linux-kernel - perf regression after "genirq/affinity: Handle affinity setting on inactive interrupts correctly"

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-next>] [day] [month] [year] [list]

Message-ID: <c3bd7f1d-3126-d20d-67f3-2fed967a3ab9@nbd.name>
Date:   Fri, 21 Aug 2020 19:38:44 +0200
From:   Felix Fietkau <nbd@....name>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-arm-kernel@...ts.infradead.org
Subject: perf regression after "genirq/affinity: Handle affinity setting on
 inactive interrupts correctly"

Hi Thomas,

after updating my MT7622 ARM64 device from Linux 5.4.52 to 5.4.59, perf
stopped working. Whenever I started it, I got an IRQ storm on the second
arm-pmu IRQ. After some digging, I figured out that reverting commit
9f8d3d2f79ba189ecc122d214d32396e5737963b ("genirq/affinity: Handle
affinity setting on inactive interrupts correctly") made it work again.

Without the revert, the IRQ storm triggered on CPU0 (no interrupts on
CPU1).
With the revert, IRQs fire normally on CPU1 and perf works again.

It seems that setting the IRQ affinity for the arm-pmu IRQs does not
work anymore.

Please let me know if you need me to test anything.

- Felix