mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Manikanta Maddireddy <mmaddireddy@nvidia.com>
To: <lpieralisi@kernel.org>, <kwilczynski@kernel.org>,
	<mani@kernel.org>, <bhelgaas@google.com>,
	<linux-pci@vger.kernel.org>
Cc: <robh@kernel.org>, <thierry.reding@kernel.org>,
	<jonathanh@nvidia.com>, <kishon@kernel.org>,
	<Frank.Li@kernel.org>, <vidyas@nvidia.com>, <cassel@kernel.org>,
	<den@valinux.co.jp>, <marco.crivellari@suse.com>,
	<andriy.shevchenko@linux.intel.com>,
	<linux-tegra@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	<mmaddireddy@nvidia.com>
Subject: [PATCH v2 03/10] PCI: tegra194: Fix Endpoint PERST# IRQ suspend race
Date: Wed, 21 Oct 2026 11:43:42 +0530	[thread overview]
Message-ID: <20261021061349.1265704-4-mmaddireddy@nvidia.com> (raw)
In-Reply-To: <20261021061349.1265704-1-mmaddireddy@nvidia.com>

Endpoint system suspend is allowed only while the endpoint is disabled.
Checking ep_state before disabling the PERST# IRQ leaves a window in which
PERST# deassertion starts the threaded handler after the check. The
handler can enable the controller, clocks and PHYs before disable_irq()
returns, but suspend still succeeds based on the earlier disabled state.
The system then proceeds into suspend with an active endpoint that should
have caused the suspend callback to reject the transition.

Disable the PERST# IRQ before checking ep_state. disable_irq() waits for
an already-running threaded handler to finish, so the state check sees
its completed transition and no new handler can enable the endpoint.
Re-enable the IRQ immediately if the endpoint is enabled and reject
suspend. Otherwise, keep it disabled until the normal resume callback.
This also balances the IRQ disable when another device aborts suspend
and PM rolls back through .resume without calling .resume_early.

This preserves the existing policy that PERST# is not serviced while the
endpoint is suspended. It serializes the software state check with the
IRQ thread; it does not make PERST# a wakeup source or guarantee delivery
of transitions during suspension.

Wrap the PM ops pointer with pm_sleep_ptr() so the callbacks are used
only when system sleep is enabled.

Fixes: c76f8eae7d46 ("PCI: tegra194: Allow system suspend when the Endpoint link is not up")
Link: https://lore.kernel.org/r/20260324190755.1094879-10-mmaddireddy@nvidia.com
Signed-off-by: Manikanta Maddireddy <mmaddireddy@nvidia.com>
---
Changes V1 -> V2:
- Expand the commit message to explain the active-endpoint suspend race,
  IRQ-thread synchronization and suspend rollback handling.
- Clarify that PERST# is not serviced while the endpoint is suspended.
- Add a comment explaining why the IRQ thread is drained before checking
  endpoint state.

 drivers/pci/controller/dwc/pcie-tegra194.c | 34 ++++++++++++++--------
 1 file changed, 22 insertions(+), 12 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
index 89dbef83dd9e..0f895c766a71 100644
--- a/drivers/pci/controller/dwc/pcie-tegra194.c
+++ b/drivers/pci/controller/dwc/pcie-tegra194.c
@@ -2374,19 +2374,30 @@ static int tegra_pcie_dw_suspend(struct device *dev)
 {
 	struct tegra_pcie_dw *pcie = dev_get_drvdata(dev);
 
-	if (pcie->of_data->mode == DW_PCIE_EP_TYPE) {
-		if (pcie->ep_state == EP_STATE_ENABLED) {
-			dev_err(dev, "Tegra PCIe is in EP mode, suspend not allowed\n");
-			return -EPERM;
-		}
-
-		disable_irq(pcie->pex_rst_irq);
+	if (pcie->of_data->mode != DW_PCIE_EP_TYPE)
 		return 0;
+
+	/* Drain the PERST# thread before checking whether the endpoint is idle. */
+	disable_irq(pcie->pex_rst_irq);
+	if (pcie->ep_state == EP_STATE_ENABLED) {
+		enable_irq(pcie->pex_rst_irq);
+		dev_err(dev, "Tegra PCIe is in EP mode, suspend not allowed\n");
+		return -EPERM;
 	}
 
 	return 0;
 }
 
+static int tegra_pcie_dw_resume(struct device *dev)
+{
+	struct tegra_pcie_dw *pcie = dev_get_drvdata(dev);
+
+	if (pcie->of_data->mode == DW_PCIE_EP_TYPE)
+		enable_irq(pcie->pex_rst_irq);
+
+	return 0;
+}
+
 static int tegra_pcie_dw_suspend_late(struct device *dev)
 {
 	struct tegra_pcie_dw *pcie = dev_get_drvdata(dev);
@@ -2462,10 +2473,8 @@ static int tegra_pcie_dw_resume_early(struct device *dev)
 	struct tegra_pcie_dw *pcie = dev_get_drvdata(dev);
 	u32 val;
 
-	if (pcie->of_data->mode == DW_PCIE_EP_TYPE) {
-		enable_irq(pcie->pex_rst_irq);
+	if (pcie->of_data->mode == DW_PCIE_EP_TYPE)
 		return 0;
-	}
 
 	if (!pcie->link_state)
 		return 0;
@@ -2569,10 +2578,11 @@ static const struct of_device_id tegra_pcie_dw_of_match[] = {
 
 static const struct dev_pm_ops tegra_pcie_dw_pm_ops = {
 	.suspend = tegra_pcie_dw_suspend,
+	.resume = tegra_pcie_dw_resume,
 	.suspend_late = tegra_pcie_dw_suspend_late,
+	.resume_early = tegra_pcie_dw_resume_early,
 	.suspend_noirq = tegra_pcie_dw_suspend_noirq,
 	.resume_noirq = tegra_pcie_dw_resume_noirq,
-	.resume_early = tegra_pcie_dw_resume_early,
 };
 
 static struct platform_driver tegra_pcie_dw_driver = {
@@ -2581,7 +2591,7 @@ static struct platform_driver tegra_pcie_dw_driver = {
 	.shutdown = tegra_pcie_dw_shutdown,
 	.driver = {
 		.name	= "tegra194-pcie",
-		.pm = &tegra_pcie_dw_pm_ops,
+		.pm = pm_sleep_ptr(&tegra_pcie_dw_pm_ops),
 		.of_match_table = tegra_pcie_dw_of_match,
 	},
 };
-- 
2.34.1


  parent reply	other threads:[~2026-10-21  6:14 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-21  6:13 [PATCH v2 00/10] PCI: tegra194: Fix EP and Root Port corner cases Manikanta Maddireddy
2026-10-21  6:13 ` [PATCH v2 01/10] PCI: tegra194: Propagate REFCLK select GPIO errors Manikanta Maddireddy
2026-10-21 15:14   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 02/10] PCI: tegra194: Check core reset deassertion Manikanta Maddireddy
2026-10-21 15:16   ` Frank Li
2026-10-21  6:13 ` Manikanta Maddireddy [this message]
2026-10-21 15:19   ` [PATCH v2 03/10] PCI: tegra194: Fix Endpoint PERST# IRQ suspend race Frank Li
2026-10-21  6:13 ` [PATCH v2 04/10] PCI: tegra194: Do not skip no-link Root Port remove cleanup Manikanta Maddireddy
2026-10-21 15:22   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 05/10] PCI: tegra194: Check for 16 GT/s capability before programming Manikanta Maddireddy
2026-10-21 15:24   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 06/10] PCI: tegra194: Check for L1SS " Manikanta Maddireddy
2026-10-21 15:28   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 07/10] PCI: tegra194: Always disable Tegra234 Endpoint L1.2 Manikanta Maddireddy
2026-10-21 15:32   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 08/10] PCI: tegra194: Guard Endpoint PLL-off error path Manikanta Maddireddy
2026-10-21 15:36   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 09/10] PCI: tegra194: Balance core monitor clock on failures Manikanta Maddireddy
2026-10-21 15:43   ` Frank Li
2026-10-21  6:13 ` [PATCH v2 10/10] PCI: tegra194: Reject zero MSI interrupt number Manikanta Maddireddy
2026-10-21 15:44   ` Frank Li

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=20261021061349.1265704-4-mmaddireddy@nvidia.com \
    --to=mmaddireddy@nvidia.com \
    --cc=Frank.Li@kernel.org \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=bhelgaas@google.com \
    --cc=cassel@kernel.org \
    --cc=den@valinux.co.jp \
    --cc=jonathanh@nvidia.com \
    --cc=kishon@kernel.org \
    --cc=kwilczynski@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=linux-tegra@vger.kernel.org \
    --cc=lpieralisi@kernel.org \
    --cc=mani@kernel.org \
    --cc=marco.crivellari@suse.com \
    --cc=robh@kernel.org \
    --cc=thierry.reding@kernel.org \
    --cc=vidyas@nvidia.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

all inboxes | Powered by JetHome®