AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: James Zhu <jamesz-5C7GfCeVMHo@public.gmane.org>
To: "Zhu, James" <James.Zhu-5C7GfCeVMHo@public.gmane.org>,
	Alex Deucher
	<alexdeucher-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: "Shamis, Leonid" <Leonid.Shamis-5C7GfCeVMHo@public.gmane.org>,
	"Gabra, Maroun" <Maroun.Gabra-5C7GfCeVMHo@public.gmane.org>,
	"amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org"
	<amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org>,
	"Liu, Shaoyun" <Shaoyun.Liu-5C7GfCeVMHo@public.gmane.org>
Subject: Re: [PATCH] drm/amdgpu: Add GDS clearing workaround in later init for gfx9
Date: Fri, 7 Jun 2019 19:42:07 +0000	[thread overview]
Message-ID: <65266293-ac7a-c3ef-752b-1691299b1f28@amd.com> (raw)
In-Reply-To: <19203faa-970e-732b-a0e0-d2d1fe69ce30-5C7GfCeVMHo@public.gmane.org>


On 2019-06-07 3:12 p.m., Zhu, James wrote:
> On 2019-06-07 2:16 p.m., Alex Deucher wrote:
>> On Fri, Jun 7, 2019 at 12:38 PM Zhu, James <James.Zhu@amd.com> wrote:
>>> Since Hardware bug, GDS exist ECC error after cold boot up,
>>> adding GDS clearing workaround in later init for gfx9.
>>>
>>> Signed-off-by: James Zhu <James.Zhu@amd.com>
>>> ---
>>>    drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 48 +++++++++++++++++++++++++++++++++++
>>>    1 file changed, 48 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
>>> index 76722fc..81f6ba8 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
>>> @@ -3634,6 +3634,50 @@ static const struct soc15_reg_entry sec_ded_counter_registers[] = {
>>>       { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6},
>>>    };
>>>
>>> +
>>> +static int gfx_v9_0_do_edc_gds_workarounds(struct amdgpu_device *adev)
>>> +{
>>> +       struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
>>> +       int r;
>>> +
>>> +       r = amdgpu_ring_alloc(ring, 17);
>>> +       if (r) {
>>> +               DRM_ERROR("amdgpu: GDS workarounds failed to lock ring %s (%d).\n",
>>> +                       ring->name, r);
>>> +               return r;
>>> +       }
>>> +
>>> +       amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
>>> +       amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(0) |
>>> +                          WRITE_DATA_DST_SEL(0));
>>> +       amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE));
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, 0x10000);
>> hardcoded size, please use the size from the driver.
>>
>>> +
>>> +       amdgpu_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
>>> +       amdgpu_ring_write(ring, (PACKET3_DMA_DATA_CP_SYNC |
>>> +                               PACKET3_DMA_DATA_DST_SEL(1) |
>>> +                               PACKET3_DMA_DATA_SRC_SEL(2) |
>>> +                               PACKET3_DMA_DATA_ENGINE(0)));
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, PACKET3_DMA_DATA_CMD_RAW_WAIT | 0x10000);
>> Instead of hardcoding the size, can you use the gds size from the
>> driver (adev->gds.gds_size).
> Hi Alex,
>
> Do you mean adev->gds.mem.total_size?
>
> But I see below operation in gfx_v9_0_ngg_init.
>
> adev->gds.mem.total_size -= ....
>
> Or you want me to add gds_size in struct amdgpu_gds?
>
> James

Yeah, The amd-staging-drm-next branch has adev->gds.gds_size.

James

>
>> With that fixed:
>> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
>>
>>> +
>>> +       amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
>>> +       amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(0) |
>>> +                               WRITE_DATA_DST_SEL(0));
>>> +       amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE));
>>> +       amdgpu_ring_write(ring, 0);
>>> +       amdgpu_ring_write(ring, 0x0);
>>> +
>>> +       amdgpu_ring_commit(ring);
>>> +
>>> +       return 0;
>>> +}
>>> +
>>> +
>>>    static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
>>>    {
>>>           struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
>>> @@ -3810,6 +3854,10 @@ static int gfx_v9_0_ecc_late_init(void *handle)
>>>                   return 0;
>>>           }
>>>
>>> +       r = gfx_v9_0_do_edc_gds_workarounds(adev);
>>> +       if (r)
>>> +               return r;
>>> +
>>>           /* requires IBs so do in late init after IB pool is initialized */
>>>           r = gfx_v9_0_do_edc_gpr_workarounds(adev);
>>>           if (r)
>>> --
>>> 2.7.4
>>>
>>> _______________________________________________
>>> 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:[~2019-06-07 19:42 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-07 16:38 [PATCH] drm/amdgpu: Add GDS clearing workaround in later init for gfx9 Zhu, James
     [not found] ` <1559925472-20077-1-git-send-email-James.Zhu-5C7GfCeVMHo@public.gmane.org>
2019-06-07 18:16   ` Alex Deucher
     [not found]     ` <CADnq5_NBWcUtXsBbQP5uBvaOFgWXw3Lhj=ZOisVyTctqZph=xQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2019-06-07 19:12       ` James Zhu
     [not found]         ` <19203faa-970e-732b-a0e0-d2d1fe69ce30-5C7GfCeVMHo@public.gmane.org>
2019-06-07 19:42           ` James Zhu [this message]
     [not found]             ` <65266293-ac7a-c3ef-752b-1691299b1f28-5C7GfCeVMHo@public.gmane.org>
2019-06-07 20:16               ` Shamis, Leonid
     [not found]                 ` <3598A781-7AF5-4DDE-83AE-8835E85573D3-5C7GfCeVMHo@public.gmane.org>
2019-06-08 12:50                   ` Christian König
     [not found]                     ` <3ab2a128-3433-0e45-b555-2aff0892f8e6-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2019-06-10 14:21                       ` James Zhu
2019-06-10 14:16                   ` James Zhu

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=65266293-ac7a-c3ef-752b-1691299b1f28@amd.com \
    --to=jamesz-5c7gfcevmho@public.gmane.org \
    --cc=James.Zhu-5C7GfCeVMHo@public.gmane.org \
    --cc=Leonid.Shamis-5C7GfCeVMHo@public.gmane.org \
    --cc=Maroun.Gabra-5C7GfCeVMHo@public.gmane.org \
    --cc=Shaoyun.Liu-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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox