2008-08-14 18:11:50

by David Witbrodt

[permalink] [raw]
Subject: Re: HPET regression in 2.6.26 versus 2.6.25 -- revert for 2.6.26-rc1 failed



> can you try enable kexec and kdump in you .config.
>
> it should works. my .config have config_kexec

First, I apologize. I see that you were assuming I had KEXEC enabled when you
gave me the patch against 2.6.27-rc3. I have only posted parts of my custom
.config to LKML, thinking that posting the whole thing was not appropriate.

Did you mean CRASH_DUMP instead of "kdump" above?

[N] kexec system call (CONFIG_KEXEC) (bool) (Help)
[N] kernel crash dumps (EXPERIMENTAL) (CONFIG_CRASH_DUMP) (bool) (Help)

My custom .config has both of these disabled -- either because they default
to "N" when running 'make oldconfig', or because I read the help text with
those options and decided I would not be using those features.


> > Some experiments I did last night may render these questions moot, though.
> > My problem is very specific to the hardware on two of my machines, and it
> > has something to do with setting up the system resources that
> > insert_resource() touches.
>
> please try to bisect on current tree. and every time apply the revert patch...

I will try your patch against 2.6.27-rc3 again, with KEXEC and CRASH_DUMP
enabled, when I get home from work. I still have not retried using 'patch'
instead of 'git apply', though I am sure that 'git apply' worked just fine.

I was trying other experiments last night, and have posted the results of
those already -- and I'm guessing that your 2.6.27-rc3 patch will not work
unless it factors in what I discovered there. (But I will do all of these
changes that you asked, as well.)

I do not know how to bisect with your patch if I have a "bad" but no "good"
to start with. Can you explain how I should proceed when I _do_ get home?
(I can just enabled those config options and try the patch again, but I am
confused about the bisect you are asking me to perform.)

Does my other post, about trying to revert those February commits, mean
anything?


Thanks,
Dave W.


2008-08-14 18:30:12

by Yinghai Lu

[permalink] [raw]
Subject: Re: HPET regression in 2.6.26 versus 2.6.25 -- revert for 2.6.26-rc1 failed

> I do not know how to bisect with your patch if I have a "bad" but no "good"
> to start with. Can you explain how I should proceed when I _do_ get home?
> (I can just enabled those config options and try the patch again, but I am
> confused about the bisect you are asking me to perform.)
>
just like the old way doing git-bisect, but before compiling, apply
the batch, and before git-bisect good or bad, revert the patch.

YH