Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1171154 > unrolled thread

Re: [GIT PULL] EDAC updates for 4.2

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [GIT PULL] EDAC updates for 4.2 Borislav Petkov <bp@suse.de> - 2015-06-24 09:50 +0200
    Re: [GIT PULL] EDAC updates for 4.2 Ingo Molnar <mingo@kernel.org> - 2015-06-24 10:30 +0200
      Re: [GIT PULL] EDAC updates for 4.2 Borislav Petkov <bp@suse.de> - 2015-06-24 11:00 +0200
      Re: [GIT PULL] EDAC updates for 4.2 Thomas Gleixner <tglx@linutronix.de> - 2015-06-24 11:00 +0200

#1171154 — Re: [GIT PULL] EDAC updates for 4.2

On Tue, Jun 23, 2015 at 03:49:50PM -0700, Linus Torvalds wrote:
> On Mon, Jun 22, 2015 at 2:11 AM, Borislav Petkov <bp@suse.de> wrote:
> >
> > Important: Please merge this stuff *after* you have merged the tip pile
> > because it depends on it.
> 
> What does this mean?

It means that it depends on functionality which went in through tip.

And to be more specific, the dependencies are the DEFERRED_ERROR_VECTOR
define and the amd_get_nodes_per_socket() helper. And it all is confined
to the AMD error injection module drivers/edac/mce_amd_inj.c.

So this is a debugging module for error injection - not something anyone
would ever use in production.

Everything else works fine.

> If it doesn't work or compile without the tip pile, then I'm not
> pulling it at all, since that means that any problems are not
> bisectable.

Ok, how would you prefer this solved - should I merge the relevant tip
branches into it?

Or should I remove the drivers/edac/mce_amd_inj.c changes from the pull
request?

In general, how would you prefer EDAC stuff handled properly when it
depends on x86 functionality which goes through tip?

> The patches are based on 4.1-rc1. If it doesn't work on top of that,
> then that means that you clearly have tested *none* of this. Which
> just makes me go "yeah, I'm not pulling untested crap".

Of course it has been tested but with the relevant tip branches merged.

Thanks.

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1171174

* Borislav Petkov <bp@suse.de> wrote:

> > If it doesn't work or compile without the tip pile, then I'm not pulling it at 
> > all, since that means that any problems are not bisectable.
> 
> Ok, how would you prefer this solved - should I merge the relevant tip branches 
> into it?

So the broken commit is 1862c7c3483c. Given that only a handful of commits follow 
it, I'd suggest the following resolution, to preserve half of your tree:

  git checkout -b tmp.tmp your/edac/tree
  git reset --hard 1862c7c3483c^1
  git merge linus # where 'linus' is 407a2c720556e - your DEFERRED_ERROR_VECTOR dependency
  git checkout -b tmp.tmp2 your/edac/tree
  git rebase --onto tmp.tmp 1862c7c3483c^1

Then re-test and re-submit the result to Linus as -v2.

Note that 407a2c720556e is very early in the v4.2 merge window, with only -tip 
(and some VFS) bits in it, which I have tested and which is looking pretty robust.

In the future we can do such dependencies in -tip as well, so that they don't 
reach Linus - I didn't realize your tree had this dependency.

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1171200

On Wed, Jun 24, 2015 at 10:50:38AM +0200, Thomas Gleixner wrote:
> One of the solution I use for cross tree dependencies is:
> 
>     - Apply the commits which create a dependency to a seperate branch
> 
>     - Let the depending tree pull that branch
> 
>     - Merge the branch into the proper tip/ branch

Yeah, in hindsight, we probably should've simply routed all the changes
to amd_mce_inj (the error injection module which depends on the tip
changes) straight through tip/x86/ras and wouldn't have any issues. Oh
well, I'll know that now for the future.

So I'll wait out to hear what Linus prefers and do that.

Thanks guys.

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1171204

On Wed, 24 Jun 2015, Ingo Molnar wrote:

> * Borislav Petkov <bp@suse.de> wrote:
> 
> > > If it doesn't work or compile without the tip pile, then I'm not pulling it at 
> > > all, since that means that any problems are not bisectable.
> > 
> > Ok, how would you prefer this solved - should I merge the relevant tip branches 
> > into it?
> 
> So the broken commit is 1862c7c3483c. Given that only a handful of commits follow 
> it, I'd suggest the following resolution, to preserve half of your tree:
> 
>   git checkout -b tmp.tmp your/edac/tree
>   git reset --hard 1862c7c3483c^1
>   git merge linus # where 'linus' is 407a2c720556e - your DEFERRED_ERROR_VECTOR dependency
>   git checkout -b tmp.tmp2 your/edac/tree
>   git rebase --onto tmp.tmp 1862c7c3483c^1
> 
> Then re-test and re-submit the result to Linus as -v2.
> 
> Note that 407a2c720556e is very early in the v4.2 merge window, with only -tip 
> (and some VFS) bits in it, which I have tested and which is looking pretty robust.
> 
> In the future we can do such dependencies in -tip as well, so that they don't 
> reach Linus - I didn't realize your tree had this dependency.

One of the solution I use for cross tree dependencies is:

    - Apply the commits which create a dependency to a seperate branch

    - Let the depending tree pull that branch

    - Merge the branch into the proper tip/ branch

Thanks,

	tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web