2006-02-20 14:35:47

by Mark Lord

[permalink] [raw]
Subject: 2.6.16-rc[34]: resume-from-RAM unreliable

For the past week, I've been trying to keep with the latest -rc*-git*
releases of 2.6.16 on my notebook, and something new in those is
impacting resume-from-RAM.

It works most of the time, but *rarely* when suspended for more
than a few hours (like overnight). Which makes testing somewhat
challenging, as this is also my main work machine.

Black screen on resume, no response to alt-sysrq-b,
and no convenient serial port in the machine.

Works 100% with 2.6.15 kernels.

If I can figure out anything more that's useful, I'll do so,
but that's all I know about it right now.

Debugging resume on a "legacy free" notebook is not fun.

-ml


2006-02-21 05:11:24

by Andrew Morton

[permalink] [raw]
Subject: Re: 2.6.16-rc[34]: resume-from-RAM unreliable

Mark Lord <[email protected]> wrote:
>
> For the past week, I've been trying to keep with the latest -rc*-git*
> releases of 2.6.16 on my notebook, and something new in those is
> impacting resume-from-RAM.
>
> It works most of the time, but *rarely* when suspended for more
> than a few hours (like overnight). Which makes testing somewhat
> challenging, as this is also my main work machine.
>
> Black screen on resume, no response to alt-sysrq-b,
> and no convenient serial port in the machine.
>
> Works 100% with 2.6.15 kernels.
>
> If I can figure out anything more that's useful, I'll do so,
> but that's all I know about it right now.
>
> Debugging resume on a "legacy free" notebook is not fun.
>

It's awful, isn't it? git-bisct is the best bet, but at one bisection
point per day it could get a bit dull.

2006-02-21 23:48:13

by Mark Lord

[permalink] [raw]
Subject: Re: 2.6.16-rc[34]: resume-from-RAM unreliable

Andrew Morton wrote:
> Mark Lord <[email protected]> wrote:
>> For the past week, I've been trying to keep with the latest -rc*-git*
>> releases of 2.6.16 on my notebook, and something new in those is
>> impacting resume-from-RAM.

Mmm.. working 100% now with Randy Dunlop's ACPI additions from here:

http://www.xenotime.net/linux/SATA/2.6.16-rc3/

Note that these are the latest evolution from a simpler patch by Jens
which I first started using in *spring 2005* to get suspend/resume working.

Are we *ever* going to add these to mainstream?
It's been almost a friggin' year!

Cheers

2006-02-21 23:50:14

by Randy Dunlap

[permalink] [raw]
Subject: Re: 2.6.16-rc[34]: resume-from-RAM unreliable

On Tue, 21 Feb 2006, Mark Lord wrote:

> Andrew Morton wrote:
> > Mark Lord <[email protected]> wrote:
> >> For the past week, I've been trying to keep with the latest -rc*-git*
> >> releases of 2.6.16 on my notebook, and something new in those is
> >> impacting resume-from-RAM.
>
> Mmm.. working 100% now with Randy Dunlop's ACPI additions from here:
>
> http://www.xenotime.net/linux/SATA/2.6.16-rc3/
>
> Note that these are the latest evolution from a simpler patch by Jens
> which I first started using in *spring 2005* to get suspend/resume working.
>
> Are we *ever* going to add these to mainstream?
> It's been almost a friggin' year!
>
> Cheers

Yes, I understand (I think). Fortunately I haven't been working
on it quite that long.

I'm updating the patchset to 2.6.16-rc4 and then I'll post it,
probably tomorrow.

--
~Randy