From: Jordan Crouse <jcrouse-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
To: Sharat Masetty <smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
Cc: linux-arm-msm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
freedreno-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org,
dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Subject: Re: [PATCH 2/3] drm/msm/a6xx: Move power counter selectable to resume()
Date: Wed, 17 Oct 2018 08:09:43 -0600 [thread overview]
Message-ID: <20181017140943.GC4751@jcrouse-lnx.qualcomm.com> (raw)
In-Reply-To: <1539781441-13076-3-git-send-email-smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
On Wed, Oct 17, 2018 at 06:34:00PM +0530, Sharat Masetty wrote:
> Moving this to resume helps with both GPU DCVS and the performance
> sampling. For GPU DCVS this makes sure that the frequency does not scale
> when there are no GPU submissions. In the case of performance profiling,
> the GPU is UP, but its possible that the hw_init() was not called yet,
> so we need this to get sane perf values in all cases.
This description is worded oddly to me but I think the gist is that you
want the counters running before you start sampling. I agree which is why I
think that the /* FIXME: */ is right and this should go into in the GMU code
(a6xx_rpmh_start to be exact). The timing should still work out and the GMU code
can stay with the GMU code which is always a bonus.
> Signed-off-by: Sharat Masetty <smasetty@codeaurora.org>
> ---
> drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 16 ++++++++--------
> 1 file changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
> index c0cd3ac..2c52b7c 100644
> --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
> @@ -387,14 +387,6 @@ static int a6xx_hw_init(struct msm_gpu *gpu)
> /* Select CP0 to always count cycles */
> gpu_write(gpu, REG_A6XX_CP_PERFCTR_CP_SEL_0, PERF_CP_ALWAYS_COUNT);
>
> - /* FIXME: not sure if this should live here or in a6xx_gmu.c */
> - gmu_write(&a6xx_gpu->gmu, REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_MASK,
> - 0xff000000);
> - gmu_rmw(&a6xx_gpu->gmu, REG_A6XX_GMU_CX_GMU_POWER_COUNTER_SELECT_0,
> - 0xff, 0x20);
> - gmu_write(&a6xx_gpu->gmu, REG_A6XX_GMU_CX_GMU_POWER_COUNTER_ENABLE,
> - 0x01);
> -
> gpu_write(gpu, REG_A6XX_RB_NC_MODE_CNTL, 2 << 1);
> gpu_write(gpu, REG_A6XX_TPL1_NC_MODE_CNTL, 2 << 1);
> gpu_write(gpu, REG_A6XX_SP_NC_MODE_CNTL, 2 << 1);
> @@ -655,6 +647,14 @@ static int a6xx_pm_resume(struct msm_gpu *gpu)
>
> gpu->needs_hw_init = true;
>
> + /* FIXME: not sure if this should live here or in a6xx_gmu.c */
If in case you don't move this, this line needs to go away.
> + gmu_write(&a6xx_gpu->gmu, REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_MASK,
> + 0xff000000);
> + gmu_rmw(&a6xx_gpu->gmu, REG_A6XX_GMU_CX_GMU_POWER_COUNTER_SELECT_0,
> + 0xff, 0x20);
> + gmu_write(&a6xx_gpu->gmu, REG_A6XX_GMU_CX_GMU_POWER_COUNTER_ENABLE,
> + 0x01);
> +
> msm_gpu_resume_devfreq(gpu);
>
> return ret;
> --
> 1.9.1
>
--
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno
next prev parent reply other threads:[~2018-10-17 14:09 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-10-17 13:03 [PATCH 0/3] drm/msm: Revive GPU workload profiling Sharat Masetty
[not found] ` <1539781441-13076-1-git-send-email-smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-10-17 13:03 ` [PATCH 1/3] drm/msm: Change gpu_busy() function Sharat Masetty
[not found] ` <1539781441-13076-2-git-send-email-smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-10-17 14:39 ` Jordan Crouse
2018-10-17 13:04 ` [PATCH 2/3] drm/msm/a6xx: Move power counter selectable to resume() Sharat Masetty
[not found] ` <1539781441-13076-3-git-send-email-smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-10-17 14:09 ` Jordan Crouse [this message]
2018-10-17 13:04 ` [PATCH 3/3] drm/msm: Use Hardware counters for perf profiling Sharat Masetty
[not found] ` <1539781441-13076-4-git-send-email-smasetty-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-10-17 14:35 ` Jordan Crouse
[not found] ` <20181017143548.GD4751-9PYrDHPZ2Orvke4nUoYGnHL1okKdlPRT@public.gmane.org>
2018-10-26 13:46 ` Sharat Masetty
[not found] ` <f57bf8b2-c44a-68f5-a051-2a17c1ac5704-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-11-28 20:09 ` Rob Clark
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=20181017140943.GC4751@jcrouse-lnx.qualcomm.com \
--to=jcrouse-sgv2jx0feol9jmxxk+q4oq@public.gmane.org \
--cc=dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=freedreno-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=linux-arm-msm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=smasetty-sgV2jX0FEOL9JmXXK+q4OQ@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