Re: [PATCH] fbcon: fix NULL pointer dereference for a console without vc_data

[Helge Deller]

On 6/24/26 23:11, Ian Bridges wrote:
> fbcon_new_modelist() runs when a framebuffer's modelist changes. For each
> console mapped to it with fb_display[i].mode set, it reads vc_cons[i].d and
> passes the vc_num to fbcon_set_disp(). This assumes a console with a mode
> set has a vc_data, but it can be NULL. fbcon_set_disp() sets
> fb_display[i].mode before it checks vc_data, and fbcon_deinit() leaves the
> mode set after the vc_data is freed. fbcon_new_modelist() then dereferences
> the NULL vc_data.
>
> Keep fb_display[i].mode set only while the console has a vc_data. Check
> vc_data before setting the mode in fbcon_set_disp(), and clear the mode in
> fbcon_deinit(). The existing mode check in fbcon_new_modelist() then skips
> such consoles.
>
> Reported-by: syzbot+42525d636f430fd5d983@xxxxxxxxxxxxxxxxxxxxxxxxx
> Closes: https://syzkaller.appspot.com/bug?extid=42525d636f430fd5d983
> Cc: stable@xxxxxxxxxxxxxxx
> Assisted-by: Claude:claude-opus-4-8
> Signed-off-by: Ian Bridges <icb@xxxxxxxxxxxx>
> ---
> This patch fixes a NULL pointer dereference in the framebuffer console code.
> fbcon_new_modelist() dereferences a NULL vc_data. It was found while writing
> a reproducer for a separate use-after-free in store_modes(). Sashiko
> independently flagged the same dereference in its review of the fix for that
> use-after-free [1].
>
> The dereference happens when a console has fb_display[i].mode set but no
> vc_data, and the modelist is then replaced, as follows.
>
> 1. A console ends up with fb_display[i].mode set while vc_cons[i].d is NULL.
>     Either fbcon_set_disp() sets the mode (fbcon.c:1446) before it checks
>     vc_data, so mapping an unused console with FBIOPUT_CON2FBMAP leaves the
>     mode set, or fbcon_deinit() (fbcon.c:1268) frees the vc_data but keeps the
>     mode.
> 2. A write to the modes attribute calls store_modes() (fbsysfs.c:91), which
>     replaces the modelist and calls fb_new_modelist() (fbsysfs.c:108).
> 3. fb_new_modelist() calls fbcon_new_modelist() (fbmem.c:770).
> 4. fbcon_new_modelist() walks the consoles mapped to the framebuffer, takes
>     vc = vc_cons[i].d for the one with the mode set, and reads vc->vc_num with
>     vc NULL (fbcon.c:3046). This is a NULL pointer dereference.
>
> This patch does not change that line:
>
>      fbcon_set_disp(info, &var, vc->vc_num);
>
> vc is vc_cons[i].d, which is NULL.
>
> syzbot reported the same crash, with the call stack store_modes() ->
> fb_new_modelist() -> fbcon_new_modelist() [2]. That report had no reproducer
> and was closed as obsolete.
>
> It reproduces on a KASAN kernel with two framebuffers, two ways:
>
> 1. Map an unused console to one framebuffer and back to the other with
>     FBIOPUT_CON2FBMAP, then write a modelist to that framebuffer's modes
>     attribute.
> 2. Open and close /dev/ttyN, then write a modelist to its framebuffer's
>     modes attribute.
>
> The reproducer was written with the help of a coding agent (Claude Code). The
> patch is against commit 3726ce7f6cef on the for-next branch of
> git://git.kernel.org/pub/scm/linux/kernel/git/deller/linux-fbdev.git. The
> file offsets above are from that commit. The dereference is present in the
> initial 2.6.12-rc2 import, so there is no Fixes tag.
>
> [1] https://lore.kernel.org/all/20260622080749.D7FC61F000E9@xxxxxxxxxxxxxxx/
> [2] https://syzkaller.appspot.com/bug?extid=42525d636f430fd5d983
>
>   drivers/video/fbdev/core/fbcon.c | 7 ++++---
>   1 file changed, 4 insertions(+), 3 deletions(-)

applied.

Thanks!
Helge