Skip to contentSkip to conversation

video: fbdev: neofb: add a check against divide error

PATCH In review Zheyu Ma 2 participants 2 messages Jul 21, 2021 Framebuffer layer

lore
Share & tools
lore

Zheyu Ma

author
#0 lore raw

PATCH 1/1 video: fbdev: neofb: add a check against divide error

The userspace program could pass any values to the driver through ioctl() interface. If the driver doesn't check the value of 'pixclock', it may cause divide error because of the 'PICOS2KHZ' macro.

Fix this by checking whether 'pixclock' is zero first.

The following log reveals it:

[   53.093806] divide error: 0000 [#1] PREEMPT SMP KASAN PTI
[   53.093838] CPU: 3 PID: 11763 Comm: hang Not tainted 5.14.0-rc2-00478-g2734d6c1b1a0 #215
[   53.093859] RIP: 0010:neofb_check_var+0x80/0xe50
[   53.093951] Call Trace:
[   53.093956]  ? neofb_setcolreg+0x2b0/0x2b0
[   53.093968]  fb_set_var+0x2e4/0xeb0
[   53.093977]  ? fb_blank+0x1a0/0x1a0
[   53.093984]  ? lock_acquire+0x1ef/0x530
[   53.093996]  ? lock_release+0x810/0x810
[   53.094005]  ? lock_is_held_type+0x100/0x140
[   53.094016]  ? ___might_sleep+0x1ee/0x2d0
[   53.094028]  ? __mutex_lock+0x620/0x1190
[   53.094036]  ? do_fb_ioctl+0x313/0x700
[   53.094044]  ? mutex_lock_io_nested+0xfa0/0xfa0
[   53.094051]  ? __this_cpu_preempt_check+0x1d/0x30
[   53.094060]  ? _raw_spin_unlock_irqrestore+0x46/0x60
[   53.094069]  ? lockdep_hardirqs_on+0x59/0x100
[   53.094076]  ? _raw_spin_unlock_irqrestore+0x46/0x60
[   53.094085]  ? trace_hardirqs_on+0x6a/0x1c0
[   53.094096]  do_fb_ioctl+0x31e/0x700
---
 drivers/video/fbdev/neofb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/video/fbdev/neofb.c +1 -1
index c0f4f402da3f..966df2a07360 100644
@@ -585,7 +585,7 @@ neofb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
 
 	DBG("neofb_check_var");
 
-	if (PICOS2KHZ(var->pixclock) > par->maxClock)
+	if (var->pixclock && PICOS2KHZ(var->pixclock) > par->maxClock)
 		return -EINVAL;
 
 	/* Is the mode larger than the LCD panel? */
Signature · 1 line
2.17.6
#1 lore raw

Hi Zheyu,

Quoted 29 lines · Zheyu Ma

The userspace program could pass any values to the driver through ioctl() interface. If the driver doesn't check the value of 'pixclock', it may cause divide error because of the 'PICOS2KHZ' macro.

Fix this by checking whether 'pixclock' is zero first.

The following log reveals it:

[   53.093806] divide error: 0000 [#1] PREEMPT SMP KASAN PTI
[   53.093838] CPU: 3 PID: 11763 Comm: hang Not tainted 5.14.0-rc2-00478-g2734d6c1b1a0 #215
[   53.093859] RIP: 0010:neofb_check_var+0x80/0xe50
[   53.093951] Call Trace:
[   53.093956]  ? neofb_setcolreg+0x2b0/0x2b0
[   53.093968]  fb_set_var+0x2e4/0xeb0
[   53.093977]  ? fb_blank+0x1a0/0x1a0
[   53.093984]  ? lock_acquire+0x1ef/0x530
[   53.093996]  ? lock_release+0x810/0x810
[   53.094005]  ? lock_is_held_type+0x100/0x140
[   53.094016]  ? ___might_sleep+0x1ee/0x2d0
[   53.094028]  ? __mutex_lock+0x620/0x1190
[   53.094036]  ? do_fb_ioctl+0x313/0x700
[   53.094044]  ? mutex_lock_io_nested+0xfa0/0xfa0
[   53.094051]  ? __this_cpu_preempt_check+0x1d/0x30
[   53.094060]  ? _raw_spin_unlock_irqrestore+0x46/0x60
[   53.094069]  ? lockdep_hardirqs_on+0x59/0x100
[   53.094076]  ? _raw_spin_unlock_irqrestore+0x46/0x60
[   53.094085]  ? trace_hardirqs_on+0x6a/0x1c0
[   53.094096]  do_fb_ioctl+0x31e/0x700
Quoted 1 line

I looked if we could move this check up to the caller, but it seems better to keep it per driver.
Added the patch to drm-misc-next, it will appera in -next in around one week.

Sam

Quoted 19 lines
---
 drivers/video/fbdev/neofb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/video/fbdev/neofb.c +1 -1
index c0f4f402da3f..966df2a07360 100644
@@ -585,7 +585,7 @@ neofb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
 
 	DBG("neofb_check_var");
 
-	if (PICOS2KHZ(var->pixclock) > par->maxClock)
+	if (var->pixclock && PICOS2KHZ(var->pixclock) > par->maxClock)
 		return -EINVAL;
 
 	/* Is the mode larger than the LCD panel? */
Signature · 1 line
2.17.6