linux-kernel - Re: [PATCH v2] PCI: qcom: Ensure that PERST is asserted for at least 100 ms

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20190530154926.GE13993@redmoon>
Date:   Thu, 30 May 2019 16:49:26 +0100
From:   Lorenzo Pieralisi <lorenzo.pieralisi@....com>
To:     Bjorn Helgaas <helgaas@...nel.org>
Cc:     Niklas Cassel <niklas.cassel@...aro.org>,
        Stanimir Varbanov <svarbanov@...sol.com>,
        Andy Gross <agross@...nel.org>,
        David Brown <david.brown@...aro.org>, stable@...r.kernel.org,
        linux-pci@...r.kernel.org, linux-arm-msm@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] PCI: qcom: Ensure that PERST is asserted for at least
 100 ms

On Wed, May 29, 2019 at 07:52:32AM -0500, Bjorn Helgaas wrote:
> On Wed, May 29, 2019 at 11:43:52AM +0200, Niklas Cassel wrote:
> > Currently, there is only a 1 ms sleep after asserting PERST.
> > 
> > Reading the datasheets for different endpoints, some require PERST to be
> > asserted for 10 ms in order for the endpoint to perform a reset, others
> > require it to be asserted for 50 ms.
> > 
> > Several SoCs using this driver uses PCIe Mini Card, where we don't know
> > what endpoint will be plugged in.
> 
> I think this patch is absolutely the right thing to do.  We don't know
> what might be plugged in, so we should support arbitrary endpoints.
> 
> One could imagine some sort of DT property to identify closed
> environments where there is no need to support arbitrary endpoints and
> it might be desirable to reduce the delay.  But that should wait
> until there's a need for it and if the need *does* arise, hopefully
> the definition of such a property could be generic across all SoCs.
> 
> > The PCI Express Card Electromechanical Specification specifies:
> > "On power up, the deassertion of PERST# is delayed 100 ms (TPVPERL) from
> > the power rails achieving specified operating limits."
> 
> It'd be nice to include the complete citation, i.e., "PCI Express Card
> Electromechanical Specification r2.0, sec 2.2"

I will add it and merge the patch.

Thanks,
Lorenzo

> > Add a sleep of 100 ms before deasserting PERST, in order to ensure that
> > we are compliant with the spec.
> > 
> > Fixes: 82a823833f4e ("PCI: qcom: Add Qualcomm PCIe controller driver")
> > Signed-off-by: Niklas Cassel <niklas.cassel@...aro.org>
> > Acked-by: Stanimir Varbanov <svarbanov@...sol.com>
> > Cc: stable@...r.kernel.org # 4.5+
> > ---
> > Changes since v1:
> > Move the sleep into qcom_ep_reset_deassert()
> > 
> >  drivers/pci/controller/dwc/pcie-qcom.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> > index 0ed235d560e3..5d1713069d14 100644
> > --- a/drivers/pci/controller/dwc/pcie-qcom.c
> > +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> > @@ -178,6 +178,8 @@ static void qcom_ep_reset_assert(struct qcom_pcie *pcie)
> >  
> >  static void qcom_ep_reset_deassert(struct qcom_pcie *pcie)
> >  {
> > +	/* Ensure that PERST has been asserted for at least 100 ms */
> > +	msleep(100);
> >  	gpiod_set_value_cansleep(pcie->reset, 0);
> >  	usleep_range(PERST_DELAY_US, PERST_DELAY_US + 500);
> >  }
> > -- 
> > 2.21.0
> >