From: Mika Westerberg <mika.westerberg@linux.intel.com>
To: Bjorn Helgaas <helgaas@kernel.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>,
Len Brown <lenb@kernel.org>, Lukas Wunner <lukas@wunner.de>,
Keith Busch <keith.busch@intel.com>,
Alex Williamson <alex.williamson@redhat.com>,
Alexandru Gagniuc <mr.nuke.me@gmail.com>,
Kai-Heng Feng <kai.heng.feng@canonical.com>,
Matthias Andree <matthias.andree@gmx.de>,
Paul Menzel <pmenzel@molgen.mpg.de>,
Nicholas Johnson <nicholas.johnson-opensource@outlook.com.au>,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 2/2] PCI: Add missing link delays required by the PCIe spec
Date: Wed, 30 Oct 2019 13:33:53 +0200 [thread overview]
Message-ID: <20191030113353.GY2593@lahna.fi.intel.com> (raw)
In-Reply-To: <20191029205456.GA100782@google.com>
On Tue, Oct 29, 2019 at 03:54:56PM -0500, Bjorn Helgaas wrote:
> On Fri, Oct 04, 2019 at 03:39:47PM +0300, Mika Westerberg wrote:
> > Currently Linux does not follow PCIe spec regarding the required delays
> > after reset. A concrete example is a Thunderbolt add-in-card that
> > consists of a PCIe switch and two PCIe endpoints:
> > ...
>
> > @@ -1025,15 +1025,11 @@ static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
> > if (state == PCI_D0) {
> > pci_platform_power_transition(dev, PCI_D0);
> > /*
> > - * Mandatory power management transition delays, see
> > - * PCI Express Base Specification Revision 2.0 Section
> > - * 6.6.1: Conventional Reset. Do not delay for
> > - * devices powered on/off by corresponding bridge,
> > - * because have already delayed for the bridge.
> > + * Mandatory power management transition delays are handled
> > + * in pci_pm_runtime_resume() of the corresponding
> > + * downstream/root port.
> > */
> > if (dev->runtime_d3cold) {
> > - if (dev->d3cold_delay && !dev->imm_ready)
> > - msleep(dev->d3cold_delay);
>
> This removes the only use of d3cold_delay. I assume that's
> intentional? If we no longer need it, we might as well remove it from
> the pci_dev and remove the places that set it. It'd be nice if that
> could be a separate patch, even if we waited a little longer than
> necessary at that one bisection point.
Yes, it is intentional. In the previous version I had function
pcie_get_downstream_delay() that used both d3cold_delay and imm_ready to
calculate the downstream device delay but you said:
I'm not sold on the idea that this delay depends on what's *below* the
bridge. We're using sec 6.6.1 to justify the delay, and that section
doesn't say anything about downstream devices.
So I dropped it and use 100 ms instead.
Now that you mention, I think if we want to continue support that _DSM,
we should still take d3cold_delay into account in this patch. There is
also one driver (drivers/mfd/intel-lpss-pci.c) that sets it to 0.
> It also removes one of the three uses of imm_ready, leaving only the
> two in FLR. I suspect there are other places we should use imm_ready,
> e.g., transitions to/from D1 and D2, but that would be beyond the
> scope of this patch.
Right, I think imm_ready does not apply here. If I understand correctly
it is exactly for D1, D2 and D3hot transitions which we should take into
account in pci_dev_d3_sleep() (which we don't do right now).
> > + /*
> > + * For PCIe downstream and root ports that do not support speeds
> > + * greater than 5 GT/s need to wait minimum 100 ms. For higher
> > + * speeds (gen3) we need to wait first for the data link layer to
> > + * become active.
> > + *
> > + * However, 100 ms is the minimum and the PCIe spec says the
> > + * software must allow at least 1s before it can determine that the
> > + * device that did not respond is a broken device. There is
> > + * evidence that 100 ms is not always enough, for example certain
> > + * Titan Ridge xHCI controller does not always respond to
> > + * configuration requests if we only wait for 100 ms (see
> > + * https://bugzilla.kernel.org/show_bug.cgi?id=203885).
> > + *
> > + * Therefore we wait for 100 ms and check for the device presence.
> > + * If it is still not present give it an additional 100 ms.
> > + */
> > + if (pci_pcie_type(dev) != PCI_EXP_TYPE_ROOT_PORT &&
> > + pci_pcie_type(dev) != PCI_EXP_TYPE_DOWNSTREAM)
> > + return;
>
> Shouldn't this be:
>
> if (!pcie_downstream_port(dev))
> return
>
> so we include PCI/PCI-X to PCIe bridges?
Yes, I'll change it in v3.
next prev parent reply other threads:[~2019-10-30 11:33 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-04 12:39 [PATCH v2 0/2] PCI: Add missing link delays Mika Westerberg
2019-10-04 12:39 ` [PATCH v2 1/2] PCI: Introduce pcie_wait_for_link_delay() Mika Westerberg
2020-08-08 20:22 ` Marc MERLIN
2020-08-08 20:23 ` Marc MERLIN
2020-08-09 16:31 ` Marc MERLIN
2020-09-06 18:18 ` pcieport 0000:00:01.0: PME: Spurious native interrupt (nvidia with nouveau and thunderbolt on thinkpad P73) Marc MERLIN
2020-09-06 18:26 ` Matthias Andree
2020-09-07 19:14 ` [Nouveau] " Karol Herbst
2020-09-07 20:58 ` Marc MERLIN
2020-09-07 23:51 ` Karol Herbst
2020-09-08 0:29 ` Marc MERLIN
[not found] ` <20200529180315.GA18804@merlins.org>
2020-12-26 11:12 ` 5.9.11 still hanging 2mn at each boot and looping on nvidia-gpu 0000:01:00.3: PME# enabled (Quadro RTX 4000 Mobile) Marc MERLIN
2020-12-27 18:28 ` [Nouveau] " Ilia Mirkin
2021-01-27 21:33 ` Bjorn Helgaas
2021-01-28 20:59 ` Bjorn Helgaas
2021-01-29 0:56 ` Marc MERLIN
2021-01-29 21:20 ` Bjorn Helgaas
2021-01-30 2:04 ` Marc MERLIN
2020-12-29 15:51 ` Marc MERLIN
2020-12-29 16:33 ` Ilia Mirkin
2019-10-04 12:39 ` [PATCH v2 2/2] PCI: Add missing link delays required by the PCIe spec Mika Westerberg
2019-10-26 14:19 ` Bjorn Helgaas
2019-10-28 11:28 ` Mika Westerberg
2019-10-28 13:42 ` Bjorn Helgaas
2019-10-28 18:06 ` Mika Westerberg
2019-10-28 20:16 ` Bjorn Helgaas
2019-10-29 11:15 ` Mika Westerberg
2019-10-29 20:27 ` Bjorn Helgaas
2019-10-30 11:15 ` Mika Westerberg
2019-10-31 22:31 ` Bjorn Helgaas
2019-11-01 11:19 ` Mika Westerberg
2019-11-05 0:00 ` Bjorn Helgaas
2019-11-05 9:54 ` Mika Westerberg
2019-11-05 12:58 ` Mika Westerberg
2019-11-05 20:01 ` Bjorn Helgaas
2019-11-06 13:31 ` Mika Westerberg
2019-11-05 15:00 ` Bjorn Helgaas
2019-11-05 15:28 ` Mika Westerberg
2019-11-05 16:10 ` Bjorn Helgaas
2019-11-06 13:29 ` Mika Westerberg
2019-10-29 20:54 ` Bjorn Helgaas
2019-10-30 11:33 ` Mika Westerberg [this message]
2019-10-04 12:57 ` [PATCH v2 0/2] PCI: Add missing link delays Matthias Andree
2019-10-04 13:06 ` Mika Westerberg
2019-10-05 7:34 ` Matthias Andree
2019-10-07 9:32 ` Mika Westerberg
2019-10-07 15:15 ` Matthias Andree
2019-10-08 9:05 ` Mika Westerberg
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=20191030113353.GY2593@lahna.fi.intel.com \
--to=mika.westerberg@linux.intel.com \
--cc=alex.williamson@redhat.com \
--cc=helgaas@kernel.org \
--cc=kai.heng.feng@canonical.com \
--cc=keith.busch@intel.com \
--cc=lenb@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lukas@wunner.de \
--cc=matthias.andree@gmx.de \
--cc=mr.nuke.me@gmail.com \
--cc=nicholas.johnson-opensource@outlook.com.au \
--cc=pmenzel@molgen.mpg.de \
--cc=rjw@rjwysocki.net \
/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).