[PATCH v2 RESEND] gfs2: reject oversized dinode sizes before i_size_write

Jiaming Zhang posted 1 patch 3 weeks, 5 days ago
fs/gfs2/glops.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
[PATCH v2 RESEND] gfs2: reject oversized dinode sizes before i_size_write
Posted by Jiaming Zhang 3 weeks, 5 days ago
A corrupted GFS2 image can store a dinode size that is larger than what VFS
i_size can represent. gfs2_dinode_in() reads the on-disk di_size as a u64 and
writes it directly into inode->i_size. If the value is larger than S64_MAX, the
incore i_size becomes negative. That negative value can bypass the existing
stuffed inode size check:

inode->i_size > gfs2_max_stuffed_size(ip)

Later, gfs2_quotad may try to sync the quota file and unstuff the quota inode.
gfs2_unstuffer_folio() reads the negative i_size into an unsigned length and
passes it to memcpy(), turning it into a huge copy size and triggering a
out-of-bound issue.

Reject dinodes whose size exceeds sb->s_maxbytes before storing the value in
inode->i_size. Also make the stuffed inode check use the raw on-disk size while
it is still unsigned.

Fixes: 70376c7ff312 ("gfs2: Always check inode size of inline inodes")
Closes: https://lore.kernel.org/lkml/CANypQFaF6bvORKKbRALvEL0k_epFaneFiOQqco4gjdmKVbdURg@mail.gmail.com/
Assisted-by: Codex:gpt-5.5-xhigh
Cc: stable@vger.kernel.org
Signed-off-by: Jiaming Zhang <r772577952@gmail.com>
---
 fs/gfs2/glops.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c
index 28f32424ee64..8c5d257451d5 100644
--- a/fs/gfs2/glops.c
+++ b/fs/gfs2/glops.c
@@ -393,11 +393,16 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
 	umode_t mode = be32_to_cpu(str->di_mode);
 	struct inode *inode = &ip->i_inode;
 	bool is_new = inode_state_read_once(inode) & I_NEW;
+	u64 size = be64_to_cpu(str->di_size);
 
 	if (unlikely(ip->i_no_addr != be64_to_cpu(str->di_num.no_addr))) {
 		gfs2_consist_inode(ip);
 		return -EIO;
 	}
+	if (unlikely(size > (u64)inode->i_sb->s_maxbytes)) {
+		gfs2_consist_inode(ip);
+		return -EIO;
+	}
 	if (unlikely(!is_new && inode_wrong_type(inode, mode))) {
 		gfs2_consist_inode(ip);
 		return -EIO;
@@ -418,7 +423,7 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
 	i_uid_write(inode, be32_to_cpu(str->di_uid));
 	i_gid_write(inode, be32_to_cpu(str->di_gid));
 	set_nlink(inode, be32_to_cpu(str->di_nlink));
-	i_size_write(inode, be64_to_cpu(str->di_size));
+	i_size_write(inode, size);
 	gfs2_set_inode_blocks(inode, be64_to_cpu(str->di_blocks));
 	atime.tv_sec = be64_to_cpu(str->di_atime);
 	atime.tv_nsec = be32_to_cpu(str->di_atime_nsec);
@@ -462,7 +467,7 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
 		return -EIO;
 	}
 
-	if (gfs2_is_stuffed(ip) && inode->i_size > gfs2_max_stuffed_size(ip)) {
+	if (gfs2_is_stuffed(ip) && size > gfs2_max_stuffed_size(ip)) {
 		gfs2_consist_inode(ip);
 		return -EIO;
 	}
-- 
2.43.0
Re: [PATCH v2 RESEND] gfs2: reject oversized dinode sizes before i_size_write
Posted by Andreas Gruenbacher 3 weeks, 5 days ago
Hello,

On Mon, Aug 31, 2026 at 12:54 PM Jiaming Zhang <r772577952@gmail.com> wrote:
> A corrupted GFS2 image can store a dinode size that is larger than what VFS
> i_size can represent. gfs2_dinode_in() reads the on-disk di_size as a u64 and
> writes it directly into inode->i_size. If the value is larger than S64_MAX, the
> incore i_size becomes negative. That negative value can bypass the existing
> stuffed inode size check:
>
> inode->i_size > gfs2_max_stuffed_size(ip)
>
> Later, gfs2_quotad may try to sync the quota file and unstuff the quota inode.
> gfs2_unstuffer_folio() reads the negative i_size into an unsigned length and
> passes it to memcpy(), turning it into a huge copy size and triggering a
> out-of-bound issue.
>
> Reject dinodes whose size exceeds sb->s_maxbytes before storing the value in
> inode->i_size. Also make the stuffed inode check use the raw on-disk size while
> it is still unsigned.

applied with the changes noted below.

> Fixes: 70376c7ff312 ("gfs2: Always check inode size of inline inodes")
> Closes: https://lore.kernel.org/lkml/CANypQFaF6bvORKKbRALvEL0k_epFaneFiOQqco4gjdmKVbdURg@mail.gmail.com/
> Assisted-by: Codex:gpt-5.5-xhigh
> Cc: stable@vger.kernel.org
> Signed-off-by: Jiaming Zhang <r772577952@gmail.com>
> ---
>  fs/gfs2/glops.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c
> index 28f32424ee64..8c5d257451d5 100644
> --- a/fs/gfs2/glops.c
> +++ b/fs/gfs2/glops.c
> @@ -393,11 +393,16 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
>         umode_t mode = be32_to_cpu(str->di_mode);
>         struct inode *inode = &ip->i_inode;
>         bool is_new = inode_state_read_once(inode) & I_NEW;
> +       u64 size = be64_to_cpu(str->di_size);
>
>         if (unlikely(ip->i_no_addr != be64_to_cpu(str->di_num.no_addr))) {
>                 gfs2_consist_inode(ip);
>                 return -EIO;
>         }
> +       if (unlikely(size > (u64)inode->i_sb->s_maxbytes)) {

The (u64) cast here is unnecessary, so I'm removing that.

> +               gfs2_consist_inode(ip);
> +               return -EIO;
> +       }
>         if (unlikely(!is_new && inode_wrong_type(inode, mode))) {
>                 gfs2_consist_inode(ip);
>                 return -EIO;
> @@ -418,7 +423,7 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
>         i_uid_write(inode, be32_to_cpu(str->di_uid));
>         i_gid_write(inode, be32_to_cpu(str->di_gid));
>         set_nlink(inode, be32_to_cpu(str->di_nlink));

 Also, I'm moving the additional consistency check here.

> -       i_size_write(inode, be64_to_cpu(str->di_size));
> +       i_size_write(inode, size);
>         gfs2_set_inode_blocks(inode, be64_to_cpu(str->di_blocks));
>         atime.tv_sec = be64_to_cpu(str->di_atime);
>         atime.tv_nsec = be32_to_cpu(str->di_atime_nsec);
> @@ -462,7 +467,7 @@ static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
>                 return -EIO;
>         }
>
> -       if (gfs2_is_stuffed(ip) && inode->i_size > gfs2_max_stuffed_size(ip)) {
> +       if (gfs2_is_stuffed(ip) && size > gfs2_max_stuffed_size(ip)) {
>                 gfs2_consist_inode(ip);
>                 return -EIO;
>         }
> --
> 2.43.0
>

Thanks,
Andreas