From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pavel Machek Subject: Re: [PATCH v1] x86/power/64: Restore processor state before using per-cpu variables Date: Fri, 12 Aug 2016 08:29:19 +0200 Message-ID: <20160812062919.GC30992@amd> References: <1470952169-39061-1-git-send-email-thgarnie@google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:39438 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751252AbcHLG3X (ORCPT ); Fri, 12 Aug 2016 02:29:23 -0400 Content-Disposition: inline In-Reply-To: <1470952169-39061-1-git-send-email-thgarnie@google.com> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Thomas Garnier Cc: "Rafael J . Wysocki" , Len Brown , linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org, keescook@chromium.org, kernel-hardening@lists.openwall.com, jikos@kernel.org, bpetkov@suse.de, yinghai@kernel.org Hi! > Restore the processor state before calling any other function to ensure > per-cpu variables can be used with KASLR memory randomization. > > Tracing functions use per-cpu variables (gs based) and one was called > just before restoring the processor state fully. It resulted in a double > fault when both the tracing & the exception handler functions tried to > use a per-cpu variable. > > Signed-off-by: Thomas Garnier > --- > Based on next-20160808 > > Thanks to Rafael, Jiri & Borislav in tracking down this bug. > --- > kernel/power/hibernate.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c > index a881c6a..33c79b6 100644 > --- a/kernel/power/hibernate.c > +++ b/kernel/power/hibernate.c > @@ -300,12 +300,12 @@ static int create_image(int platform_mode) > save_processor_state(); > trace_suspend_resume(TPS("machine_suspend"), PM_EVENT_HIBERNATE, true); > error = swsusp_arch_suspend(); > + /* Restore control flow magically appears here */ > + restore_processor_state(); > trace_suspend_resume(TPS("machine_suspend"), PM_EVENT_HIBERNATE, false); > if (error) > printk(KERN_ERR "PM: Error %d creating hibernation image\n", > error); > - /* Restore control flow magically appears here */ > - restore_processor_state(); > if (!in_suspend) > events_check_enabled = false; > Ugh. Plus it also fixes very confusing situation where /* Restore control flow magically appears here */ comment was 4 lines away from where it _actually_ magically appeared. Good catch. Acked-by: Pavel Machek -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html