This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Failure introduced by the sahf patch(es)
H. J. Lu wrote:
There has been a recent addition to add this "sahf" to the
instructions used by the i386 back-end.
This is mine, but from asm dump:
.L120:
fprem
fnstsw %ax
sahf
jp .L120
fstp %st(1)
xorpd %xmm2, %xmm2
This is perfectly good asm for MOD() a.k.a. fmod() function.
64bit sahf was enabled by
http://sourceware.org/ml/binutils/2004-11/msg00313.html
I think we should either use .byte with a commment and check if
assembler supports 64bit sahf if we want to generate 64bit sahf.
Attached patch fixes this annoyance.
2007-03-14 Uros Bizjak <ubizjak@gmail.com>
* configure.ac (HAVE_AS_IX86_SAHF): On x86 targets check whether
the configured assembler supports the sahf mnemonic.
* configure: Regenerate.
* config.in: Regenerate.
* config/i386/i386.md (x86_sahf_1): Depending on HAVE_AS_IX86_SAHF,
emit "sahf" or ".byte\t0x9e" as asm template.
Patch is regression tested on x86_64-pc-linux-gnu for all default
languages. Patch is commited to SVN.
Uros.
Index: config/i386/i386.md
===================================================================
--- config/i386/i386.md (revision 122924)
+++ config/i386/i386.md (working copy)
@@ -988,7 +988,7 @@
(unspec:CC [(match_operand:HI 0 "register_operand" "a")]
UNSPEC_SAHF))]
"TARGET_SAHF"
- "sahf"
+ "* return HAVE_AS_IX86_SAHF ? \"sahf\" : \".byte\t0x9e\";"
[(set_attr "length" "1")
(set_attr "athlon_decode" "vector")
(set_attr "amdfam10_decode" "direct")
Index: config.in
===================================================================
--- config.in (revision 122924)
+++ config.in (working copy)
@@ -228,6 +228,12 @@
#endif
+/* Define if your assembler supports the sahf mnemonic. */
+#ifndef USED_FOR_TARGET
+#undef HAVE_AS_IX86_SAHF
+#endif
+
+
/* Define if your assembler supports the lituse_jsrdirect relocation. */
#ifndef USED_FOR_TARGET
#undef HAVE_AS_JSRDIRECT_RELOCS
Index: configure
===================================================================
--- configure (revision 122924)
+++ configure (working copy)
@@ -15237,6 +15237,39 @@
fi
+ echo "$as_me:$LINENO: checking assembler for sahf mnemonic" >&5
+echo $ECHO_N "checking assembler for sahf mnemonic... $ECHO_C" >&6
+if test "${gcc_cv_as_ix86_sahf+set}" = set; then
+ echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+ gcc_cv_as_ix86_sahf=no
+ if test x$gcc_cv_as != x; then
+ echo 'sahf' > conftest.s
+ if { ac_try='$gcc_cv_as -o conftest.o conftest.s >&5'
+ { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+ (eval $ac_try) 2>&5
+ ac_status=$?
+ echo "$as_me:$LINENO: \$? = $ac_status" >&5
+ (exit $ac_status); }; }
+ then
+ gcc_cv_as_ix86_sahf=yes
+ else
+ echo "configure: failed program was" >&5
+ cat conftest.s >&5
+ fi
+ rm -f conftest.o conftest.s
+ fi
+fi
+echo "$as_me:$LINENO: result: $gcc_cv_as_ix86_sahf" >&5
+echo "${ECHO_T}$gcc_cv_as_ix86_sahf" >&6
+if test $gcc_cv_as_ix86_sahf = yes; then
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_AS_IX86_SAHF 1
+_ACEOF
+
+fi
+
echo "$as_me:$LINENO: checking assembler for different section symbol subtraction" >&5
echo $ECHO_N "checking assembler for different section symbol subtraction... $ECHO_C" >&6
if test "${gcc_cv_as_ix86_diff_sect_delta+set}" = set; then
Index: configure.ac
===================================================================
--- configure.ac (revision 122924)
+++ configure.ac (working copy)
@@ -2808,6 +2808,12 @@
[AC_DEFINE(HAVE_AS_IX86_FFREEP, 1,
[Define if your assembler supports the ffreep mnemonic.])])
+ gcc_GAS_CHECK_FEATURE([sahf mnemonic],
+ gcc_cv_as_ix86_sahf,,,
+ [sahf],,
+ [AC_DEFINE(HAVE_AS_IX86_SAHF, 1,
+ [Define if your assembler supports the sahf mnemonic.])])
+
gcc_GAS_CHECK_FEATURE([different section symbol subtraction],
gcc_cv_as_ix86_diff_sect_delta,,,
[.section .rodata