From: sashiko-bot@kernel.org
To: "Yury Murashka" <yurypm@arista.com>
Cc: linux-pci@vger.kernel.org
Subject: Re: [PATCH v2 2/2] PCI: Add pci=nodpc kernel boot option
Date: Thu, 09 Jul 2026 19:07:30 +0000 [thread overview]
Message-ID: <20260709190730.E19A11F000E9@smtp.kernel.org> (raw)
In-Reply-To: <20260709185429.627968-3-yurypm@arista.com>
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The `pci=nodpc` parameter disables native OS DPC but still claims DPC control via ACPI `_OSC` and enables it via `_DSM`, causing permanent unhandled link errors.
--
--- Patch [2]: [PATCH v2 2/2] PCI: Add pci=nodpc kernel boot option ---
commit 9eea006773bbca0de0b0f501e83971ad7b74166e
Author: Yury Murashka <yurypm@arista.com>
PCI: Add pci=nodpc kernel boot option
This patch introduces the pci=nodpc kernel boot option to disable PCI
Downstream Port Containment (DPC). When specified, DPC initialization,
state operations, and driver registration are bypassed, allowing userspace
to handle PCIe errors.
> diff --git a/drivers/pci/pcie/dpc.c b/drivers/pci/pcie/dpc.c
> index 2b779bd1d861b..759d9f18812ee 100644
> --- a/drivers/pci/pcie/dpc.c
> +++ b/drivers/pci/pcie/dpc.c
[ ... ]
> @@ -404,6 +411,9 @@ void pci_dpc_init(struct pci_dev *pdev)
> {
> u16 cap;
>
> + if (pcie_dpc_disable)
> + return;
> +
[Severity: High]
Does returning early here leave pdev->dpc_cap as 0 and cause issues with Error
Disconnect Recover (EDR)?
When pci=nodpc is set, calculate_control() in drivers/acpi/pci_root.c still
requests DPC control from firmware based solely on the compile-time flag:
drivers/acpi/pci_root.c:calculate_control() {
...
if (IS_ENABLED(CONFIG_PCIE_DPC) && IS_ENABLED(CONFIG_PCIE_EDR))
control |= OSC_PCI_EXPRESS_DPC_CONTROL;
...
}
And pci_acpi_add_edr_notifier() in drivers/pci/pcie/edr.c unconditionally
enables DPC in hardware:
drivers/pci/pcie/edr.c:pci_acpi_add_edr_notifier() {
...
if (acpi_enable_dpc(pdev))
acpi_remove_notify_handler(adev->handle, ACPI_SYSTEM_NOTIFY,
edr_handle_event);
...
}
If a PCIe error occurs and ACPI firmware notifies the OS, edr_handle_event()
will trigger but immediately abort because edev->dpc_cap is 0:
drivers/pci/pcie/edr.c:edr_handle_event() {
...
if (!edev->dpc_cap) {
pci_err(edev, FW_BUG "This device doesn't support DPC\n");
goto send_ost;
}
...
}
Will this cause EDR to abort recovery, incorrectly log a firmware bug, and
permanently abandon the link instead of letting userspace handle it?
> pdev->dpc_cap = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_DPC);
> if (!pdev->dpc_cap)
> return;
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260709185429.627968-1-yurypm@arista.com?part=2
next prev parent reply other threads:[~2026-07-09 19:07 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-07-09 18:54 [PATCH v2 0/2] PCI: Add boot options to disable DPC and AER recovery Yury Murashka
2026-07-09 18:54 ` [PATCH v2 1/2] PCI: Add pci=noaer_recovery kernel boot option Yury Murashka
2026-07-09 19:06 ` sashiko-bot
2026-07-09 18:54 ` [PATCH v2 2/2] PCI: Add pci=nodpc " Yury Murashka
2026-07-09 19:07 ` sashiko-bot [this message]
2026-07-09 22:42 ` [PATCH v2 0/2] PCI: Add boot options to disable DPC and AER recovery Bjorn Helgaas
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=20260709190730.E19A11F000E9@smtp.kernel.org \
--to=sashiko-bot@kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=sashiko-reviews@lists.linux.dev \
--cc=yurypm@arista.com \
/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