linux-kernel - Re: [PATCH] x86, MCE: Kill CPU_POST

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]

Message-ID: <537E687B.9080300@linux.vnet.ibm.com>
Date:	Fri, 23 May 2014 02:43:31 +0530
From:	"Srivatsa S. Bhat" <srivatsa.bhat@...ux.vnet.ibm.com>
To:	Borislav Petkov <bp@...en8.de>, "Luck, Tony" <tony.luck@...el.com>,
	Andi Kleen <andi@...stfloor.org>
CC:	Peter Zijlstra <peterz@...radead.org>,
	Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>,
	Jacob Pan <jacob.jun.pan@...ux.intel.com>,
	LKML <linux-kernel@...r.kernel.org>,
	Borislav Petkov <bp@...e.de>, Ingo Molnar <mingo@...nel.org>,
	"Wysocki, Rafael J" <rafael.j.wysocki@...el.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	"ego@...ux.vnet.ibm.com" <ego@...ux.vnet.ibm.com>,
	Oleg Nesterov <oleg@...hat.com>
Subject: Re: [PATCH] x86, MCE: Kill CPU_POST_DEAD

On 05/23/2014 01:25 AM, Borislav Petkov wrote:
> On Thu, May 22, 2014 at 03:50:21PM +0000, Luck, Tony wrote:
>>>> So I think we can reduce it to just the one rwsem (with recursion) if we
>>>> shoot CPU_POST_DEAD in the head.
>>>
>>> Here's the first bullet. Stressing my box here with Steve's hotplug
>>> script seems to work fine.
>>>
>>> Tony, any objections?
>>
>> what was this comment referring to:
>>
>>     /* intentionally ignoring frozen here */
>>
>> After you move the cmci_rediscover() call, it is now in a place where we are
>> no longer ignoring frozen (i.e. the old placement did the rediscover even if the
>> CPU_TASKS_FROZEN bit was set - with the new placement we will skip rediscovery.
>>

That's not quite true. The existing code already ignores FROZEN for all the cases,
by ignoring it at the top of the switch-case itself:

        switch (action & ~CPU_TASKS_FROZEN) {
        case CPU_ONLINE:
		[...]
		break;
        case CPU_DEAD:
                if (threshold_cpu_callback)
                        threshold_cpu_callback(action, cpu);
                mce_device_remove(cpu);
                mce_intel_hcpu_update(cpu);
                break;

Then I started wondering what the comment really meant, and commit 1a65f970d1
made things clear: its actually the _other_ way around! That is, cmci_rediscover()
didn't have to be invoked* during suspend/resume, so it was kept separate from
the rest.

* or maybe it was not _supposed_ to be invoked; I don't know which is the case..
the original commit 88ccbedd9 didn't explain that.

Either way, cmci_rediscover() doesn't seem to have any reason why it should
be called specifically in the POST_DEAD stage only. So I'm sure we can get rid
of that one way or other easily.

Regards,
Srivatsa S. Bhat

>> So we were working around some interaction between cpu hotplug and frozen.
>> Do we no longer need to do that?
> 
> Hmm, that FROZEN thing is supposedly for hotplug operations while
> suspend is happening. I guess it makes a little sense to rediscover CMCI
> banks while suspend is in progress. Whatever.
> 
> Let's keep it before more crap ensues, that was a good catch, thanks.
> 
> So, I guess something like that instead.
> 
> Which means, I'd need to run a couple of suspend/resume rounds while
> hotplugging cores to see whether we're still kosher.
> 
> More fun tomorrow.
> 
> ---
> From: Borislav Petkov <bp@...e.de>
> Date: Thu, 22 May 2014 16:40:54 +0200
> Subject: [PATCH] x86, MCE: Kill CPU_POST_DEAD
> 
> In conjunction with cleaning up CPU hotplug, we want to get rid of
> CPU_POST_DEAD. Kill this instance here and rediscover CMCI banks at the
> end of CPU_DEAD.
> 
> Signed-off-by: Borislav Petkov <bp@...e.de>
> ---
>  arch/x86/kernel/cpu/mcheck/mce.c | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
> index 68317c80de7f..bfde4871848f 100644
> --- a/arch/x86/kernel/cpu/mcheck/mce.c
> +++ b/arch/x86/kernel/cpu/mcheck/mce.c
> @@ -2391,6 +2391,10 @@ mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
>  			threshold_cpu_callback(action, cpu);
>  		mce_device_remove(cpu);
>  		mce_intel_hcpu_update(cpu);
> +
> +		/* intentionally ignoring frozen here */
> +		if (!(action & CPU_TASKS_FROZEN))
> +			cmci_rediscover();
>  		break;
>  	case CPU_DOWN_PREPARE:
>  		smp_call_function_single(cpu, mce_disable_cpu, &action, 1);
> @@ -2402,11 +2406,6 @@ mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
>  		break;
>  	}
> 
> -	if (action == CPU_POST_DEAD) {
> -		/* intentionally ignoring frozen here */
> -		cmci_rediscover();
> -	}
> -
>  	return NOTIFY_OK;
>  }
> 

--
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/