linux-kernel - Re: [RFC PATCH] riscv: Fix PCI warning by enabling PCI_MSI_ARCH

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: <87v7vn917f.ffs@tglx>
Date: Fri, 13 Dec 2024 14:12:04 +0100
From: Thomas Gleixner <tglx@...utronix.de>
To: Alexandre Ghiti <alexghiti@...osinc.com>, Paul Walmsley
 <paul.walmsley@...ive.com>, Palmer Dabbelt <palmer@...belt.com>, Anup
 Patel <anup@...infault.org>, Sunil V L <sunilvl@...tanamicro.com>,
 linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org
Cc: Alexandre Ghiti <alexghiti@...osinc.com>
Subject: Re: [RFC PATCH] riscv: Fix PCI warning by enabling
 PCI_MSI_ARCH_FALLBACKS

On Fri, Dec 13 2024 at 12:57, Alexandre Ghiti wrote:
> When the interrupt controller is not using the IMSIC and ACPI is enabled,
> the following warning appears:
>
> [    0.866401] WARNING: CPU: 1 PID: 1 at drivers/pci/msi/msi.h:121 pci_msi_setup_msi_irqs+0x2c/0x32
> [    0.867071] Modules linked in:
> [    0.867389] CPU: 1 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.13.0-rc2-00001-g795582ce7e24-dirty #44
> [    0.867538] Hardware name: QEMU QEMU Virtual Machine, BIOS
> [    0.867672] epc : pci_msi_setup_msi_irqs+0x2c/0x32
> [    0.867738]  ra : __pci_enable_msix_range+0x30c/0x596

Removing a ton of badly formatted stack trace:

https://www.kernel.org/doc/html/latest/process/submitting-patches.html#backtraces-in-commit-messages

>
> So enable PCI_MSI_ARCH_FALLBACKS to get rid of this.

No. PCI_MSI_ARCH_FALLBACKS is really only meant for architectures which
implement the legacy fallbacks and not to paper over the underlying
logic bug in the pci/msi code. Of course the loongson folks ran into the
same problem two years ago and went for the sloppy fix without talking
to anyone...

Thanks for bringing it up instead of silently slapping it into the RISCV
tree !

The uncompiled patch below should fix this for real.

Thanks,

        tglx
---
--- a/arch/loongarch/Kconfig
+++ b/arch/loongarch/Kconfig
@@ -185,7 +185,6 @@ config LOONGARCH
 	select PCI_DOMAINS_GENERIC
 	select PCI_ECAM if ACPI
 	select PCI_LOONGSON
-	select PCI_MSI_ARCH_FALLBACKS
 	select PCI_QUIRKS
 	select PERF_USE_VMALLOC
 	select RTC_LIB
--- a/drivers/pci/msi/irqdomain.c
+++ b/drivers/pci/msi/irqdomain.c
@@ -350,8 +350,11 @@ bool pci_msi_domain_supports(struct pci_
 
 	domain = dev_get_msi_domain(&pdev->dev);
 
-	if (!domain || !irq_domain_is_hierarchy(domain))
-		return mode == ALLOW_LEGACY;
+	if (!domain || !irq_domain_is_hierarchy(domain)) {
+		if (IS_ENABLED(CONFIG_PCI_MSI_ARCH_FALLBACKS))
+			return mode == ALLOW_LEGACY;
+		return false;
+	}
 
 	if (!irq_domain_is_msi_parent(domain)) {
 		/*
--- a/drivers/pci/msi/msi.c
+++ b/drivers/pci/msi/msi.c
@@ -442,6 +442,10 @@ int __pci_enable_msi_range(struct pci_de
 	if (nvec > maxvec)
 		nvec = maxvec;
 
+	/* Test for the availability of MSI support */
+	if (!pci_msi_domain_supports(dev, 0, ALLOW_LEGACY))
+		return -ENOTSUPP;
+
 	rc = pci_setup_msi_context(dev);
 	if (rc)
 		return rc;