linux-kernel - [PATCH] PCI: dwc: Use level-triggered handler for MSI IRQs

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>] [day] [month] [year] [list]

Message-ID: <20241015141215.1.Id60295bee6aacf44aa3664e702012cb4710529c3@changeid>
Date: Tue, 15 Oct 2024 14:12:16 -0700
From: Brian Norris <briannorris@...omium.org>
To: Jingoo Han <jingoohan1@...il.com>,
	Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
Cc: Lorenzo Pieralisi <lpieralisi@...nel.org>,
	Bjorn Helgaas <bhelgaas@...gle.com>,
	linux-kernel@...r.kernel.org,
	linux-pci@...r.kernel.org,
	Rob Herring <robh@...nel.org>,
	Marc Zyngier <marc.zyngier@....com>,
	Krzysztof Wilczyński <kw@...ux.com>,
	Brian Norris <briannorris@...gle.com>,
	Brian Norris <briannorris@...omium.org>
Subject: [PATCH] PCI: dwc: Use level-triggered handler for MSI IRQs

From: Brian Norris <briannorris@...gle.com>

Per Synopsis's documentation, the msi_ctrl_int signal is
level-triggered, not edge-triggered.

The use of handle_edge_trigger() was chosen in commit 7c5925afbc58
("PCI: dwc: Move MSI IRQs allocation to IRQ domains hierarchical API"),
which actually dropped preexisting use of handle_level_trigger().
Looking at the patch history, this change was only made by request:

  Subject: Re: [PATCH v6 1/9] PCI: dwc: Add IRQ chained API support
  https://lore.kernel.org/all/04d3d5b6-9199-218d-476f-c77d04b8d2e7@arm.com/

  "Are you sure about this "handle_level_irq"? MSIs are definitely edge
   triggered, not level."

However, while the underlying MSI protocol is edge-triggered in a sense,
the DesignWare IP is actually level-triggered.

So, let's switch back to level-triggered.

In many cases, the distinction doesn't really matter here, because this
signal is hidden behind another (chained) top-level IRQ which can be
masked on its own. But it's still wise to manipulate this interrupt line
according to its actual specification -- specifically, to mask it while
we handle it.

Signed-off-by: Brian Norris <briannorris@...gle.com>
Signed-off-by: Brian Norris <briannorris@...omium.org>
---

 drivers/pci/controller/dwc/pcie-designware-host.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
index 3e41865c7290..0fb79a26d05e 100644
--- a/drivers/pci/controller/dwc/pcie-designware-host.c
+++ b/drivers/pci/controller/dwc/pcie-designware-host.c
@@ -198,7 +198,7 @@ static int dw_pcie_irq_domain_alloc(struct irq_domain *domain,
 	for (i = 0; i < nr_irqs; i++)
 		irq_domain_set_info(domain, virq + i, bit + i,
 				    pp->msi_irq_chip,
-				    pp, handle_edge_irq,
+				    pp, handle_level_irq,
 				    NULL, NULL);

 	return 0;
-- 
2.47.0.rc1.288.g06298d1525-goog