| Message ID | 20220404084723.79089-2-zheyuma97@gmail.com (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Headers | show |
| Series | Fix divide errors in fbdev drivers | expand |
On 4/4/22 10:47, Zheyu Ma wrote: > 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. > > Fix this by checking whether 'pixclock' is zero in the function > i740fb_check_var(). > > The following log reveals it: > > divide error: 0000 [#1] PREEMPT SMP KASAN PTI > RIP: 0010:i740fb_decode_var drivers/video/fbdev/i740fb.c:444 [inline] > RIP: 0010:i740fb_set_par+0x272f/0x3bb0 drivers/video/fbdev/i740fb.c:739 > Call Trace: > fb_set_var+0x604/0xeb0 drivers/video/fbdev/core/fbmem.c:1036 > do_fb_ioctl+0x234/0x670 drivers/video/fbdev/core/fbmem.c:1112 > fb_ioctl+0xdd/0x130 drivers/video/fbdev/core/fbmem.c:1191 > vfs_ioctl fs/ioctl.c:51 [inline] > __do_sys_ioctl fs/ioctl.c:874 [inline] > > Signed-off-by: Zheyu Ma <zheyuma97@gmail.com> Hello Zheyu, I've applied the patches #2-#7 of this series, but left out this specific patch (for now). As discussed on the mailing list we can try to come up with a better fix (to round up the pixclock when it's invalid). If not, I will apply this one later. Thanks! Helge > --- > drivers/video/fbdev/i740fb.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/video/fbdev/i740fb.c b/drivers/video/fbdev/i740fb.c > index 52cce0db8bd3..b595437a5752 100644 > --- a/drivers/video/fbdev/i740fb.c > +++ b/drivers/video/fbdev/i740fb.c > @@ -657,6 +657,9 @@ static int i740fb_decode_var(const struct fb_var_screeninfo *var, > > static int i740fb_check_var(struct fb_var_screeninfo *var, struct fb_info *info) > { > + if (!var->pixclock) > + return -EINVAL; > + > switch (var->bits_per_pixel) { > case 8: > var->red.offset = var->green.offset = var->blue.offset = 0;
On Fri, Apr 8, 2022 at 3:50 AM Helge Deller <deller@gmx.de> wrote: > > On 4/4/22 10:47, Zheyu Ma wrote: > > 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. > > > > Fix this by checking whether 'pixclock' is zero in the function > > i740fb_check_var(). > > > > The following log reveals it: > > > > divide error: 0000 [#1] PREEMPT SMP KASAN PTI > > RIP: 0010:i740fb_decode_var drivers/video/fbdev/i740fb.c:444 [inline] > > RIP: 0010:i740fb_set_par+0x272f/0x3bb0 drivers/video/fbdev/i740fb.c:739 > > Call Trace: > > fb_set_var+0x604/0xeb0 drivers/video/fbdev/core/fbmem.c:1036 > > do_fb_ioctl+0x234/0x670 drivers/video/fbdev/core/fbmem.c:1112 > > fb_ioctl+0xdd/0x130 drivers/video/fbdev/core/fbmem.c:1191 > > vfs_ioctl fs/ioctl.c:51 [inline] > > __do_sys_ioctl fs/ioctl.c:874 [inline] > > > > Signed-off-by: Zheyu Ma <zheyuma97@gmail.com> > > Hello Zheyu, > > I've applied the patches #2-#7 of this series, but left > out this specific patch (for now). > As discussed on the mailing list we can try to come up with a > better fix (to round up the pixclock when it's invalid). > If not, I will apply this one later. I'm also looking forward to a more appropriate patch for this driver! Thanks, Zheyu Ma
On Friday 08 April 2022 03:58:10 Zheyu Ma wrote: > On Fri, Apr 8, 2022 at 3:50 AM Helge Deller <deller@gmx.de> wrote: > > > > On 4/4/22 10:47, Zheyu Ma wrote: > > > 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. > > > > > > Fix this by checking whether 'pixclock' is zero in the function > > > i740fb_check_var(). > > > > > > The following log reveals it: > > > > > > divide error: 0000 [#1] PREEMPT SMP KASAN PTI > > > RIP: 0010:i740fb_decode_var drivers/video/fbdev/i740fb.c:444 [inline] > > > RIP: 0010:i740fb_set_par+0x272f/0x3bb0 drivers/video/fbdev/i740fb.c:739 > > > Call Trace: > > > fb_set_var+0x604/0xeb0 drivers/video/fbdev/core/fbmem.c:1036 > > > do_fb_ioctl+0x234/0x670 drivers/video/fbdev/core/fbmem.c:1112 > > > fb_ioctl+0xdd/0x130 drivers/video/fbdev/core/fbmem.c:1191 > > > vfs_ioctl fs/ioctl.c:51 [inline] > > > __do_sys_ioctl fs/ioctl.c:874 [inline] > > > > > > Signed-off-by: Zheyu Ma <zheyuma97@gmail.com> > > > > Hello Zheyu, > > > > I've applied the patches #2-#7 of this series, but left > > out this specific patch (for now). > > As discussed on the mailing list we can try to come up with a > > better fix (to round up the pixclock when it's invalid). > > If not, I will apply this one later. > > I'm also looking forward to a more appropriate patch for this driver! I was not able to reproduce it at first but finally found it: the monitor must be unplugged. If a valid EDID is present, fb_validate_mode() call in i740fb_check_var() will refuse zero pixclock. Haven't found any obvious way to correct zero pixclock value. Most other drivers simply return -EINVAL. > Thanks, > Zheyu Ma >
On 4/10/22 11:02, Ondrej Zary wrote: > On Friday 08 April 2022 03:58:10 Zheyu Ma wrote: >> On Fri, Apr 8, 2022 at 3:50 AM Helge Deller <deller@gmx.de> wrote: >>> >>> On 4/4/22 10:47, Zheyu Ma wrote: >>>> 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. >>>> >>>> Fix this by checking whether 'pixclock' is zero in the function >>>> i740fb_check_var(). >>>> >>>> The following log reveals it: >>>> >>>> divide error: 0000 [#1] PREEMPT SMP KASAN PTI >>>> RIP: 0010:i740fb_decode_var drivers/video/fbdev/i740fb.c:444 [inline] >>>> RIP: 0010:i740fb_set_par+0x272f/0x3bb0 drivers/video/fbdev/i740fb.c:739 >>>> Call Trace: >>>> fb_set_var+0x604/0xeb0 drivers/video/fbdev/core/fbmem.c:1036 >>>> do_fb_ioctl+0x234/0x670 drivers/video/fbdev/core/fbmem.c:1112 >>>> fb_ioctl+0xdd/0x130 drivers/video/fbdev/core/fbmem.c:1191 >>>> vfs_ioctl fs/ioctl.c:51 [inline] >>>> __do_sys_ioctl fs/ioctl.c:874 [inline] >>>> >>>> Signed-off-by: Zheyu Ma <zheyuma97@gmail.com> >>> >>> Hello Zheyu, >>> >>> I've applied the patches #2-#7 of this series, but left >>> out this specific patch (for now). >>> As discussed on the mailing list we can try to come up with a >>> better fix (to round up the pixclock when it's invalid). >>> If not, I will apply this one later. >> >> I'm also looking forward to a more appropriate patch for this driver! > > I was not able to reproduce it at first but finally found it: the > monitor must be unplugged. If a valid EDID is present, > fb_validate_mode() call in i740fb_check_var() will refuse zero > pixclock. > > Haven't found any obvious way to correct zero pixclock value. Most other drivers simply return -EINVAL. Thanks for checking, Ondrej! So, I'll apply the EINVAL-patch from Zheyu for now. Helge
diff --git a/drivers/video/fbdev/i740fb.c b/drivers/video/fbdev/i740fb.c index 52cce0db8bd3..b595437a5752 100644 --- a/drivers/video/fbdev/i740fb.c +++ b/drivers/video/fbdev/i740fb.c @@ -657,6 +657,9 @@ static int i740fb_decode_var(const struct fb_var_screeninfo *var, static int i740fb_check_var(struct fb_var_screeninfo *var, struct fb_info *info) { + if (!var->pixclock) + return -EINVAL; + switch (var->bits_per_pixel) { case 8: var->red.offset = var->green.offset = var->blue.offset = 0;
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. Fix this by checking whether 'pixclock' is zero in the function i740fb_check_var(). The following log reveals it: divide error: 0000 [#1] PREEMPT SMP KASAN PTI RIP: 0010:i740fb_decode_var drivers/video/fbdev/i740fb.c:444 [inline] RIP: 0010:i740fb_set_par+0x272f/0x3bb0 drivers/video/fbdev/i740fb.c:739 Call Trace: fb_set_var+0x604/0xeb0 drivers/video/fbdev/core/fbmem.c:1036 do_fb_ioctl+0x234/0x670 drivers/video/fbdev/core/fbmem.c:1112 fb_ioctl+0xdd/0x130 drivers/video/fbdev/core/fbmem.c:1191 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] Signed-off-by: Zheyu Ma <zheyuma97@gmail.com> --- drivers/video/fbdev/i740fb.c | 3 +++ 1 file changed, 3 insertions(+)