linux-kernel - Re: [PATCH] genirq: Keep handle_nested_irq() from touching desc->threads

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]

Date: Mon, 10 Jun 2024 20:23:09 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Peng Liu <iwtbavbm@...il.com>, linux-kernel@...r.kernel.org,
 stable@...r.kernel.org
Cc: gregkh@...uxfoundation.org, maz@...nel.org, vincent.whitchurch@...s.com,
 iwtbavbm@...il.com, 158710936@...com
Subject: Re: [PATCH] genirq: Keep handle_nested_irq() from touching
 desc->threads_active

On Mon, Jun 10 2024 at 02:30, Peng Liu wrote:
> handle_nested_irq() is supposed to be running inside the parent thread
> handler context. It per se has no dedicated kernel thread, thus shouldn't
> touch desc->threads_active. The parent kernel thread has already taken
> care of this.

No it has not. The parent thread has marked itself in the parent threads
interrupt descriptor.

How does that help synchronizing the nested interrupt, which has a
separate interrupt descriptor?

> Fixes: e2c12739ccf7 ("genirq: Prevent nested thread vs synchronize_hardirq() deadlock")
> Cc: stable@...r.kernel.org

There is nothing to fix.

> Signed-off-by: Peng Liu <iwtbavbm@...il.com>
> ---
>
> Despite of its correctness, I'm afraid the testing on my only PC can't
> cover the affected code path. So the patch may be totally -UNTESTED-.

Which correctness?

The change log of the commit you want to "fix" says:

    Remove the incorrect usage in the nested threaded interrupt case and
    instead re-use the threads_active / wait_for_threads mechanism to
    wait for nested threaded interrupts to complete.

It's very clearly spelled out, no?

Thanks,

        tglx