All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Koenig, Christian" <Christian.Koenig-5C7GfCeVMHo@public.gmane.org>
To: "Kazlauskas,
	Nicholas" <Nicholas.Kazlauskas-5C7GfCeVMHo@public.gmane.org>,
	Alex Deucher
	<alexdeucher-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: amd-gfx list <amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org>
Subject: Re: [PATCH] drm/amdgpu: Clear VRAM for DRM dumb_create buffers
Date: Mon, 11 Mar 2019 17:31:26 +0000	[thread overview]
Message-ID: <e021eacf-8200-fe28-6aad-0bfbcb27798c@amd.com> (raw)
In-Reply-To: <a5aa3caf-253e-8832-c63e-77269d4e865f-5C7GfCeVMHo@public.gmane.org>

Am 11.03.19 um 18:04 schrieb Kazlauskas, Nicholas:
> On 3/11/19 12:58 PM, Christian König wrote:
>> Am 08.03.19 um 17:47 schrieb Alex Deucher:
>>> On Fri, Mar 8, 2019 at 10:38 AM Nicholas Kazlauskas
>>> <nicholas.kazlauskas@amd.com> wrote:
>>>> The dumb_create API isn't intended for high performance rendering
>>>> and it's more useful for userspace (ie. IGT) to have them precleared.
>>>>
>>>> The bonus here is that we also won't needlessly leak whatever was
>>>> previously in VRAM, but it also probably wasn't sensitive if it was
>>>> going through this API.
>>>>
>>>> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
>>> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
>> NAK, IIRC we used to have this bit here before.
>>
>> In general I agree that this would be a good idea, but the problem is
>> the that first dump buffer is sometimes created before the engine
>> (usually SDMA) to clear VRAM is initialized.
>>
>> So you can run into a nice crash on suspend/resume.
>>
>> Christian.
> FWIW I never hit it myself during any suspend/resume tests I did but
> I'll take your word for it.
>
> How about doing a map/memset/unmap immediately after buffer creation
> here? It'd be much slower, but it's also what we'd be doing in userspace
> anyway.

And history repeat itself :) Yeah, that was also suggested before and 
didn't worked either for some reason.

But that was ~2 years ago and I really don't remember the details.

What could work is maybe to add a check if the engine is available if 
the flag is given and ignore it when it isn't.

Christian.

>
> Nicholas Kazlauskas
>
>
>>>> ---
>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 3 ++-
>>>>    1 file changed, 2 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>>> index fcaaac30e84b..a58072bbc9b8 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>>> @@ -743,7 +743,8 @@ int amdgpu_mode_dumb_create(struct drm_file
>>>> *file_priv,
>>>>           domain = amdgpu_bo_get_preferred_pin_domain(adev,
>>>>                                   
>>>> amdgpu_display_supported_domains(adev));
>>>>           r = amdgpu_gem_object_create(adev, args->size, 0, domain,
>>>> -
>>>> AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
>>>> +
>>>> AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
>>>> +                                    AMDGPU_GEM_CREATE_VRAM_CLEARED,
>>>>                                        ttm_bo_type_device, NULL, &gobj);
>>>>           if (r)
>>>>                   return -ENOMEM;
>>>> -- 
>>>> 2.17.1
>>>>
>>>> _______________________________________________
>>>> amd-gfx mailing list
>>>> amd-gfx@lists.freedesktop.org
>>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>>> _______________________________________________
>>> amd-gfx mailing list
>>> amd-gfx@lists.freedesktop.org
>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  parent reply	other threads:[~2019-03-11 17:31 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-03-08 15:38 [PATCH] drm/amdgpu: Clear VRAM for DRM dumb_create buffers Nicholas Kazlauskas
     [not found] ` <20190308153816.7978-1-nicholas.kazlauskas-5C7GfCeVMHo@public.gmane.org>
2019-03-08 16:47   ` Alex Deucher
     [not found]     ` <CADnq5_PxyfDKVLgQ9iBqEsSGX-jYMbr-RsxrvM5j=0N7AF-w5Q-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2019-03-11 16:58       ` Christian König
     [not found]         ` <73f25762-0277-7e12-dfff-a016d245478c-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2019-03-11 17:04           ` Kazlauskas, Nicholas
     [not found]             ` <a5aa3caf-253e-8832-c63e-77269d4e865f-5C7GfCeVMHo@public.gmane.org>
2019-03-11 17:31               ` Koenig, Christian [this message]
     [not found]                 ` <e021eacf-8200-fe28-6aad-0bfbcb27798c-5C7GfCeVMHo@public.gmane.org>
2019-03-11 17:53                   ` Kazlauskas, Nicholas
     [not found]                     ` <7e271708-34ad-db55-54ca-eaccd954eff6-5C7GfCeVMHo@public.gmane.org>
2019-03-11 18:04                       ` Christian König

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e021eacf-8200-fe28-6aad-0bfbcb27798c@amd.com \
    --to=christian.koenig-5c7gfcevmho@public.gmane.org \
    --cc=Nicholas.Kazlauskas-5C7GfCeVMHo@public.gmane.org \
    --cc=alexdeucher-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org \
    --cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.