Re: [igt-dev] [PATCH i-g-t 1/2] lib: intel: report GT size from the database if available

[Lionel Landwerlin <lionel.g.landwerlin@intel.com>]

On 14/05/18 16:41, Chris Wilson wrote:
> Quoting Lionel Landwerlin (2018-05-14 14:15:01)
>> Ping?
>>
>> On 20/03/18 13:47, Lionel Landwerlin wrote:
>>> The scheme if getting the GT size from the PCI-id doesn't work anymore
>>> on Coffeelake.
>>>
>>> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
>>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105451, https://bugs.freedesktop.org/show_bug.cgi?id=101740
>>> ---
>>>    lib/intel_device_info.c | 9 +++++++++
>>>    1 file changed, 9 insertions(+)
>>>
>>> diff --git a/lib/intel_device_info.c b/lib/intel_device_info.c
>>> index 1c710733..c104c9c4 100644
>>> --- a/lib/intel_device_info.c
>>> +++ b/lib/intel_device_info.c
>>> @@ -410,8 +410,17 @@ unsigned intel_gen(uint16_t devid)
>>>     */
>>>    unsigned intel_gt(uint16_t devid)
>>>    {
>>> +     const struct intel_device_info *devinfo = intel_get_device_info(devid);
>>>        unsigned mask = intel_gen(devid);
>>>    
>>> +     /* If in the database, just use that information. */
>>> +     if (devinfo->gt != 0)
>>> +             return devinfo->gt - 1;
>>> +
>>> +     /*
>>> +      * This scheme doesn't work on Coffeelake, we should probably
>>> +      * not rely on this anymore.
>>> +      */
> I don't get this comment in light of the above. If you are adding a
> preferred mechanism, why complain about the stuff you haven't convert?
> -Chris
>
I only added gt information on hsw+. So a number of devices won't have 
the information.
I do need to add it for CNL+ too, then this comment could be turn into 
an assert(gen < 8 && !haswell)

_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev