Re: [GHC] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work +--- Reporter: [EMAIL PROTECTED] | Owner: Type: bug | Status: closed Priority: normal | Milestone: 6.6.1 Component: Runtime System |Version: 6.6 Severity: normal | Resolution: fixed Keywords: | Difficulty: Unknown Testcase: N/A | Architecture: x86 Os: Solaris | +--- Changes (by igloo): * resolution: = fixed * status: new = closed Comment: The only open issue is now bug #1015. -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993 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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by [EMAIL PROTECTED]):
Well spotted! That fixes the problem. (I did not look at Schedule
anymore.)
The testsuite (for ghc-6.6) still showed problems for
{{{concprog002(threaded2)}}} (looping) and {{{regex001(ghci,threaded1)}}}.
regex001(threaded1) failed with:
{{{
Starting program: /home/maeder/haskell/pc-solaris/ghc-6.6/testsuite/tests
/ghc-regress/lib/Regex/regex001
warning: Lowest section in /lib/libpthread.so.1 is .dynamic at 0074
[New LWP 1]
[New LWP 2]
[New LWP 3]
[New LWP 4]
Program received signal SIGSEGV, Segmentation fault.
[Switching to LWP 3]
0xbfd4386a in _free_unlocked () from /lib/libc.so.1
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
Indeed, that avoided the immediate crash. I'll try a completely new
ghc-6.6 build now.
GHC schrieb:
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by simonmar):
Looking at the manual for GNU as, it doesn't say anything about using a
slash (/) between the prefix and instruction mnemonic, so you might try
changing that code to
{{{
#if i386_HOST_ARCH || x86_64_HOST_ARCH
__asm__ __volatile__ (
lock cmpxchg %3,%1
:=a(o), =m (*(volatile unsigned int *)p)
:0 (o), r (n));
return o;
}}}
in fact, I don't know where that slash came from!
___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by simonmar):
So apparently the inline asseembly did not generate the correct
instruction, because you have this:
{{{
0x08ca8dc7 scheduleDoGC+24: lock test %eax,%eax
}}}
which is why you get an illegal instruction error: the test instruction
cannot have a lock prefix.
I'm at a loss to know why this happens. Maybe you could look at the
assembly file Schedule.s and see if that looks reasonable, and if so then
look at the object file Schedule.o using objdump --disassemble and see if
the instruction looks ok there.
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by simonmar):
Looking at the manual for GNU as, it doesn't say anything about using a
slash (/) between the prefix and instruction mnemonic, so you might try
changing that code to
{{{
#if i386_HOST_ARCH || x86_64_HOST_ARCH
__asm__ __volatile__ (
lock cmpxchg %3,%1
:=a(o), =m (*(volatile unsigned int *)p)
:0 (o), r (n));
return o;
}}}
in fact, I don't know where that slash came from!
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by simonmar):
Strange, it appears to be failing at the inline assembly
{{{
#if i386_HOST_ARCH || x86_64_HOST_ARCH
__asm__ __volatile__ (
lock/cmpxchg %3,%1
:=a(o), =m (*(volatile unsigned int *)p)
:0 (o), r (n));
return o;
}}}
but since this is an i386 processor, I don't see why it shouldn't let you
execute that instruction. Perhaps you could disassemble the instruction
that causes the crash, and see if it looks plausible?
Incedentally, your workaround will probably lead to problems later,
because it'll default to a non-atomic cas() implementation.
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by [EMAIL PROTECTED]):
here are the testsuite results (without the extralibs being checked out):
{{{
OVERALL SUMMARY for test run started at Fri Nov 10 16:09:48 CET 2006
1521 total tests, which gave rise to
8218 test cases, of which
0 caused framework failures
1502 were skipped
6351 expected passes
45 expected failures
7 unexpected passes
313 unexpected failures
Unexpected passes:
cholewo-eval(profasm)
tc175(normal,opt,hpc,optasm,prof,profasm)
Unexpected failures:
Chan001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
GMapAssoc(prof,profasm)
GMapTop(prof,profasm)
MVar001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
Over(normal)
QSem001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
QSemN001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
SampleVar001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
Simple2(hpc,profasm)
Simple6(normal)
Simple7(normal)
Simple8(normal)
TH_exn(normal)
TH_fail(normal)
TH_recompile(normal)
arith011(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
base01(normal)
bytestring001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
bytestring004(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
bytestring005(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
cabal01(normal)
cabal02(normal)
cc005(normal,opt,hpc,optasm,prof,profasm)
cc008(normal,opt,hpc,optasm,prof,profasm)
cc012(normal,opt,hpc,optasm,prof,profasm)
cg057(prof,profasm)
cg058(ghci)
conc023(threaded2)
conc053(threaded2)
conc056(threaded1,threaded2)
conc063(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
concprog001(ghci)
concprog002(threaded2)
countReaders001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
drvfail002(normal)
drvfail003(normal)
drvfail006(normal)
drvfail008(normal)
ds043(normal)
ds060(normal)
ds061(normal)
enum01(hpc)
ffi-deriv1(normal,opt,hpc,optasm,prof,profasm)
gadt13(normal)
gadt7(normal)
ghci008(ghci)
ghci011(ghci)
ghci014(ghci)
ghci016(ghci)
ghcpkg01(normal)
ghcpkg03(normal)
ghcpkg04(normal)
hGetBuf001(threaded2)
ind2(hpc,prof,profasm)
joao-circular(ghci)
maessen_hashtab(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
mod133(normal)
mod174(normal)
mod175(normal)
mod177(normal)
mod49(normal)
net001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
net002(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
num012(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
openFile003(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
pkg02_b(normal,opt,hpc,optasm,prof,profasm)
prof001(normal)
prof002(normal)
read032(normal)
regex001(ghci,threaded1)
reify(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
rnfail028(normal)
simplrun006(opt)
tc162(normal)
tc168(normal,opt,hpc,optasm,prof,profasm)
tc183(normal,opt,hpc,optasm,prof,profasm)
tc215(normal,opt,hpc,optasm,prof,profasm)
tc216(normal,opt,hpc,optasm,prof,profasm)
tc217(normal,opt,hpc,optasm,prof,profasm)
tcfail032(normal)
tcfail055(normal)
tcfail077(normal)
tcfail083(normal)
tcfail115(normal)
tcfail126(normal)
tcfail132(normal)
tcfail140(normal)
tcrun022(hpc)
tcrun032(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
tree(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
uri001(normal,opt,hpc,optasm,prof,profasm,ghci,threaded1,threaded2)
utf8_002(normal)
utf8_003(normal)
utf8_004(normal)
utf8_005(normal)
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: N/A | Architecture: x86
Os: Solaris |
+---
Comment (by [EMAIL PROTECTED]):
more gdb output:
{{{
(gdb) info line
Line 105 of ../includes/SMP.h starts at address 0x8ca8dbb
scheduleDoGC+12
and ends at 0x8ca8dc8 scheduleDoGC+25.
(gdb) disas 0x8ca8dbb 0x8ca8dc8
Dump of assembler code from 0x8ca8dbb to 0x8ca8dc8:
0x08ca8dbb scheduleDoGC+12: mov$0x0,%edx
0x08ca8dc0 scheduleDoGC+17: mov$0x1,%ecx
0x08ca8dc5 scheduleDoGC+22: mov%edx,%eax
0x08ca8dc7 scheduleDoGC+24: lock test %eax,%eax
End of assembler dump.
(gdb) info line
Line 106 of ../includes/SMP.h
is at address 0x8ca9954 threadStackOverflow+49 but contains no code.
(gdb) info line
Line 107 of ../includes/SMP.h
is at address 0x8ca9954 threadStackOverflow+49 but contains no code.
(gdb) info line
Line 108 of ../includes/SMP.h
is at address 0x8ca9954 threadStackOverflow+49 but contains no code.
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work +--- Reporter: [EMAIL PROTECTED] | Owner: Type: bug | Status: new Priority: normal | Milestone: 6.6.1 Component: Runtime System |Version: 6.6 Severity: normal | Resolution: Keywords: | Difficulty: Unknown Testcase: N/A | Architecture: x86 Os: Solaris | +--- Changes (by igloo): * testcase: = N/A -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993 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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Solaris |
+---
Comment (by [EMAIL PROTECTED]):
I've linked the stage2 compiler of the HEAD with:
{{{ -threaded -debug -optl-L/usr/local/lib -optl-lbfd -optl-liberty -optl-
lintl }}}
and gdb showed the following location:
{{{
[New LWP 1]
[New LWP 2]
ghc-6.7: missing -Bdir option
[New LWP 3]
Program received signal SIGILL, Illegal instruction.
0x08ca8dc7 in scheduleDoGC (cap=0x0, task=0x8e275b8, force_major=rtsFalse)
at ../includes/SMP.h:105
105 __asm__ __volatile__ (
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
+---
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.6.1
Component: Runtime System |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Solaris |
+---
Comment (by [EMAIL PROTECTED]):
commenting out (actually renaming) {{{i386_HOST_ARCH}}} in
{{{includes/SMP.h}}} fixes the illegal instruction problem! The first test
cases of the test-suite look good.
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work --+- Reporter: [EMAIL PROTECTED] | Owner: Type: bug| Status: new Priority: normal | Milestone: Component: Compiler | Version: 6.6 Severity: normal |Keywords: Difficulty: Unknown|Testcase: Architecture: x86| Os: Solaris --+- the stage2 compiler produces Illegal Instruction as soon as libHSrts_thr.a is linked in -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993 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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
---+
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug| Status: new
Priority: normal | Milestone:
Component: Compiler |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Solaris|
---+
Comment (by [EMAIL PROTECTED]):
It is sufficient to comment out -threaded in compiler/Makefile.ghcbin.
The working stage2 compiler behaves as follows:
{{{
-bash-3.1$ compiler/stage2/ghc-inplace --make -threaded hello.hs
[1 of 1] Compiling Main ( hello.hs, hello.o )
Linking hello ...
-bash-3.1$ ./hello
hello
Illegal Instruction
-bash-3.1$ rm hello
-bash-3.1$ compiler/stage2/ghc-inplace --make hello.hs
Linking hello ...
-bash-3.1$ ./hello
hello
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
---+
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug| Status: new
Priority: normal | Milestone:
Component: Compiler |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Solaris|
---+
Comment (by [EMAIL PROTECTED]):
During compilation I get the following warning
{{{
RetainerProfile.c: In function 'retainClosure':
RetainerProfile.c:435:0:
warning: 'se.info.type' may be used uninitialized in this function
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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] #993: threaded RTS under pc-solaris does not work
#993: threaded RTS under pc-solaris does not work
---+
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug| Status: new
Priority: normal | Milestone:
Component: Compiler |Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Solaris|
---+
Comment (by [EMAIL PROTECTED]):
just to complete the list of compiler warnings:
{{{
../../compiler/ghc-inplace -H16m -O -fglasgow-exts -cpp -Iinclude
-#include Hs
Base.h -funbox-strict-fields -package-name base-2.0 -O -Rghc-timing
-fgenerics
-fgenerics -split-objs-c GHC/Conc.lhs -o GHC/Conc.o -ohi GHC/Conc.hi
/tmp/ghc27985_0/ghc27985_0.hc: In function 'base_GHCziConc_par_entry':
/tmp/ghc27985_0/ghc27985_0.hc:3339:0:
warning: implicit declaration of function 'newSpark'
}}}
{{{
../../compiler/ghc-inplace -H16m -O -fglasgow-exts -cpp -Iinclude
-#include Hs
Base.h -funbox-strict-fields -package-name base-2.0 -O -Rghc-timing
-fgenerics
-fgenerics -split-objs-c Control/Parallel/Strategies.hs -o
Control/Parallel
/Strategies.o -ohi Control/Parallel/Strategies.hi
/tmp/ghc2134_0/ghc2134_0.hc: In function 's2D5_entry':
/tmp/ghc2134_0/ghc2134_0.hc:171:0:
warning: implicit declaration of function 'newSpark'
}}}
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/993
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
