From: Stanimir Varbanov <stanimir.varbanov@linaro.org>
To: Rajendra Nayak <rnayak@codeaurora.org>,
sboyd@codeaurora.org, mturquette@baylibre.com
Cc: linux-clk@vger.kernel.org, linux-arm-msm@vger.kernel.org,
linux-kernel@vger.kernel.org, sricharan@codeaurora.org
Subject: Re: [PATCH] clk: qcom: gdsc: Fix handling of hw control enable/disable
Date: Tue, 10 Jan 2017 18:43:52 +0200 [thread overview]
Message-ID: <db0b904e-4d28-b404-802b-7c2bcf0801e9@linaro.org> (raw)
In-Reply-To: <1484027679-18397-1-git-send-email-rnayak@codeaurora.org>
Hi Rajendra,
On 01/10/2017 07:54 AM, Rajendra Nayak wrote:
> Once a gdsc is brought in and out of HW control, there is a
> power down and up cycle which can take upto 1us. Polling on
> the gdsc status immediately after the hw control enable/disable
> can mislead software/firmware to belive the gdsc is already either on
> or off, while its yet to complete the power cycle.
> To avoid this add a 1us delay post a enable/disable of HW control
> mode.
>
> Also after the HW control mode is disabled, poll on the status to
> check gdsc status reflects its 'on' before force disabling it
> in software.
>
> Reported-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
> Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
> ---
>
> Stan,
> If there was a specific issue you saw with venus because of the missing
> delay and poll, can you check if this fixes any of that.
Something more about the issue.
I had re-designed venus driver on three platform drivers venus-core,
venus-dec and venus-enc in order to be able to control those three
power-domains (VENUS_GDSC, VENUS_CORE0_GDSC and VENUS_CORE1_GDSC).
After that I abstracted MMAGIC hw on a separate mmagic driver. This
driver just controls mmagic clocks and GDSC in its runtime_suspend and
runtime_resume methods.
The DT nodes looks like:
mmagic_video {
compatible = "qcom,msm8996-mmagic";
clocks = <&rpmcc MSM8996_RPM_SMD_MMAXI_CLK>,
<&mmcc MMSS_MMAGIC_AHB_CLK>,
<&mmcc MMSS_MMAGIC_CFG_AHB_CLK>,
<&mmcc MMAGIC_VIDEO_NOC_CFG_AHB_CLK>,
<&mmcc MMSS_MMAGIC_MAXI_CLK>,
<&mmcc MMAGIC_VIDEO_AXI_CLK>,
<&mmcc SMMU_VIDEO_AHB_CLK>,
<&mmcc SMMU_VIDEO_AXI_CLK>;
power-domains = <&mmcc MMAGIC_VIDEO_GDSC>;
video-codec {
compatible = "qcom,msm8996-venus";
clocks = <&mmcc VIDEO_CORE_CLK>,
<&mmcc VIDEO_AHB_CLK>,
<&mmcc VIDEO_AXI_CLK>,
<&mmcc VIDEO_MAXI_CLK>;
power-domains = <&mmcc VENUS_GDSC>;
...
video-decoder {
compatible = "venus-decoder";
clocks = "subcore0";
clock-names = <&mmcc VIDEO_SUBCORE0_CLK>;
power-domains = <&mmcc VENUS_CORE0_GDSC>;
};
video-encoder {
compatible = "venus-encoder";
clocks = "subcore1";
clock-names = <&mmcc VIDEO_SUBCORE1_CLK>;
power-domains = <&mmcc VENUS_CORE1_GDSC>;
};
};
};
Note that mmagic_video is a parent dt node for smmu_video DT node so
that clocks and mmagic_video gdsc will be enabled once smmu driver is
instantiated by venus-core diriver.
Now when video-dec driver calls pm_runtime_get_sync() the sequence of
enabling is:
MMAGIC_VIDEO_GDSC -> MMAGIC clocks and SMMU clocks -> VENUS_GDSC ->
VIDEO clocks -> VENUS_CORE0_GDSC -> VIDEO subcore0 clock
When video-dec platform driver calls pm_runtime_put_sync() we should
disabling of GDSC and clocks in the reversed oder.
The issue comes when I have ran video decoder, the decoder hw finish
stream decoding and we want to suspend venus core. The issue is that
when I start disabling SMMU_VIDEO_AXI_CLK and/or MMAGIC_VIDEO_AXI_CLK
the system reboots.
I have added a delay (200ms) before disabling mmagic clocks and then
everything is fine again.
Any idea?
--
regards,
Stan
next prev parent reply other threads:[~2017-01-10 16:43 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-01-10 5:54 [PATCH] clk: qcom: gdsc: Fix handling of hw control enable/disable Rajendra Nayak
2017-01-10 9:31 ` Stanimir Varbanov
2017-01-20 23:20 ` Stephen Boyd
2017-01-23 4:03 ` Rajendra Nayak
2017-01-23 4:06 ` Rajendra Nayak
2017-01-10 16:43 ` Stanimir Varbanov [this message]
2017-01-10 19:29 ` Sricharan
2017-01-11 8:55 ` Stanimir Varbanov
2017-01-12 8:47 ` Sricharan
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=db0b904e-4d28-b404-802b-7c2bcf0801e9@linaro.org \
--to=stanimir.varbanov@linaro.org \
--cc=linux-arm-msm@vger.kernel.org \
--cc=linux-clk@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mturquette@baylibre.com \
--cc=rnayak@codeaurora.org \
--cc=sboyd@codeaurora.org \
--cc=sricharan@codeaurora.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;
as well as URLs for NNTP newsgroup(s).