On 13/03/14 01:05, Ilia Mirkin wrote:
[snip]
>
> Not really. drm->drm_version will be 0 if ver fails.
>
Indeed, dev is calloc'ated by its callers, and if they mess around with
that's their own fault.
Sorry for the noise.
-Emil
On 13/03/14 00:45, Ilia Mirkin wrote:
> On Wed, Mar 12, 2014 at 4:45 PM, Emil Velikov
> wrote:
>> In theory it's possible for any of the nouveau_getparam calls to
>> fail whist the last one being successful.
>>
>> Thus at least one of the following (hard requirements) drmVersion,
>> chipset and v
On Wed, Mar 12, 2014 at 9:24 PM, Emil Velikov
wrote:
> On 13/03/14 01:05, Ilia Mirkin wrote:
> [snip]
>
>>
>> Not really. drm->drm_version will be 0 if ver fails.
>>
> Indeed, dev is calloc'ated by its callers, and if they mess around with
> that's their own fault.
The callers? It's calloc'd ins
On Wed, Mar 12, 2014 at 9:04 PM, Emil Velikov
wrote:
> On 13/03/14 00:45, Ilia Mirkin wrote:
>>
>> On Wed, Mar 12, 2014 at 4:45 PM, Emil Velikov
>> wrote:
>>>
>>> In theory it's possible for any of the nouveau_getparam calls to
>>> fail whist the last one being successful.
>>>
>>> Thus at least
In theory it's possible for any of the nouveau_getparam calls to
fail whist the last one being successful.
Thus at least one of the following (hard requirements) drmVersion,
chipset and vram/gart memory size will be filled with garbage and
sent to the userspace drivers.
Signed-off-by: Emil Veliko
On Wed, Mar 12, 2014 at 4:45 PM, Emil Velikov
wrote:
> In theory it's possible for any of the nouveau_getparam calls to
> fail whist the last one being successful.
>
> Thus at least one of the following (hard requirements) drmVersion,
> chipset and vram/gart memory size will be filled with garbag
