linux-kernel - [PATCH 6.16 2/2] genirq/cpuhotplug: Restore affinity even for suspended IRQ

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: <20250612183303.3433234-3-briannorris@chromium.org>
Date: Thu, 12 Jun 2025 11:32:52 -0700
From: Brian Norris <briannorris@...omium.org>
To: Thomas Gleixner <tglx@...utronix.de>
Cc: Douglas Anderson <dianders@...omium.org>,
	David Stevens <stevensd@...omium.org>,
	linux-kernel@...r.kernel.org,
	Tsai Sung-Fu <danielsftsai@...gle.com>,
	Aleksandrs Vinarskis <alex.vinarskis@...il.com>,
	Johan Hovold <johan@...nel.org>,
	Brian Norris <briannorris@...omium.org>
Subject: [PATCH 6.16 2/2] genirq/cpuhotplug: Restore affinity even for suspended IRQ

Commit 788019eb559f ("genirq: Retain disable depth for managed
interrupts across CPU hotplug") tried to make managed shutdown/startup
properly reference counted, but it missed the fact that the unplug and
hotplug code had an intentional imbalance -- that they skip
IRQS_SUSPENDED interrupts on the "restore" path. This means that if a
managed-affinity interrupt was both suspended and managed-shutdown (such
as may happen during system suspend / S3), we would skip calling
irq_startup_managed() when resuming, and would again have an unbalanced
depth -- this time, with a positive value (i.e., remaining unexpectedly
masked).

This IRQS_SUSPENDED check was introduced in commit a60dd06af674
("genirq/cpuhotplug: Skip suspended interrupts when restoring affinity")
for essentially the same reason as commit 788019eb559f -- that
irq_startup() would unconditionally re-enable an interrupt too early.
Because irq_startup_managed() now respsects the disable-depth count, we
no longer need the IRQS_SUSPENDED check, and instead, it causes harm.

Thus, drop the IRQS_SUSPENDED check, and restore balance.

This effectively reverts commit a60dd06af674 ("genirq/cpuhotplug: Skip
suspended interrupts when restoring affinity"), because it is replaced
by commit 788019eb559f ("genirq: Retain disable depth for managed
interrupts across CPU hotplug").

Fixes: 788019eb559f ("genirq: Retain disable depth for managed interrupts across CPU hotplug")
Reported-by: Aleksandrs Vinarskis <alex.vinarskis@...il.com>
Closes: https://lore.kernel.org/lkml/24ec4adc-7c80-49e9-93ee-19908a97ab84@gmail.com/
Signed-off-by: Brian Norris <briannorris@...omium.org>
---

 kernel/irq/cpuhotplug.c | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c
index f07529ae4895..755346ea9819 100644
--- a/kernel/irq/cpuhotplug.c
+++ b/kernel/irq/cpuhotplug.c
@@ -210,13 +210,6 @@ static void irq_restore_affinity_of_irq(struct irq_desc *desc, unsigned int cpu)
 	    !irq_data_get_irq_chip(data) || !cpumask_test_cpu(cpu, affinity))
 		return;

-	/*
-	 * Don't restore suspended interrupts here when a system comes back
-	 * from S3. They are reenabled via resume_device_irqs().
-	 */
-	if (desc->istate & IRQS_SUSPENDED)
-		return;
-
 	if (irqd_is_managed_and_shutdown(data))
 		irq_startup_managed(desc);

-- 
2.50.0.rc1.591.g9c95f17f64-goog