I've now committed the proposed changes,
see tickets: #32989, #32514, #34072
If you have a PPC based box, please give it a try.
Thanks,
leo
Le Thursday 3 March 2005 11:56, Leopold Toetsch a =E9crit :
> I've now committed the proposed changes,
> see tickets: #32989, #32514, #34072
>
> If you have a PPC based box, please give it a try.
My ppc box is a Mandrake cooker (devel distro): I ran:
Unfortunally, I don't have the C/asm knowledge to help more :(
cvs update -dC
make clean
perl Configure.PL
make
ops/core_ops_switch.c
as -o src/asmfun.o src/asmfun.s
src/asmfun.s: Assembler messages:
src/asmfun.s:6: Error: unsupported relocation against f14
src/asmfun.s:6: Error: unsupported relocation against r1
src/asmfun.s:7: Error: unsupported relocation against f15
src/asmfun.s:7: Error: unsupported relocation against r1
src/asmfun.s:8: Error: unsupported relocation against f16
src/asmfun.s:8: Error: unsupported relocation against r1
src/asmfun.s:9: Error: unsupported relocation against f17
src/asmfun.s:9: Error: unsupported relocation against r1
src/asmfun.s:10: Error: unsupported relocation against f18
src/asmfun.s:10: Error: unsupported relocation against r1
src/asmfun.s:11: Error: unsupported relocation against f19
src/asmfun.s:11: Error: unsupported relocation against r1
src/asmfun.s:12: Error: unsupported relocation against f20
src/asmfun.s:12: Error: unsupported relocation against r1
src/asmfun.s:13: Error: unsupported relocation against f21
src/asmfun.s:13: Error: unsupported relocation against r1
src/asmfun.s:14: Error: unsupported relocation against f22
src/asmfun.s:14: Error: unsupported relocation against r1
src/asmfun.s:15: Error: unsupported relocation against f23
src/asmfun.s:15: Error: unsupported relocation against r1
src/asmfun.s:16: Error: unsupported relocation against f24
src/asmfun.s:16: Error: unsupported relocation against r1
src/asmfun.s:17: Error: unsupported relocation against f25
src/asmfun.s:17: Error: unsupported relocation against r1
src/asmfun.s:18: Error: unsupported relocation against f26
src/asmfun.s:18: Error: unsupported relocation against r1
src/asmfun.s:19: Error: unsupported relocation against f27
src/asmfun.s:19: Error: unsupported relocation against r1
src/asmfun.s:20: Error: unsupported relocation against f28
src/asmfun.s:20: Error: unsupported relocation against r1
src/asmfun.s:21: Error: unsupported relocation against f29
src/asmfun.s:21: Error: unsupported relocation against r1
src/asmfun.s:22: Error: unsupported relocation against f30
src/asmfun.s:22: Error: unsupported relocation against r1
src/asmfun.s:23: Error: unsupported relocation against f31
src/asmfun.s:23: Error: unsupported relocation against r1
make: *** [src/asmfun.o] Error 1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQBCJznCk29cDOWzfVARArxdAKCS46BzxMAzLf3/5BXfXpyEZcNvjwCeKwZ1
KMBP75zITOxPnWKmgqpIMNQ=
=6rvG
-----END PGP SIGNATURE-----
On Thu, 2005-03-03 at 17:22 +0100, Olivier Thauvin wrote:
> ops/core_ops_switch.c
> as -o src/asmfun.o src/asmfun.s
> src/asmfun.s: Assembler messages:
> src/asmfun.s:6: Error: unsupported relocation against f14
Adding -mregnames to the as call makes it compile for me, but it
apparently doesn't export the symbol properly:
blib/lib/libparrot.a(jit_cpu.o)(.text+0x2722): In function
`Parrot_end_jit':
src/jit_cpu.c:74: undefined reference to
`Parrot_ppc_jit_restore_nonvolatile_registers'
blib/lib/libparrot.a(jit_cpu.o)(.text+0x2726):src/jit_cpu.c:74:
undefined reference to `Parrot_ppc_jit_restore_nonvolatile_registers
-- c
On Thu, 2005-03-03 at 09:30 -0800, chromatic wrote:
> Adding -mregnames to the as call makes it compile for me, but it
> apparently doesn't export the symbol properly:
Ah, here it is. For whatever reason, it doesn't strip the leading
underscore from the symbol name. (GNU as allows symbols to start with
word characters, but I don't see a flag to remove a prefix.)
Compiling the file manually with the -mregnames flag and applying this
patch fixes the build for me.
-- c
Index: jit/ppc/asm.s
===================================================================
RCS file: /cvs/public/parrot/jit/ppc/asm.s,v
retrieving revision 1.1
diff -u -u -r1.1 asm.s
--- jit/ppc/asm.s 3 Mar 2005 10:55:37 -0000 1.1
+++ jit/ppc/asm.s 3 Mar 2005 17:33:12 -0000
@@no-spam -1,7 +1,7 @@no-spam
.text
.align 12
- .globl _Parrot_ppc_jit_restore_nonvolatile_registers
-_Parrot_ppc_jit_restore_nonvolatile_registers:
+ .globl Parrot_ppc_jit_restore_nonvolatile_registers
+Parrot_ppc_jit_restore_nonvolatile_registers:
lfd f14,-84(r1)
lfd f15,-92(r1)
[plaintext linux_ppc_asm.patch]
Chromatic <chromatic@no-spam> wrote:
> On Thu, 2005-03-03 at 09:30 -0800, chromatic wrote:
>> Adding -mregnames to the as call makes it compile for me, but it
>> apparently doesn't export the symbol properly:
> Ah, here it is. For whatever reason, it doesn't strip the leading
> underscore from the symbol name. (GNU as allows symbols to start with
> word characters, but I don't see a flag to remove a prefix.)
> Compiling the file manually with the -mregnames flag and applying this
> patch fixes the build for me.
Ok. I've duplicate jit/ppc/asm.s to jit/ppc/ppc-linux.s, removed the
leading underscore and the register name prefixes.
config/auto/jit.pl should now use this file for the platform
"ppc-linux".
> -- c
Thanks to Olivier and your for providing feedback,
leo
On Fri, 4 Mar 2005 08:48:06 +0100, Leopold Toetsch <lt@no-spam> wrote:
> Ok. I've duplicate jit/ppc/asm.s to jit/ppc/ppc-linux.s, removed the
> leading underscore and the register name prefixes.
>
> config/auto/jit.pl should now use this file for the platform
> "ppc-linux".
Can someone that #32514 can be closed?
https://rt.perl.org/rt3/Ticket/Display.html?id=32989
And #27414 along with it:
https://rt.perl.org/rt3/Ticket/Display.html?id=27414
Thanks.
--
matt diephouse
http://matt.diephouse.com
On Fri, 4 Mar 2005 16:24:30 -0500, Matt Diephouse <mddiep@no-spam> wrote:
> On Fri, 4 Mar 2005 08:48:06 +0100, Leopold Toetsch <lt@no-spam> wrote:
> > Ok. I've duplicate jit/ppc/asm.s to jit/ppc/ppc-linux.s, removed the
> > leading underscore and the register name prefixes.
> >
> > config/auto/jit.pl should now use this file for the platform
> > "ppc-linux".
>
> Can someone that #32514 can be closed?
Eek. That should read "Can someone *confirm* that #32514 can be closed?".
> https://rt.perl.org/rt3/Ticket/Display.html?id=32989
>
> And #27414 along with it:
>
> https://rt.perl.org/rt3/Ticket/Display.html?id=27414
>
Sorry.
--
matt diephouse
http://matt.diephouse.com
On Fri, 2005-03-04 at 16:24 -0500, Matt Diephouse wrote:
> Can someone that #32514 can be closed?
>
> https://rt.perl.org/rt3/Ticket/Display.html?id=32989
Yes, close it.
> And #27414 along with it:
>
> https://rt.perl.org/rt3/Ticket/Display.html?id=27414
I don't see the failures (and have never seen them). Both of the tests
mentioned there pass with make testj. (Only t/pmc/eval.t #4 fails, and
I'll look at that later.)
-- c