Re: [CHECKER] Probable security holes in 2.6.5

[Chris Wright]

* Ken Ashcraft (ken@xxxxxxxxxxxx) wrote:
> [BUG] 
> /home/kash/linux/linux-2.6.5/drivers/scsi/gdth.c:5068:ioc_general:
> ERROR:TAINT: 5059:5068:Passing unbounded user value "((gen).data_len +
> (gen).sense_len)" as arg 2 to function "copy_from_user", which uses it
> unsafely in model [SOURCE_MODEL=(lib,copy_from_user,user,taintscalar)]
> [SINK_MODEL=(lib,copy_from_user,user,trustingsink)]    [PATH=
> "((gen).data_len + (gen).sense_len) != 0" on line 5064 is false =>
> "(gen).ionode >= gdth_ctr_count" on line 5059 is false =>
> "copy_from_user != 0" on line 5059 is false] 
> #else
> 	Scsi_Cmnd scp;
> 	Scsi_Device sdev;
> #endif
>         
> Start --->
>         if (copy_from_user(&gen, (char *)arg,
> sizeof(gdth_ioctl_general)) ||
>             gen.ionode >= gdth_ctr_count)
>             return -EFAULT;
>         hanum = gen.ionode; 
>         ha = HADATA(gdth_ctr_tab[hanum]);
>         if (gen.data_len + gen.sense_len != 0) {
>             if (!(buf = gdth_ioctl_alloc(hanum, gen.data_len +
> gen.sense_len, 
>                                          FALSE, &paddr)))
>                 return -EFAULT;
> Error --->
>             if (copy_from_user(buf, (char *)arg +
> sizeof(gdth_ioctl_general),  
>                                gen.data_len + gen.sense_len)) {
>                 gdth_ioctl_free(hanum, gen.data_len+gen.sense_len, buf,
> paddr);
>                 return -EFAULT;

Agreed this looks like it should be limited, but I'm not sure what valid
limits might be.  Seems this will specify size to pci_alloc_consitent(),
and then use that buffer.

thanks,
-chris
-- 
Linux Security Modules     http://lsm.immunix.org     http://lsm.bkbits.net
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/