linux-kernel - Entry conditions for perf_event_do

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: <20100113041445.GA17829@brick.ozlabs.ibm.com>
Date:	Wed, 13 Jan 2010 15:14:45 +1100
From:	Paul Mackerras <paulus@...ba.org>
To:	Ingo Molnar <mingo@...e.hu>, Peter Zijlstra <peterz@...radead.org>
Cc:	linux-kernel@...r.kernel.org, Milton Miller <miltonm@....com>,
	Benjamin Herrenschmidt <benh@...nel.crashing.org>
Subject: Entry conditions for perf_event_do_pending?

We're seeing some perf-related crashes on powerpc related to having
irqs in an inconsistent state (soft-enable vs. hard-enable
vs. trace-irqs state) when entering perf_event_do_pending().
We're fixing that, but along the way we have struck a question about
what conditions are required on entry to perf_event_do_pending.

Its use of __get_cpu_var implies that it at least needs to be called
with either interrupts or preemption disabled.  Does it in fact need
to be called with irqs off?  Do we need to call irq_enter()/irq_exit()
around it?  Are there any other requirements that people can think of?

Paul.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/