Daily report for head: tnaur x86 Linux head: fail (failed stage1)

[Simon Marlow]

Thorkil Naur wrote:
> Hello,
> 
> On Tuesday 31 July 2007 13:05, Thorkil Naur wrote:
>> Hello,
>>
>> On Tuesday 31 July 2007 09:30, BuildBot Collator wrote:
>>> ...
>>> tnaur x86 Linux head:          fail (failed stage1)
>>> ...
>> The latest 4 builds (38, 37, 36, 35) failed building stage 1 with 
>>
> (http://darcs.haskell.org/buildbot/head/tnaur%20x86%20Linux%20head/builds/38/step-stage1/0):
>>> ...
>>> ../compiler/ghc-inplace -H16m -O  -static -I../gmp/gmpbuild -I. -#include 
>> HCIncludes.h -fvia-C -dcmm-lint  -hisuf debug_hi -hcsuf debug_hc -osuf 
>> debug_o -optc-DDEBUG   -c Updates.cmm -o Updates.debug_o
>>> ../utils/genapply/genapply  >AutoApply_debug.cmm
>>> ../compiler/ghc-inplace -H16m -O  -static -I../gmp/gmpbuild -I. -#include 
>> HCIncludes.h -fvia-C -dcmm-lint  -hisuf debug_hi -hcsuf debug_hc -osuf 
>> debug_o -optc-DDEBUG   -c AutoApply_debug.cmm -o AutoApply_debug.debug_o
>>> Prologue junk?: .globl stg_ap_p_ret
>>> 	.type	stg_ap_p_ret, @function
>>> stg_ap_p_ret:
>>> 	pushl	%ebp
>>> 	movl	%esp, %ebp
>>>
>>> make[2]: *** [AutoApply_debug.debug_o] Error 255
>>> make[1]: *** [all] Error 1
>>> make: *** [stage1] Error 1
>>> program finished with exit code 2
>> Build 34 failed earlier. Build 33 succeeded.
>>
>> I haven't tried to figure out what goes on here (my guess is some code 
>> generation error and/or an error/limitation in the mangler). Please indicate 
>> if there is something I can do to investigate further.
>>
>> Thanks and best regards
>> Thorkil
>>
>> _______________________________________________
>> Cvs-ghc mailing list
>> Cvs-ghc at haskell.org
>> http://www.haskell.org/mailman/listinfo/cvs-ghc
>>
> 
> I have tried to unpull patches back to, and including,
> 
> < Fri Jul 27 12:41:57 CEST 2007  Simon Marlow <simonmar at microsoft.com>
> <   * Pointer Tagging
> 
> and the above error disappeared. (Or, rather, it was replced by another error, 
> reported somewhat further along in the build process.) I then tried to pull 
> the "Pointer Tagging" patch back in, and the error re-appeared. Additional 
> details are available if desired.

In a tree with that patch, could you try running the fast testsuite using 
the stage1 compiler?  I want to find a small testcase that fails, for 
debugging.

Then if you or someone else could give me remote access to a PPC machine 
that I can use to debug this, that would be great.

Cheers,
	Simon