linux-kernel - Re: [PATCH] x86/ioapic: add IO-APIC 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] [day] [month] [year] [list]

Date:   Wed, 18 Jan 2017 15:36:23 +0100 (CET)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     "Ruslan Ruslichenko -X (rruslich - GLOBALLOGIC INC at Cisco)" 
        <rruslich@...co.com>
cc:     Ingo Molnar <mingo@...hat.com>, "H. Peter Anvin" <hpa@...or.com>,
        x86@...nel.org, linux-kernel@...r.kernel.org,
        xe-linux-external@...co.com
Subject: Re: [PATCH] x86/ioapic: add IO-APIC irq_chip retrigger hook

On Wed, 18 Jan 2017, Ruslan Ruslichenko -X (rruslich - GLOBALLOGIC INC at Cisco) wrote:
> On 01/18/2017 12:46 PM, Thomas Gleixner wrote:
> > On Tue, 17 Jan 2017, Ruslan Ruslichenko wrote:
> > 
> > > After commit d32932d02e18  ("x86/irq: Convert IOAPIC to use hierarchical
> > > irqdomain interfaces") IO-APIC is switched to hierarchical irqdomain
> > > implementation. Also this patch removed irq_retrigger hook for IO-APIC
> > > chip which seem to be wrong. Without this hook some edge-triggered irq's
> > > might be lost, since kernel uses it to retrigger edge irq's triggered
> > > while they are being handled or disabled by driver
> > Restoring the callbacks is the right thing to do, but there is no problem
> > because if the retrigger callback is empty then pending edge interrupts are
> > resent by software. It's just less efficient, but not a issue.
> 
> Software resend will only work if CONFIG_HARDIRQS_SW_RESEND is enabled,
> which is not selected for x86 architectures.

Dammit, right you are. Good question why nobody noticed so far.

Lemme amend that commit, so it gets back into stable then.

Thanks,

	tglx