[SRU][X][PATCH 1/1] ocfs2: validate fast symlink target during inode read

[Javier Carrasco]

From: Zhang Cen <rollkingzzc at gmail.com>

ocfs2_validate_inode_block() already rejects several inconsistent
self-contained dinodes before they are exposed to the rest of the
filesystem.  Fast symlinks need the same treatment.

A zero-cluster symlink is treated as a fast symlink and later read through
page_get_link() and ocfs2_fast_symlink_read_folio().  That path uses
strnlen() on the inline payload and then copies len + 1 bytes into the
folio.  If a corrupt dinode stores an i_size that does not fit the inline
area or omits the terminating NUL at i_size, that copy reads past the end
of the inode block buffer.

Reject zero-cluster symlink dinodes whose i_size exceeds the inline
fast-symlink capacity or whose inline payload is not NUL-terminated
exactly at i_size when the inode block is validated.  This keeps malformed
fast symlinks from reaching the read path.

Validation reproduced this kernel report:
KASAN use-after-free in ocfs2_fast_symlink_read_folio+0x12c/0x1f0
RIP: 0033:0x7f5c6d859aa7
Read of size 3905
Call trace:
  dump_stack_lvl+0x66/0xa0 (?:?)
  print_report+0xce/0x630 (?:?)
  ocfs2_fast_symlink_read_folio+0x12c/0x1f0 (fs/ocfs2/inode.c:?)
  srso_alias_return_thunk+0x5/0xfbef5 (?:?)
  __virt_addr_valid+0x19f/0x330 (?:?)
  kasan_report+0xe0/0x110 (?:?)
  kasan_check_range+0x105/0x1b0 (?:?)
  __asan_memcpy+0x23/0x60 (?:?)
  filemap_read_folio+0x27/0xe0 (?:?)
  filemap_read_folio+0x35/0xe0 (?:?)
  do_read_cache_folio+0x138/0x230 (?:?)
  __page_get_link+0x26/0x110 (?:?)
  page_get_link+0x2e/0x70 (?:?)
  vfs_readlink+0x15e/0x250 (?:?)
  touch_atime+0x4d/0x370 (?:?)
  do_readlinkat+0x186/0x200 (?:?)
  do_user_addr_fault+0x65a/0x890 (?:?)
  __x64_sys_readlink+0x46/0x60 (?:?)
  do_syscall_64+0x115/0x6a0 (arch/x86/entry/syscall_64.c:87)
  entry_SYSCALL_64_after_hwframe+0x77/0x7f (?:?)

Link: https://lore.kernel.org/20260528151230.361127-1-rollkingzzc@gmail.com
Fixes: ea022dfb3c2a ("ocfs: simplify symlink handling")
Assisted-by: Codex:gpt-5.5
Signed-off-by: Zhang Cen <rollkingzzc at gmail.com>
Reviewed-by: Joseph Qi <joseph.qi at linux.alibaba.com>
Cc: Gui-Dong Han <2045gemini at gmail.com>
Cc: Mark Fasheh <mark at fasheh.com>
Cc: Joel Becker <jlbec at evilplan.org>
Cc: Junxiao Bi <junxiao.bi at oracle.com>
Cc: Changwei Ge <gechangwei at live.cn>
Cc: Jun Piao <piaojun at huawei.com>
Cc: Heming Zhao <heming.zhao at suse.com>
Signed-off-by: Andrew Morton <akpm at linux-foundation.org>
(backported from commit e234973f286ed2e8961a24561ec91594ec3e3ff8)
[javier-carrasco: manually applied as the surrounding code (independent
if statements) is not available.]
CVE-2026-74350
Signed-off-by: Javier Carrasco <javier.carrasco at canonical.com>
---
 fs/ocfs2/inode.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
index 8f87e05ee25d..8ad4f41cf196 100644
--- a/fs/ocfs2/inode.c
+++ b/fs/ocfs2/inode.c
@@ -1390,6 +1390,29 @@ int ocfs2_validate_inode_block(struct super_block *sb,
 		goto bail;
 	}

+	if (S_ISLNK(le16_to_cpu(di->i_mode)) &&
+	    !le32_to_cpu(di->i_clusters)) {
+		int max_inline = ocfs2_fast_symlink_chars(sb);
+		u64 i_size = le64_to_cpu(di->i_size);
+
+		if (i_size >= max_inline) {
+			rc = ocfs2_error(sb,
+					 "Invalid dinode #%llu: fast symlink i_size %llu exceeds max %d\n",
+					 (unsigned long long)bh->b_blocknr,
+					 (unsigned long long)i_size,
+					 max_inline - 1);
+			goto bail;
+		}
+
+		if (strnlen((char *)di->id2.i_symlink, i_size + 1) != i_size) {
+			rc = ocfs2_error(sb,
+					 "Invalid dinode #%llu: fast symlink is not NUL-terminated at i_size %llu\n",
+					 (unsigned long long)bh->b_blocknr,
+					 (unsigned long long)i_size);
+			goto bail;
+		}
+	}
+
 	rc = 0;

 bail:
--
2.43.0