Re: [GHC] #1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?)
#1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?) --+- Reporter: Isaac Dupree | Owner: Type: bug | Status: new Priority: normal| Milestone: Component: Compiler |Version: 6.6 Severity: normal| Resolution: Keywords:| Difficulty: Unknown Testcase:| Architecture: powerpc Os: Linux | --+- Comment (by Isaac Dupree): Problem also reproducible with 6.7.20070418 for me. -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/1282 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?)
#1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?) --+- Reporter: Isaac Dupree | Owner: Type: bug | Status: new Priority: normal| Milestone: Component: Compiler |Version: 6.6 Severity: normal| Resolution: Keywords:| Difficulty: Unknown Testcase:| Architecture: powerpc Os: Linux | --+- Comment (by Isaac Dupree): Okay, looking at different versions of the GHC source-code, it looks like the problem is that the stg_* functions have been renamed hs_* (e.g. see foreign imports in libraries/base/GHC/Word.hs, or libraries/base/include/HsBase.h and libraries/base/cbits/longlong.c), and so the binary library is trying to import uncheckedShiftL64 and uncheckedShiftRL64 using the C-names that are used in any so-far-released version of GHC, that start with stg_. -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/1282 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?)
#1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?) --+- Reporter: Isaac Dupree | Owner: Type: bug | Status: new Priority: normal| Milestone: Component: Compiler |Version: 6.6 Severity: normal| Resolution: Keywords:| Difficulty: Unknown Testcase:| Architecture: powerpc Os: Linux | --+- Changes (by Isaac Dupree): * reporter: Isaac Dupree [EMAIL PROTECTED] = Isaac Dupree Comment: Okay, there is an _error_ from my yhc buildbot because it's actually trying to link the files. I can run the command in that yhc directory now and get the same error. If I instead use the same arguments, but with the system ghc-6.6, it succeeds in producing an executable (which segfaults because it's linking in libraries compiled by a different ghc, of course). Also, using a ghc-6.6 that I built sometime (without mtl, so I deleted `-package mtl` from my command-line and ignored undefined references to mtl parts), it did not report those stg_uncheckedShift things (nor indeed anything besides mtl) to be undefined. (the Version field of this ticket isn't correct... there doesn't seem to be a 6.6 branch option...) -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/1282 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
[GHC] #1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC 6.6.20070415 (?)
#1282: stg_uncheckedShift* not found: regression from ghc-6.6 to ghc-6.6.1 RC
6.6.20070415 (?)
--+-
Reporter: Isaac Dupree [EMAIL PROTECTED] | Owner:
Type: bug| Status:
new
Priority: normal | Milestone:
Component: Compiler | Version:
6.6
Severity: normal |Keywords:
Difficulty: Unknown|Testcase:
Architecture: powerpc| Os:
Linux
--+-
{{{
darcs get http://darcs.haskell.org/binary
cd binary
ghc --make Setup.lhs
./Setup configure --ghc [--with-
compiler=/home/isaac/install/ghc-6.6.20070415/bin/ghc]
}}}
./Setup build # 6.6
{{{
Preprocessing library binary-0.3...
Building binary-0.3...
[1 of 4] Compiling Data.Binary.Builder ( src/Data/Binary/Builder.hs,
dist/build/Data/Binary/Builder.o )
[2 of 4] Compiling Data.Binary.Get ( src/Data/Binary/Get.hs,
dist/build/Data/Binary/Get.o )
[3 of 4] Compiling Data.Binary.Put ( src/Data/Binary/Put.hs,
dist/build/Data/Binary/Put.o )
[4 of 4] Compiling Data.Binary ( src/Data/Binary.hs,
dist/build/Data/Binary.o )
...
}}}
./Setup build # 6.6.20070415
{{{
Preprocessing library binary-0.3...
Building binary-0.3...
[1 of 4] Compiling Data.Binary.Builder ( src/Data/Binary/Builder.hs,
dist/build/Data/Binary/Builder.o )
/tmp/ghc3685_0/ghc3685_0.hc: In function ‘s3cf_ret’:
/tmp/ghc3685_0/ghc3685_0.hc:1568:0:
warning: implicit declaration of function
‘stg_uncheckedShiftRL64’
[2 of 4] Compiling Data.Binary.Get ( src/Data/Binary/Get.hs,
dist/build/Data/Binary/Get.o )
/tmp/ghc3685_0/ghc3685_0.hc: In function ‘s5Cn_ret’:
/tmp/ghc3685_0/ghc3685_0.hc:609:0:
warning: implicit declaration of function ‘stg_uncheckedShiftL64’
[3 of 4] Compiling Data.Binary.Put ( src/Data/Binary/Put.hs,
dist/build/Data/Binary/Put.o )
[4 of 4] Compiling Data.Binary ( src/Data/Binary.hs,
dist/build/Data/Binary.o )
/tmp/ghc3685_0/ghc3685_0.hc: In function ‘sdYB_ret’:
/tmp/ghc3685_0/ghc3685_0.hc:1142:0:
warning: implicit declaration of function
‘stg_uncheckedShiftRL64’
/tmp/ghc3685_0/ghc3685_0.hc: In function ‘rbKZ_entry’:
/tmp/ghc3685_0/ghc3685_0.hc:2759:0:
warning: implicit declaration of function ‘stg_uncheckedShiftL64’
/usr/bin/ar: creating dist/build/libHSbinary-0.3.a
}}}
See http://www.indiegigs.co.uk:8010/linux_ppc/builds/111/step-yhc/0
errors at the end from ghc-6.6.20070415 on the same binary library for
the Yhc buildbot. I don't know if this error is a regression versus 6.6
and I can't figure out how to reproduce it outside my Yhc-buildbot (I
tried building the library with a wrapper shell script ghc-6.6.20070415
$@ -O0 -fasm, and it succeeded building the Binary library without any
warnings OR errors, so I don't know what's going on...)
(I'm not entirely sure, since Ubuntu 7.04 packagers build my 6.6, so maybe
they did something to it that made it work better -- made it
unregisterised, at least, and unable to use ghci or produce -fasm output.
But the ghc-6.6.20070415 warning messages above, from the .hc look to be
-fvia-C, and produced by the C compiler before any mangler starts on the
produced assembly...)
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/1282
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
