Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-02-02 Thread Michael Cheng
As far as I know, we haven't gotten to the arm implementation yet, since we are trying to get i915 compile for arm without using random ifdefs and dummy functions. "Noob question - why is i915 the only driver calling it? Do other GPUs never need to flush CPU cache?" Unfortunately I don't hav

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-02-01 Thread Tvrtko Ursulin
On 01/02/2022 15:41, Michael Cheng wrote: Ah, thanks for the clarification! While discussion goes on about the route you suggested, could we land these patches (after addressing the reviews) to unblock compiling i915 on arm? I am 60-40 to no, since follow up can be hard. I'd prefer a little

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-02-01 Thread Michael Cheng
Ah, thanks for the clarification! While discussion goes on about the route you suggested, could we land these patches (after addressing the reviews) to unblock compiling i915 on arm? On 2022-02-01 1:25 a.m., Tvrtko Ursulin wrote: On 31/01/2022 17:02, Michael Cheng wrote: Hey Tvrtko, Are you

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-02-01 Thread Tvrtko Ursulin
On 31/01/2022 17:02, Michael Cheng wrote: Hey Tvrtko, Are you saying when adding drm_clflush_virt_range(addr, sizeof(addr), this function forces an x86 code path only? If that is the case, drm_clflush_virt_range(addr, sizeof(addr) currently has ifdefs that seperate out x86 and powerpc, so w

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-01-31 Thread Michael Cheng
Hey Tvrtko, Are you saying when adding drm_clflush_virt_range(addr, sizeof(addr), this function forces an x86 code path only? If that is the case, drm_clflush_virt_range(addr, sizeof(addr) currently has ifdefs that seperate out x86 and powerpc, so we can add an ifdef for arm in the near futur

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-01-31 Thread Tvrtko Ursulin
On 28/01/2022 22:10, Michael Cheng wrote: Use drm_clflush_virt_range instead of clflushopt and remove the memory barrier, since drm_clflush_virt_range takes care of that. Signed-off-by: Michael Cheng --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 8 +++- 1 file changed, 3 inserti

Re: [Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-01-28 Thread Bowman, Casey G
> -Original Message- > From: Cheng, Michael > Sent: Friday, January 28, 2022 2:10 PM > To: intel-gfx@lists.freedesktop.org > Cc: Cheng, Michael ; Bowman, Casey G > ; De Marchi, Lucas > ; Boyer, Wayne ; > ville.syrj...@linux.intel.com; Kuoppala, Mika ; > Auld, Matthew > Subject: [PATCH

[Intel-gfx] [PATCH v2 4/4] drm/i915/: Re-work clflush_write32

2022-01-28 Thread Michael Cheng
Use drm_clflush_virt_range instead of clflushopt and remove the memory barrier, since drm_clflush_virt_range takes care of that. Signed-off-by: Michael Cheng --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 8 +++- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/drivers/gpu