* [Xenomai] ipipe tracer call trace during boot
[not found] <1019176459.3312173.1454670838251.JavaMail.yahoo.ref@mail.yahoo.com>
@ 2016-02-05 11:13 ` Frederik Bayart
2016-02-06 13:46 ` Gilles Chanteperdrix
0 siblings, 1 reply; 6+ messages in thread
From: Frederik Bayart @ 2016-02-05 11:13 UTC (permalink / raw)
To: xenomai
I have build an 3.18.20 kernel with xenomai 2.6.4 git latest with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64 (64bit).
I try to build an version with ipipe tracer.
If I try to boot, I get a call trace : in __ipipe_init_tracer :
RIP __ipipe_tracer_hrclock_initialized
I tried already :
- disabling tracing on boot
- enabling vmalloc'ed trace buffer
- reducing depth of trace log from 14 to 10
In attachment :
- my latest kernel config
- a screenshot from the call trace
Does anyone have a suggestion what could the problem ?
I noticed there was an issue with ipipe/tracer: allow hrclock late init (http://xenomai.org/pipermail/xenomai/2015-September/035172.html) but f I'm correct this patch is integrated in ipipe-core-3.18.20-x86-4.patch
Kind regards,
Frederik
-------------- next part --------------
A non-text attachment was scrubbed...
Name: config-3.18.20-amd64-ipipe-tracer
Type: application/octet-stream
Size: 115303 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20160205/a8210c32/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IMG_20160204_110412.jpg
Type: image/jpeg
Size: 1471099 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20160205/a8210c32/attachment.jpg>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Xenomai] ipipe tracer call trace during boot
2016-02-05 11:13 ` [Xenomai] ipipe tracer call trace during boot Frederik Bayart
@ 2016-02-06 13:46 ` Gilles Chanteperdrix
2016-02-08 14:54 ` Frederik Bayart
0 siblings, 1 reply; 6+ messages in thread
From: Gilles Chanteperdrix @ 2016-02-06 13:46 UTC (permalink / raw)
To: Frederik Bayart; +Cc: xenomai
On Fri, Feb 05, 2016 at 11:13:56AM +0000, Frederik Bayart wrote:
> I have build an 3.18.20 kernel with xenomai 2.6.4 git latest with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64 (64bit).
>
> I try to build an version with ipipe tracer.
>
> If I try to boot, I get a call trace : in __ipipe_init_tracer :
>
> RIP __ipipe_tracer_hrclock_initialized
Hi,
the problem is probably the commit you found. Will try your
configuration soon.
Regards.
--
Gilles.
https://click-hack.org
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Xenomai] ipipe tracer call trace during boot
2016-02-06 13:46 ` Gilles Chanteperdrix
@ 2016-02-08 14:54 ` Frederik Bayart
2016-02-08 22:49 ` Gilles Chanteperdrix
0 siblings, 1 reply; 6+ messages in thread
From: Frederik Bayart @ 2016-02-08 14:54 UTC (permalink / raw)
To: xenomai
> On Saturday, 6 February 2016, 14:46, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
>
> On Fri, Feb 05, 2016 at 11:13:56AM +0000, Frederik Bayart wrote:
>
> > I have build an 3.18.20 kernel with xenomai 2.6.4 git latest with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64 (64bit).
> >
> > I try to build an version with ipipe tracer.
> >
> > If I try to boot, I get a call trace : in __ipipe_init_tracer :
> >
> > RIP __ipipe_tracer_hrclock_initialized
>
> Hi,
>
> the problem is probably the commit you found. Will try your
> configuration soon.
>
> Regards.
> --
> Gilles.
I tried with ipipe-core-3.18.20-x86-2.patch with doesn't contain that commit and I have still a call trace :
RIP __ipipe_init_tracer
Frederik
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Xenomai] ipipe tracer call trace during boot
2016-02-08 14:54 ` Frederik Bayart
@ 2016-02-08 22:49 ` Gilles Chanteperdrix
2016-02-09 10:23 ` Frederik Bayart
2016-02-19 10:14 ` Henning Schild
0 siblings, 2 replies; 6+ messages in thread
From: Gilles Chanteperdrix @ 2016-02-08 22:49 UTC (permalink / raw)
To: Frederik Bayart; +Cc: xenomai
On Mon, Feb 08, 2016 at 02:54:12PM +0000, Frederik Bayart wrote:
> > On Saturday, 6 February 2016, 14:46, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
> >
> > On Fri, Feb 05, 2016 at 11:13:56AM +0000, Frederik Bayart wrote:
> >
> > > I have build an 3.18.20 kernel with xenomai 2.6.4 git latest with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64 (64bit).
> > >
> > > I try to build an version with ipipe tracer.
> > >
> > > If I try to boot, I get a call trace : in __ipipe_init_tracer :
> > >
> > > RIP __ipipe_tracer_hrclock_initialized
> >
> > Hi,
> >
> > the problem is probably the commit you found. Will try your
> > configuration soon.
> >
> > Regards.
> > --
> > Gilles.
>
>
> I tried with ipipe-core-3.18.20-x86-2.patch with doesn't contain that commit and I have still a call trace :
>
> RIP __ipipe_init_tracer
Hi,
I could reproduce this issue. Running "addr2line" on the RIP points
to __BUILTIN_RETURN_ADDRESS1. I tried enabling frame pointers
(CONFIG_FRAME_POINTER=y) and the kernel does not crash. I suspect
__builtin_return_address(1) relies in fact on frame pointers.
Regards.
--
Gilles.
https://click-hack.org
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Xenomai] ipipe tracer call trace during boot
2016-02-08 22:49 ` Gilles Chanteperdrix
@ 2016-02-09 10:23 ` Frederik Bayart
2016-02-19 10:14 ` Henning Schild
1 sibling, 0 replies; 6+ messages in thread
From: Frederik Bayart @ 2016-02-09 10:23 UTC (permalink / raw)
To: xenomai
> On Monday, 8 February 2016, 23:49, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
> On Mon, Feb 08, 2016 at 02:54:12PM +0000, Frederik Bayart wrote:
> > > On Saturday, 6 February 2016, 14:46, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
> > >
> > > On Fri, Feb 05, 2016 at 11:13:56AM +0000, Frederik Bayart wrote:
> > >
> > > > I have build an 3.18.20 kernel with xenomai 2.6.4 git latest with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64 (64bit).
> > > >
> > > > I try to build an version with ipipe tracer.
> > > >
> > > > If I try to boot, I get a call trace : in __ipipe_init_tracer :
> > > >
> > > > RIP __ipipe_tracer_hrclock_initialized
> > >
> > > Hi,
> > >
> > > the problem is probably the commit you found. Will try your
> > > configuration soon.
> > > --
> > > Gilles.
> >
> > I tried with ipipe-core-3.18.20-x86-2.patch with doesn't contain that commit and I have still a call trace :
> >
> > RIP __ipipe_init_tracer
>
> Hi,
>
> I could reproduce this issue. Running "addr2line" on the RIP points
> to __BUILTIN_RETURN_ADDRESS1. I tried enabling frame pointers
> (CONFIG_FRAME_POINTER=y) and the kernel does not crash. I suspect
> __builtin_return_address(1) relies in fact on frame pointers.
> --
> Gilles.
Problem solved. Thanx for the help.
Frederik
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Xenomai] ipipe tracer call trace during boot
2016-02-08 22:49 ` Gilles Chanteperdrix
2016-02-09 10:23 ` Frederik Bayart
@ 2016-02-19 10:14 ` Henning Schild
1 sibling, 0 replies; 6+ messages in thread
From: Henning Schild @ 2016-02-19 10:14 UTC (permalink / raw)
To: Gilles Chanteperdrix; +Cc: xenomai, Frederik Bayart
On Mon, 8 Feb 2016 23:49:30 +0100
Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
> On Mon, Feb 08, 2016 at 02:54:12PM +0000, Frederik Bayart wrote:
> > > On Saturday, 6 February 2016, 14:46, Gilles Chanteperdrix
> > > <gilles.chanteperdrix@xenomai.org> wrote:
> > >
> > > On Fri, Feb 05, 2016 at 11:13:56AM +0000, Frederik Bayart wrote:
> > >
> > > > I have build an 3.18.20 kernel with xenomai 2.6.4 git latest
> > > > with ipipe-core-3.18.20-x86-4.patch on debian jessie amd64
> > > > (64bit).
> > > >
> > > > I try to build an version with ipipe tracer.
> > > >
> > > > If I try to boot, I get a call trace : in __ipipe_init_tracer :
> > > >
> > > > RIP __ipipe_tracer_hrclock_initialized
> > >
> > > Hi,
> > >
> > > the problem is probably the commit you found. Will try your
> > > configuration soon.
> > >
> > > Regards.
> > > --
> > > Gilles.
> >
> >
> > I tried with ipipe-core-3.18.20-x86-2.patch with doesn't contain
> > that commit and I have still a call trace :
> >
> > RIP __ipipe_init_tracer
>
> Hi,
>
> I could reproduce this issue. Running "addr2line" on the RIP points
> to __BUILTIN_RETURN_ADDRESS1. I tried enabling frame pointers
> (CONFIG_FRAME_POINTER=y) and the kernel does not crash. I suspect
> __builtin_return_address(1) relies in fact on frame pointers.
The issue can be solved with the right config. Is that something we
should make more explicit? i.e. config depend
Did you also look into reproducing it on 4.1?
I could test 4.1 and provide patches.
Henning
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2016-02-19 10:14 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <1019176459.3312173.1454670838251.JavaMail.yahoo.ref@mail.yahoo.com>
2016-02-05 11:13 ` [Xenomai] ipipe tracer call trace during boot Frederik Bayart
2016-02-06 13:46 ` Gilles Chanteperdrix
2016-02-08 14:54 ` Frederik Bayart
2016-02-08 22:49 ` Gilles Chanteperdrix
2016-02-09 10:23 ` Frederik Bayart
2016-02-19 10:14 ` Henning Schild
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.