Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Russell King - ARM Linux]

FYI.

Presumably this hasn't been detected because linux-next isn't building
enough ARM configurations to hit one that has kprobes enabled?

----- Forwarded message from Abhishek Sagar <sagar.abhishek@gmail.com> -----

Delivery-date: Sat, 17 May 2008 12:30:06 +0100
Date: Sat, 17 May 2008 16:59:53 +0530
From: "Abhishek Sagar" <sagar.abhishek@gmail.com>
To: "Russell King - ARM Linux" <linux@arm.linux.org.uk>
Subject: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex
Cc: linux-arm-kernel@lists.arm.linux.org.uk, nico@cam.org, 
	mathieu.desnoyers@polymtl.ca

On 5/17/08, Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> The commit you refer to does not exist in mainline.

Yea, sorry. That commit is an upstream bubble in some unrelated branch
I was tracking.

Now that I look at it, the commit I'm referring to is in
git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86.git :
auto-sched-next. I merged all the recent kprobe related changes from
there into my local arm kprobes branch based on 2.6.26-rc2 and fired a
build for arm. That's when I encountered the kprobe build error.

> In addition, the URL you refer to above, the patch doesn't
> match the  diffstat - the patch doesn't touch kernel/kprobes.c.

Here's the updated version (series), with a proper diffstat:
http://lkml.org/lkml/2008/4/24/217

> The only thing the patch seems to be doing is removing the
> mutex.  It never adds anything.

Yes and this cleanup of kprobe_mutex usage from all the arch dependent
kprobe code needs to be done for arm too, else it'll break
compilation.

Regards,
Abhishek Sagar

----- End forwarded message -----

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1066 bytes --]

Hi Russell,

On Sat, 17 May 2008 12:42:45 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
>
> Presumably this hasn't been detected because linux-next isn't building
> enough ARM configurations to hit one that has kprobes enabled?

http://kisskb.ellerman.id.au/kisskb/compiler/11/

That's what I have building for arm.  I have asked twice for a suggested
set of configs.  Consider this request number three :-)

From: "Abhishek Sagar" <sagar.abhishek@gmail.com>
> 
> Now that I look at it, the commit I'm referring to is in
> git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86.git :
> auto-sched-next. I merged all the recent kprobe related changes from
> there into my local arm kprobes branch based on 2.6.26-rc2 and fired a
> build for arm. That's when I encountered the kprobe build error.

Also, that branch isn't in linux-next.  The commit maybe but I don't know
what commit is being discussed, sorry.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 289 bytes --]

Hi Russell,

On Sat, 17 May 2008 22:02:04 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> Also, that branch isn't in linux-next.

I should qualify that with "yet".

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Russell King - ARM Linux]

On Sat, May 17, 2008 at 10:02:04PM +1000, Stephen Rothwell wrote:
> Hi Russell,
> 
> On Sat, 17 May 2008 12:42:45 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> >
> > Presumably this hasn't been detected because linux-next isn't building
> > enough ARM configurations to hit one that has kprobes enabled?
> 
> http://kisskb.ellerman.id.au/kisskb/compiler/11/
> 
> That's what I have building for arm.  I have asked twice for a suggested
> set of configs.  Consider this request number three :-)

The answer is all of them.  I can't suggest any subset, mainly because
I don't know what's in each and what each will cover.  The coverage of
each is really up to each platform maintainer I'm afraid.

> From: "Abhishek Sagar" <sagar.abhishek@gmail.com>
> > 
> > Now that I look at it, the commit I'm referring to is in
> > git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86.git :
> > auto-sched-next. I merged all the recent kprobe related changes from
> > there into my local arm kprobes branch based on 2.6.26-rc2 and fired a
> > build for arm. That's when I encountered the kprobe build error.
> 
> Also, that branch isn't in linux-next.  The commit maybe but I don't know
> what commit is being discussed, sorry.

085a6d5e4fb23c6395da1d0106feed795b57c79b

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1021 bytes --]

Hi Russell,

On Sat, 17 May 2008 14:35:27 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
>
> On Sat, May 17, 2008 at 10:02:04PM +1000, Stephen Rothwell wrote:
> > 
> > That's what I have building for arm.  I have asked twice for a suggested
> > set of configs.  Consider this request number three :-)
> 
> The answer is all of them.  I can't suggest any subset, mainly because
> I don't know what's in each and what each will cover.  The coverage of
> each is really up to each platform maintainer I'm afraid.

OK, I will see what happens if I add them all.  You may need to ping me
if any get added or removed.

> > Also, that branch isn't in linux-next.  The commit maybe but I don't know
> > what commit is being discussed, sorry.
> 
> 085a6d5e4fb23c6395da1d0106feed795b57c79b

Yeah, as far as I can tell nothing like that one is in linux-next yet
(probably on Monday).

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 992 bytes --]

On Sat, 17 May 2008 23:59:26 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> On Sat, 17 May 2008 14:35:27 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> >
> > The answer is all of them.  I can't suggest any subset, mainly because
> > I don't know what's in each and what each will cover.  The coverage of
> > each is really up to each platform maintainer I'm afraid.
> 
> OK, I will see what happens if I add them all.  You may need to ping me
> if any get added or removed.

I have added them all.  There are some results on the web page already
(http://kisskb.ellerman.id.au/kisskb/compiler/11/) which are all ok.  On
our internal page (which will be mirrored above in an hour or so) there
are some failures (so far clps7500_defconfig, collie_defconfig,
corgi_defconfig, eseries_pxa_defconfig and magician_defconfig). Check
there later.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Russell King - ARM Linux]

On Sun, May 18, 2008 at 02:20:10AM +1000, Stephen Rothwell wrote:
> On Sat, 17 May 2008 23:59:26 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> >
> > On Sat, 17 May 2008 14:35:27 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> > >
> > > The answer is all of them.  I can't suggest any subset, mainly because
> > > I don't know what's in each and what each will cover.  The coverage of
> > > each is really up to each platform maintainer I'm afraid.
> > 
> > OK, I will see what happens if I add them all.  You may need to ping me
> > if any get added or removed.
> 
> I have added them all.  There are some results on the web page already
> (http://kisskb.ellerman.id.au/kisskb/compiler/11/) which are all ok.  On
> our internal page (which will be mirrored above in an hour or so) there
> are some failures (so far clps7500_defconfig, collie_defconfig,
> corgi_defconfig, eseries_pxa_defconfig and magician_defconfig). Check
> there later.

If I'm reading this right, it's taking you about 3.5 hours to build the
set?

spitz_defconfig - trivial fix (I'll push that with my upcoming set)
magican_defconfig - ditto
omap*_defconfig - hopefully my recent merge of fixes from Tony Lindgren
 should resolve this.
eseries_pxa_defconfig - looks like a botched defconfig file ;(
corgi_defconfig - tried to get the person concerned to submit a trivial
 patch but wants to introduce some infrastructure instead
clps7500_defconfig - you may as well disable, I'll probably eventually
 remove this platform.

Lastly, collie_defconfig seems to be broken as a result of 5025/1 -
which claims to be a "fix" patch but is actually a "break" patch. ;(

Thanks.

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1250 bytes --]

Hi Russell,

On Sat, 17 May 2008 20:08:55 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
>
> If I'm reading this right, it's taking you about 3.5 hours to build the
> set?

Not really, I added them in several goes and the build scripts have
significant pauses between passes.  It also takes some time to mirror out
the results between passes.  Tomorrow's build should be more
representative.

> spitz_defconfig - trivial fix (I'll push that with my upcoming set)
> magican_defconfig - ditto
> omap*_defconfig - hopefully my recent merge of fixes from Tony Lindgren
>  should resolve this.
> eseries_pxa_defconfig - looks like a botched defconfig file ;(
> corgi_defconfig - tried to get the person concerned to submit a trivial
>  patch but wants to introduce some infrastructure instead
> clps7500_defconfig - you may as well disable, I'll probably eventually
>  remove this platform.
>
> Lastly, collie_defconfig seems to be broken as a result of 5025/1 -
> which claims to be a "fix" patch but is actually a "break" patch. ;(

I'll disable clps7500_defconfig and leave the rest for now.  Thanks.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Fwd: Re: [PATCH] kprobes: fix compilation break due to kprobe_mutex

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 680 bytes --]

Hi Russell,

On Sun, 18 May 2008 13:49:33 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> On Sat, 17 May 2008 20:08:55 +0100 Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> >
> > If I'm reading this right, it's taking you about 3.5 hours to build the
> > set?
> 
> Not really, I added them in several goes and the build scripts have
> significant pauses between passes.  It also takes some time to mirror out
> the results between passes.  Tomorrow's build should be more
> representative.

It looks more like 2 hours 10 minutes or so.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]