_dirty(dir, HFSPLUS_I_CAT_DIRTY);

if (type == HFSPLUS_FILE || type == HFSPLUS_FOLDER) {
- if (HFSPLUS_SB(sb)->attr_tree)
- hfsplus_delete_all_attrs(dir, cnid);
+ if (HFSPLUS_SB(sb)->attr_tree) {
+ int attr_err = hfsplus_delete_all_attrs(dir, cnid);
+ if (attr_err && attr_err != -ENOENT) {
+ pr_err("hfsplus: failed to delete xattrs for cnid %u: %d\n",
+ cnid, attr_err);
+ err = attr_err;
+ }
+ }
}

out:
--
2.43.0

[syzbot]

Hello,

syzbot has tested the proposed patch and the reproducer did not trigger any issue:

Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com
Tested-by: syzbot+c0ba77...@syzkaller.appspotmail.com

Tested on:

commit: d142ab35 Merge tag 'crc-for-linus' of git://git.kernel..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=10c27cd2580000
kernel config: https://syzkaller.appspot.com/x/.config?x=532b940251127b3a

dashboard link: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
compiler: Debian clang version 21.1.8 (++20251221033036+2078da43e25a-1~exp1~20251221153213.50), Debian LLD 21.1.8

patch: https://syzkaller.appspot.com/x/patch.diff?x=162e5036580000

Note: testing is done by a robot and is best-effort only.

[syzbot]

For archival purposes, forwarding an incoming command email to
linux-...@vger.kernel.org, syzkall...@googlegroups.com.

***

Subject: [PATCH] hfsplus: fix null-ptr-deref in hfsplus_cat_write_inode

Author: karti...@gmail.com

#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master

When a corrupt HFS+ image is mounted where the hidden directory is
absent, hfsplus_fill_super() handles -ENOENT from
hfsplus_get_hidden_dir_entry() silently and continues with
sbi->hidden_dir left as NULL.

hfsplus_link() and hfsplus_unlink() call
hfsplus_cat_write_inode(sbi->hidden_dir) unconditionally without
checking whether hidden_dir is NULL, triggering a general protection
fault caught by KASAN as a null-ptr-deref at offset 0x28.

Other call sites in dir.c already guard against this by checking
hidden_dir for NULL before use. Apply the same guard to the two
unprotected call sites in hfsplus_link() and hfsplus_unlink().

Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
Signed-off-by: Deepanshu Kartikey <karti...@gmail.com>
---

fs/hfsplus/dir.c | 15 ++++++++-------
1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/fs/hfsplus/dir.c b/fs/hfsplus/dir.c
index d559bf8625f8..a7feef53d8cb 100644
--- a/fs/hfsplus/dir.c
+++ b/fs/hfsplus/dir.c
@@ -362,9 +362,11 @@ static int hfsplus_link(struct dentry *src_dentry, struct inode *dst_dir,
if (res)
goto out;

- res = hfsplus_cat_write_inode(sbi->hidden_dir);
- if (res)
- goto out;
+ if (sbi->hidden_dir) {
+ res = hfsplus_cat_write_inode(sbi->hidden_dir);
+ if (res)
+ goto out;
+ }

res = hfsplus_cat_write_inode(inode);

@@ -431,11 +433,10 @@ static int hfsplus_unlink(struct inode *dir, struct dentry *dentry)
out:
if (!res) {
res = hfsplus_cat_write_inode(dir);
- if (!res) {
+ if (!res && sbi->hidden_dir)
res = hfsplus_cat_write_inode(sbi->hidden_dir);
- if (!res)
- res = hfsplus_cat_write_inode(inode);
- }
+ if (!res)
+ res = hfsplus_cat_write_inode(inode);
}

mutex_unlock(&sbi->vh_mutex);
--
2.43.0

[syzbot]

Hello,

syzbot has tested the proposed patch and the reproducer did not trigger any issue:

Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com
Tested-by: syzbot+c0ba77...@syzkaller.appspotmail.com

Tested on:

commit: e997ac58 Merge tag 'linux_kselftest-next-7.1-rc1' of g..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=10f6fcd2580000
kernel config: https://syzkaller.appspot.com/x/.config?x=8121d5de65640618

dashboard link: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
compiler: Debian clang version 21.1.8 (++20251221033036+2078da43e25a-1~exp1~20251221153213.50), Debian LLD 21.1.8

patch: https://syzkaller.appspot.com/x/patch.diff?x=14f184ce580000

[syzbot]

For archival purposes, forwarding an incoming command email to
linux-...@vger.kernel.org, syzkall...@googlegroups.com.

***

Subject: [PATCH v4] hfsplus: fix null-ptr-deref by creating hidden dir on remount rw

Author: karti...@gmail.com

#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master

hfsplus_reconfigure() does not create the hidden directory when
remounting from read-only to read-write, leaving sbi->hidden_dir
as NULL. This causes a null-ptr-deref when any subsequent
link/unlink/rename operation dereferences it.

Extract hidden directory creation into a helper and call it from
hfsplus_reconfigure() when switching to read-write mode.

Reported-by: syzbot+c0ba77...@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=c0ba772a362e70937dfb
Signed-off-by: Deepanshu Kartikey <karti...@gmail.com>
---

Changes in v4:
- Correct fix: extract hidden dir creation into helper and call
from hfsplus_reconfigure() on remount rw, as suggested by
Vyacheslav Dubeyko.

Changes in v3:
- Correct fix location: guard sbi->hidden_dir in hfsplus_link()
and hfsplus_unlink() in dir.c.

Changes in v2:
- Fixed commit message: hfsplus_delete_cat() has multiple callers,
not just hfsplus_unlink() as incorrectly stated in v1.
---
fs/hfsplus/super.c | 88 ++++++++++++++++++++++++++++------------------
1 file changed, 53 insertions(+), 35 deletions(-)

diff --git a/fs/hfsplus/super.c b/fs/hfsplus/super.c
index 7229a8ae89f9..8b5c39ce4d48 100644
--- a/fs/hfsplus/super.c
+++ b/fs/hfsplus/super.c
@@ -372,15 +372,59 @@ static int hfsplus_statfs(struct dentry *dentry, struct kstatfs *buf)
return 0;
}

+static int hfsplus_create_hidden_dir(struct super_block *sb)
+{
+ struct hfsplus_sb_info *sbi = HFSPLUS_SB(sb);
+ struct inode *root = d_inode(sb->s_root);
+ struct qstr str;
+ int err;
+
+ str.len = sizeof(HFSP_HIDDENDIR_NAME) - 1;
+ str.name = HFSP_HIDDENDIR_NAME;
+
+ mutex_lock(&sbi->vh_mutex);
+ sbi->hidden_dir = hfsplus_new_inode(sb, root, S_IFDIR);
+ if (!sbi->hidden_dir) {
+ mutex_unlock(&sbi->vh_mutex);
+ return -ENOMEM;
+ }
+
+ err = hfsplus_create_cat(sbi->hidden_dir->i_ino, root,
+ &str, sbi->hidden_dir);
+ if (err) {
+ mutex_unlock(&sbi->vh_mutex);
+ goto out_put_hidden_dir;
+ }
+
+ err = hfsplus_init_security(sbi->hidden_dir, root, &str);
+ if (err == -EOPNOTSUPP)
+ err = 0;
+ else if (err) {
+ hfsplus_delete_cat(sbi->hidden_dir->i_ino, root, &str);
+ mutex_unlock(&sbi->vh_mutex);
+ goto out_put_hidden_dir;
+ }
+
+ mutex_unlock(&sbi->vh_mutex);
+ hfsplus_mark_inode_dirty(sbi->hidden_dir,
+ HFSPLUS_I_CAT_DIRTY);
+ return 0;
+
+out_put_hidden_dir:
+ iput(sbi->hidden_dir);
+ sbi->hidden_dir = NULL;
+ return err;
+}
+
static int hfsplus_reconfigure(struct fs_context *fc)
{
struct super_block *sb = fc->root->d_sb;
+ struct hfsplus_sb_info *sbi = HFSPLUS_SB(sb);

sync_filesystem(sb);
if ((bool)(fc->sb_flags & SB_RDONLY) == sb_rdonly(sb))
return 0;
if (!(fc->sb_flags & SB_RDONLY)) {
- struct hfsplus_sb_info *sbi = HFSPLUS_SB(sb);
struct hfsplus_vh *vhdr = sbi->s_vhdr;

if (!(vhdr->attributes & cpu_to_be32(HFSPLUS_VOL_UNMNT))) {
@@ -401,6 +445,12 @@ static int hfsplus_reconfigure(struct fs_context *fc)
fc->sb_flags |= SB_RDONLY;
}
}
+ /*
+ * Create hidden dir if remounting read-write and it does
+ * not exist - required for link/unlink/rename operations.
+ */
+ if (!sb_rdonly(sb) && !sbi->hidden_dir)
+ return hfsplus_create_hidden_dir(sb);
return 0;
}

@@ -595,38 +645,9 @@ static int hfsplus_fill_super(struct super_block *sb, struct fs_context *fc)
hfsplus_sync_fs(sb, 1);

if (!sbi->hidden_dir) {
- mutex_lock(&sbi->vh_mutex);
- sbi->hidden_dir = hfsplus_new_inode(sb, root, S_IFDIR);
- if (!sbi->hidden_dir) {
- mutex_unlock(&sbi->vh_mutex);
- err = -ENOMEM;
+ err = hfsplus_create_hidden_dir(sb);
+ if (err)
goto out_put_root;
- }
- err = hfsplus_create_cat(sbi->hidden_dir->i_ino, root,
- &str, sbi->hidden_dir);
- if (err) {
- mutex_unlock(&sbi->vh_mutex);
- goto out_put_hidden_dir;
- }
-
- err = hfsplus_init_security(sbi->hidden_dir,
- root, &str);
- if (err == -EOPNOTSUPP)
- err = 0; /* Operation is not supported. */
- else if (err) {
- /*
- * Try to delete anyway without
- * error analysis.
- */
- hfsplus_delete_cat(sbi->hidden_dir->i_ino,
- root, &str);
- mutex_unlock(&sbi->vh_mutex);
- goto out_put_hidden_dir;
- }
-
- mutex_unlock(&sbi->vh_mutex);
- hfsplus_mark_inode_dirty(sbi->hidden_dir,
- HFSPLUS_I_CAT_DIRTY);
}
}

@@ -634,9 +655,6 @@ static int hfsplus_fill_super(struct super_block *sb, struct fs_context *fc)
sbi->nls = nls;
return 0;

-out_put_hidden_dir:
- cancel_delayed_work_sync(&sbi->sync_work);
- iput(sbi->hidden_dir);
out_put_root:
dput(sb->s_root);
sb->s_root = NULL;
--
2.43.0

[syzbot]

For archival purposes, forwarding an incoming command email to
linux-...@vger.kernel.org, syzkall...@googlegroups.com.

***

Subject: [PATCH] hfsplus: fix null-ptr-deref by creating hidden dir on remount rw