2012-08-31 15:59:38

by Shuah Khan

[permalink] [raw]
Subject: Re: crash dump memory reservation regression

On Mon, 2012-03-12 at 22:42 -0700, H. Peter Anvin wrote:
> On 03/12/2012 10:31 PM, Yinghai Lu wrote:
> >
> > Solution will be:
> > 1. remove the good_end setting for 64 bit again. and root cause S4 resume.
>
> This would by far be the best.
>
> -hpa
>

Any resolution on this issue. Has this been fixed? I haven't seen this
message on the systems I am running the upstream kernel on, and would
like to get a commit ID (if any) that fixed the problem.

Thanks,
-- Shuah


2012-08-31 16:37:39

by H. Peter Anvin

[permalink] [raw]
Subject: Re: crash dump memory reservation regression

Kernel Summit is this week... people are away.

Shuah Khan <[email protected]> wrote:

>On Mon, 2012-03-12 at 22:42 -0700, H. Peter Anvin wrote:
>> On 03/12/2012 10:31 PM, Yinghai Lu wrote:
>> >
>> > Solution will be:
>> > 1. remove the good_end setting for 64 bit again. and root cause S4
>resume.
>>
>> This would by far be the best.
>>
>> -hpa
>>
>
>Any resolution on this issue. Has this been fixed? I haven't seen this
>message on the systems I am running the upstream kernel on, and would
>like to get a commit ID (if any) that fixed the problem.
>
>Thanks,
>-- Shuah

--
Sent from my mobile phone. Please excuse brevity and lack of formatting.

2012-08-31 16:42:18

by Takashi Iwai

[permalink] [raw]
Subject: Re: crash dump memory reservation regression

At Fri, 31 Aug 2012 09:37:05 -0700,
H. Peter Anvin wrote:
>
> Kernel Summit is this week... people are away.

I (as a person who asked the revert) don't mind now to set the
good_end back for 64bit again. On 3.5/3.6 kernels, there seems more
other places breaking the hibernation on the machines hitting the
problem at that time. I need to dig deeply, but it's a horrible time-
consuming and unreliable task, thus postponed, so far.


thanks,

Takashi


> Shuah Khan <[email protected]> wrote:
>
> >On Mon, 2012-03-12 at 22:42 -0700, H. Peter Anvin wrote:
> >> On 03/12/2012 10:31 PM, Yinghai Lu wrote:
> >> >
> >> > Solution will be:
> >> > 1. remove the good_end setting for 64 bit again. and root cause S4
> >resume.
> >>
> >> This would by far be the best.
> >>
> >> -hpa
> >>
> >
> >Any resolution on this issue. Has this been fixed? I haven't seen this
> >message on the systems I am running the upstream kernel on, and would
> >like to get a commit ID (if any) that fixed the problem.
> >
> >Thanks,
> >-- Shuah
>
> --
> Sent from my mobile phone. Please excuse brevity and lack of formatting.
>

2012-08-31 16:50:31

by Shuah Khan

[permalink] [raw]
Subject: Re: crash dump memory reservation regression

On Fri, 2012-08-31 at 18:42 +0200, Takashi Iwai wrote:
> At Fri, 31 Aug 2012 09:37:05 -0700,
> H. Peter Anvin wrote:
> >
> > Kernel Summit is this week... people are away.
>
> I (as a person who asked the revert) don't mind now to set the
> good_end back for 64bit again. On 3.5/3.6 kernels, there seems more
> other places breaking the hibernation on the machines hitting the
> problem at that time. I need to dig deeply, but it's a horrible time-
> consuming and unreliable task, thus postponed, so far.
>

I can work on getting this change in, if you are short on time. I am
seeing this on a few systems and would like to resolve it before 3.6
window closes.

Thanks,
-- Shuah