Re: [PATCH v1 1/1] drm/i915/gt: Move wbvind_on_all_cpus #define

[Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>]

On 11/02/2022 13:33, Jani Nikula wrote:
> On Thu, 10 Feb 2022, Michael Cheng <michael.cheng@intel.com> wrote:
>> Move wbvind_on_all_cpus to intel_gt.h. This will allow other wbind_on_all_cpus
>> calls to benefit from the #define logic, and prevent compiler errors
>> when building for non-x86 architectures.
>>
>> Signed-off-by: Michael Cheng <michael.cheng@intel.com>
>> ---
>>   drivers/gpu/drm/i915/gem/i915_gem_pm.c | 7 -------
>>   drivers/gpu/drm/i915/gt/intel_gt.h     | 7 +++++++
>>   2 files changed, 7 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pm.c b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
>> index 6da68b38f00f..ff7340ae5ac8 100644
>> --- a/drivers/gpu/drm/i915/gem/i915_gem_pm.c
>> +++ b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
>> @@ -12,13 +12,6 @@
>>   
>>   #include "i915_drv.h"
>>   
>> -#if defined(CONFIG_X86)
>> -#include <asm/smp.h>
>> -#else
>> -#define wbinvd_on_all_cpus() \
>> -	pr_warn(DRIVER_NAME ": Missing cache flush in %s\n", __func__)
>> -#endif
>> -
>>   void i915_gem_suspend(struct drm_i915_private *i915)
>>   {
>>   	GEM_TRACE("%s\n", dev_name(i915->drm.dev));
>> diff --git a/drivers/gpu/drm/i915/gt/intel_gt.h b/drivers/gpu/drm/i915/gt/intel_gt.h
>> index 2dad46c3eff2..149e8c13e402 100644
>> --- a/drivers/gpu/drm/i915/gt/intel_gt.h
>> +++ b/drivers/gpu/drm/i915/gt/intel_gt.h
>> @@ -10,6 +10,13 @@
>>   #include "intel_gt_types.h"
>>   #include "intel_reset.h"
>>   
>> +#if defined(CONFIG_X86)
>> +#include <asm/smp.h>
>> +#else
>> +#define wbinvd_on_all_cpus() \
>> +         pr_warn(DRIVER_NAME ": Missing cache flush in %s\n", __func__)
>> +#endif
> 
> Don't include headers from headers if it can be avoided.
> 
> gt/intel_gt.h is included from 79 files. We don't want all of them to
> include <asm/smp.h> when only 3 files actually need
> wbinvd_on_all_cpus().
> 
> Also, gt/intel_gt.h has absolutely nothing to do with
> wbinvd_on_all_cpus() or asm/smp.h. Please don't use topical headers as
> dumping grounds for random things.
> 
> Maybe a better idea is to add a local wrapper for wbinvd_on_all_cpus()
> that does the right thing. Or add the above in a dedicated header.

+1, noting the naming angle:

WBINVD — Write Back and Invalidate Cache

Is an x86 instruction. Also interesting comment:

          * XXX: Consider doing a vmap flush or something, where possible.
          * Currently we just do a heavy handed wbinvd_on_all_cpus() here since
          * the underlying sg_table might not even point to struct pages, so we
          * can't just call drm_clflush_sg or similar, like we do elsewhere in
          * the driver.
          */
         if (i915_gem_object_can_bypass_llc(obj) ||
             (!HAS_LLC(i915) && !IS_DG1(i915)))
                 wbinvd_on_all_cpus();

The two together to me sound like the fix is to either find an equivalent existing platform agnostic API in the kernel, or if it does not exist create one and name it generically.

Either per-platform i915, if we go for my proposal, or I guess drm_cache.c if we don't.

Regards,

Tvrtko