Re: linux-next: Tree for Oct 6

Re: linux-next: Tree for Oct 6

[Sudip Mukherjee]

On Tue, Oct 06, 2015 at 06:25:22PM +1100, Stephen Rothwell wrote:
> Hi all,
> 
> Changes since 20151002:

The build for cris allmodconfig is failing with the error:

net/sched/sch_dsmark.c: In function 'dsmark_dequeue':
net/sched/sch_dsmark.c:316:1: error: unrecognizable insn:
(insn 245 244 119 15 (set (reg:QI 11 r11 [179])
        (and:QI (mem/s/j:QI (reg/f:SI 2 r2 [orig:48 D.44939 ] [48]) [0 D.44939_34->mask+0 S1 A8])
            (reg:QI 11 r11 [179]))) include/net/dsfield.h:33 -1
     (nil))
net/sched/sch_dsmark.c:316:1: internal compiler error: in extract_insn, at recog.c:2109
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
make[2]: *** [net/sched/sch_dsmark.o] Error 1
make[1]: *** [net/sched] Error 2
make[1]: *** Waiting for unfinished jobs....

It says compiler error, but with 4.3-rc4 it compiled properly.
Build log of v4.3-rc4 is at
https://travis-ci.org/sudipm-mukherjee/parport/jobs/83633972

Build log of linux-next is at:
https://travis-ci.org/sudipm-mukherjee/parport/jobs/83839470

regards
sudip

Re: linux-next: Tree for Oct 6

[David Miller]

From: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Date: Tue, 6 Oct 2015 18:03:46 +0530

> It says compiler error, but with 4.3-rc4 it compiled properly.

That doesn't make it any less of a compiler bug.

Re: linux-next: Tree for Oct 6

[Hans-Peter Nilsson]

> From: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
> Date: Tue, 6 Oct 2015 14:33:46 +0200

> On Tue, Oct 06, 2015 at 06:25:22PM +1100, Stephen Rothwell wrote:
> > Hi all,
> > 
> > Changes since 20151002:
> 
> The build for cris allmodconfig is failing with the error:
> 
> net/sched/sch_dsmark.c: In function 'dsmark_dequeue':
> net/sched/sch_dsmark.c:316:1: error: unrecognizable insn:
> (insn 245 244 119 15 (set (reg:QI 11 r11 [179])
>         (and:QI (mem/s/j:QI (reg/f:SI 2 r2 [orig:48 D.44939 ] [48]) [0 D.44939_34->mask+0 S1 A8])
>             (reg:QI 11 r11 [179]))) include/net/dsfield.h:33 -1
>      (nil))
> net/sched/sch_dsmark.c:316:1: internal compiler error: in extract_insn, at recog.c:2109
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.
> make[2]: *** [net/sched/sch_dsmark.o] Error 1
> make[1]: *** [net/sched] Error 2
> make[1]: *** Waiting for unfinished jobs....
> 
> It says compiler error, but with 4.3-rc4 it compiled properly.
> Build log of v4.3-rc4 is at
> https://travis-ci.org/sudipm-mukherjee/parport/jobs/83633972
> 
> Build log of linux-next is at:
> https://travis-ci.org/sudipm-mukherjee/parport/jobs/83839470

Thanks for the heads-up, but any chance of a bug report as per
the instructions in the message?  It'll only speed up matters.

brgds, H-P

Re: linux-next: Tree for Oct 6

[Sudip Mukherjee]

On Tue, Oct 06, 2015 at 04:46:53PM +0200, Hans-Peter Nilsson wrote:
> > From: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
> > Date: Tue, 6 Oct 2015 14:33:46 +0200
> 
> > On Tue, Oct 06, 2015 at 06:25:22PM +1100, Stephen Rothwell wrote:
> > > Hi all,
> > > 
> > > Changes since 20151002:
> > 
> > The build for cris allmodconfig is failing with the error:
> > 
> > net/sched/sch_dsmark.c: In function 'dsmark_dequeue':
> > net/sched/sch_dsmark.c:316:1: error: unrecognizable insn:
> > (insn 245 244 119 15 (set (reg:QI 11 r11 [179])
> >         (and:QI (mem/s/j:QI (reg/f:SI 2 r2 [orig:48 D.44939 ] [48]) [0 D.44939_34->mask+0 S1 A8])
> >             (reg:QI 11 r11 [179]))) include/net/dsfield.h:33 -1
> >      (nil))
> > net/sched/sch_dsmark.c:316:1: internal compiler error: in extract_insn, at recog.c:2109
> > Please submit a full bug report,
> > with preprocessed source if appropriate.
> > See <http://gcc.gnu.org/bugs.html> for instructions.
> > make[2]: *** [net/sched/sch_dsmark.o] Error 1
> > make[1]: *** [net/sched] Error 2
> > make[1]: *** Waiting for unfinished jobs....
> > 
> > It says compiler error, but with 4.3-rc4 it compiled properly.
> > Build log of v4.3-rc4 is at
> > https://travis-ci.org/sudipm-mukherjee/parport/jobs/83633972
> > 
> > Build log of linux-next is at:
> > https://travis-ci.org/sudipm-mukherjee/parport/jobs/83839470
> 
> Thanks for the heads-up, but any chance of a bug report as per
> the instructions in the message?  It'll only speed up matters.
I can try, but since I am using 4.6.3 version I doubt if I will get the
required response (Happened with another arch). And this is the latest
cris cross-compiler available in
https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/

BTW, I did a bisect and if I revert 47bbbb30b433 ("sch_dsmark: improve memory locality")
it again builds without any error.

regards
sudip

Re: linux-next: Tree for Oct 6

[Hans-Peter Nilsson]

> From: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
> Date: Wed, 7 Oct 2015 09:10:46 +0200

> On Tue, Oct 06, 2015 at 04:46:53PM +0200, Hans-Peter Nilsson wrote:
> > Thanks for the heads-up, but any chance of a bug report as per
> > the instructions in the message?  It'll only speed up matters.
> I can try, but since I am using 4.6.3 version I doubt if I will get the
> required response (Happened with another arch).

Ow, way old.

> And this is the latest
> cris cross-compiler available in
> https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/

I have nothing to do with that.

> BTW, I did a bisect and if I revert 47bbbb30b433 ("sch_dsmark: improve memory locality")
> it again builds without any error.

Sorry, that's not really useful to me, neither are build logs.
What I need to proceed with fixing is preprocessed code and the
compiler options used, just as the bug-reporting instructions
say.  Now that I hear that you get the ICE with a really old gcc
version, I'm not inclined to look further, sorry.  Except, when
you mention the version I vaguely remember; it reminds me of a
bug fixed a long time ago.  Maybe it was PR53120?  Sorry, I
don't remember anything else useful.  Still, thanks; it'll
shorten the response time if other people tell me the same
thing.

brgds, H-P