Re: [PATCH 1/2] video: fbdev: core: cfbcopyarea: fix sloppy typing

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello!

On 9/19/23 10:05 AM, Helge Deller wrote:

>> In cfb_copyarea(), when initializing *unsigned long const* bits_per_line
>> __u32 typed fb_fix_screeninfo::line_length gets multiplied by 8u -- which
>> might overflow __u32; multiplying by 8UL instead should fix that...
>> Also, that bits_per_line constant is used to advance *unsigned* src_idx
>> and dst_idx variables -- which might be overflowed as well; declaring
>> them as *unsigned long* should fix that too...
>>
>> Found by Linux Verification Center (linuxtesting.org) with the Svace static
>> analysis tool.
>>
>> Signed-off-by: Sergey Shtylyov <s.shtylyov@xxxxxx>
>> Cc: stable@xxxxxxxxxxxxxxx
>> ---
>>   drivers/video/fbdev/core/cfbcopyarea.c | 5 +++--
>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/video/fbdev/core/cfbcopyarea.c b/drivers/video/fbdev/core/cfbcopyarea.c
>> index 6d4bfeecee35..b67ba69ea2fb 100644
>> --- a/drivers/video/fbdev/core/cfbcopyarea.c
>> +++ b/drivers/video/fbdev/core/cfbcopyarea.c
>> @@ -382,10 +382,11 @@ void cfb_copyarea(struct fb_info *p, const struct fb_copyarea *area)
>>   {
>>       u32 dx = area->dx, dy = area->dy, sx = area->sx, sy = area->sy;
>>       u32 height = area->height, width = area->width;
>> -    unsigned long const bits_per_line = p->fix.line_length*8u;
>> +    unsigned long const bits_per_line = p->fix.line_length * 8UL;
> 
> you wrote:
>> __u32 typed fb_fix_screeninfo::line_length gets multiplied by 8u -- which
>> might overflow __u32; multiplying by 8UL instead should fix that...
> 
> This would only be true on 64-bit CPUs, where unsigned long is 64 bits,

   Right, Svace was run with the arm64 and x86_64 configs -- and I forgot
to make the emphasis on the 64-bit specifics here...

> while on 32-bit CPUs, it's still 32 bits (same as _u32).

   Yes, indeed. That *unsigned long const* doesn't seem justified
at all then...

> Instead we could make bits_per_line __u32 (or unsigned int) too.

   Yes. Will you accept such a patch? :-)

>>       unsigned long __iomem *base = NULL;
>>       int bits = BITS_PER_LONG, bytes = bits >> 3;
>> -    unsigned dst_idx = 0, src_idx = 0, rev_copy = 0;
>> +    unsigned long dst_idx = 0, src_idx = 0;
> 
> An "unsigned int" can address at least up to 4GB, which is fully sufficent here.

   Good to know! :-)

> So, both patches don't have any real effect.
> NAK.

   Thanks for your time!

> Helge

MBR, Sergey



[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux