> but the latest command failed with the following messages:
> --------------------------------------------------------------
> ---------------
> [...]
> ../../../ghc/driver/ghc -i../misc -recomp -cpp -fglasgow-exts
> -fvia-C -Rghc-timing -O -split-objs -odir PosixTTY
> '-#include"cbits/libposix.h"' -monly-2-regs -c PosixTTY.lhs
> -o PosixTTY.o -osuf o
> <<ghc: 263578000 bytes, 331 GCs, 2803122/5866356 avg/max
> bytes residency (9 samples), 17M in use, 0.01 INIT (0.02
> elapsed), 8.71 MUT (9.11 elapsed), 5.39 GC (5.74 elapsed) :ghc>>
> ghc: module version changed to 1; reason: no old .hi file
> Epilogue junk?:
> popl %ecx
> .Lfe337:
> .size c91T_ret,.Lfe337-c91T_ret
> .section .rodata
> .align 4
> .type c91S_info,@object
> .size c91S_info,12
This is a quirk specific to recent versions of gcc or egcs. The relevant
patch is to fptools/ghc/driver/ghc-asm.lprl:
*** ghc-asm.lprl 1999/03/15 16:31:02 1.28
--- ghc-asm.lprl 1999/03/18 11:53:34 1.29
***************
*** 703,708 ****
--- 703,709 ----
$e =~ s/^\tret\n//;
$e =~ s/^\tpopl \%edi\n//;
$e =~ s/^\tpopl \%esi\n//;
+ $e =~ s/^\tpopl \%ecx\n//;
$e =~ s/^\taddl \$\d+,\%esp\n//;
} elsif ($TargetPlatform =~ /^m68k-/) {
$e =~ s/^\tunlk a6\n//;
Since this has cropped up a couple of times now, I've made a note of it on
the 4.02 web pages.
Cheers,
Simon
