[Intel-gfx] [PATCH 1/1] drm/i915: Disable pread/pwrite ioctl's for future platforms (v2)

[Maarten Lankhorst]

Op 2021-03-12 om 04:28 schreef Dixit, Ashutosh:
> On Thu, 11 Mar 2021 12:20:17 -0800, Jason Ekstrand wrote:
>> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
>> index b2e3b5cfccb4a..78ad5a9dd4784 100644
>> --- a/drivers/gpu/drm/i915/i915_gem.c
>> +++ b/drivers/gpu/drm/i915/i915_gem.c
>> @@ -374,10 +374,19 @@ int
>>  i915_gem_pread_ioctl(struct drm_device *dev, void *data,
>> 		     struct drm_file *file)
>>  {
>> +	struct drm_i915_private *i915 = to_i915(dev);
>> 	struct drm_i915_gem_pread *args = data;
>> 	struct drm_i915_gem_object *obj;
>> 	int ret;
>>
>> +	/* Pread is disallowed for all platforms after TGL-LP */
>> +	if (INTEL_GEN(i915) >= 12 && !IS_TIGERLAKE(i915))
>> +		return -EOPNOTSUPP;
>> +
>> +	/* All discrete memory platforms are Gen12 or above */
>> +	if (WARN_ON(HAS_LMEM(i915)))
>> +		return -EOPNOTSUPP;
> Not sure but you are probably trying to make it explicit that pread/pwrite
> are truly gone on dGfx? Because real dGfx are Gen12+ the code will return
> from the first if statement and never get to the second if statement. And
> there's talk on the relocation thread about tripping fake LMEM here for
> platforms prior to Gen12.
>
> So I'd suggest get rid of this second if statement and only retain the
> first (for both pread and pwrite) since that seems to be entirely
> sufficient.
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx

I think this should probably be a -ENODEV return code, otherwise patch looks good to me.

We probably don't want to break fake lmem until it's removed..

Cc drm maintainers on next version?