Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:1 as permitted sender) client-ip=2620:137:e000::3:1;
Message-ID: <ac00b49505e926ac4f3d0d0ba315b366.sboyd@kernel.org>
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
In-Reply-To: <b8fa17a7-296e-4e17-9438-a8be04dbd476@quicinc.com>
References: <20231017131944.8356-1-quic_kriskura@quicinc.com> <e38ae24c09ef321c224a6b6aeb97e73b.sboyd@kernel.org> <b8fa17a7-296e-4e17-9438-a8be04dbd476@quicinc.com>
Subject: Re: [RFC 4/8] clk: qcom: gcc-sm8450: Keep usb30 prim gdsc on during runtime suspend
From:   Stephen Boyd <sboyd@kernel.org>
Cc:     linux-clk@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-arm-msm@vger.kernel.org, quic_ppratap@quicinc.com,
        quic_jackp@quicinc.com, Udipto Goswami <quic_ugoswami@quicinc.com>
To:     Andy Gross <agross@kernel.org>,
        Bjorn Andersson <andersson@kernel.org>,
        Konrad Dybcio <konrad.dybcio@linaro.org>,
        Krishna Kurapati PSSNV <quic_kriskura@quicinc.com>,
        Michael Turquette <mturquette@baylibre.com>
Date:   Thu, 19 Oct 2023 17:05:54 -0700
User-Agent: alot/0.10
Precedence: bulk

Quoting Krishna Kurapati PSSNV (2023-10-19 01:37:20)
>=20
>=20
> On 10/19/2023 6:28 AM, Stephen Boyd wrote:
> > Quoting Krishna Kurapati (2023-10-17 06:19:44)
> >> When runtime suspend happens in host mode, if cable is removed at this
> >> instant, it triggers a resume where a xhci register access is done when
> >> gdsc is not on leading to a crash. Keep usb30 prim gdsc as on during
> >> runtime suspend to avoid this crash.
> >>
> >> Signed-off-by: Krishna Kurapati <quic_kriskura@quicinc.com>
> >> ---
> >> Note: I am not sure if this is the right way or setting GDSC flags to
> >> PWRSTS_RET_ON is the right approach. So for now made this change to
> >=20
> > Does using PWRSTS_RET_ON work? Does the GDSC support being turned off or
> > is the lowest power state retention?
>=20
> I didn't see any issue when I put it as PWRSTS_RET_ON. If the GDSC is=20
> off, when we remove cable in host mode, we see a crash because of xhci=20
> register access. But when I try putting PWRSTS_RET_ON /=20
> CLK_RPM_ALWAYS_ON it doesn't crash.
>=20

You need to find out if the GDSC only supports retention mode. Otherwise
it sounds like the xhci register access is happening when the genpd is
off, and thus the driver needs to power on the device before accessing
it.