| Message ID | 192d0b37e3b4eb47d0c9eab58a6f359cbf130453.camel@vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
| Series | Add ppc_ieee128_ok target-supports proc | expand |
Hi! On Mon, Feb 10, 2020 at 03:39:49PM -0600, will schmidt wrote: > This adds a target_supports entry to check that the > _ieee128 keyword is understood by the target. (Two underscores, "__ieee128".) > Also adds that require-effective-target check to the > existing pr92796 testcase. > +# See if the __ieee128 keyword is understood. > +proc check_effective_target_ppc_ieee128_ok { } { > + return [check_cached_effective_target ppc_ieee128_ok { > + # disable on Darwin, AIX. > + if { [istarget powerpc-*-eabi] > + || [istarget powerpc*-*-eabispe] > + || [istarget *-*-darwin*] > + || [istarget *-*-aix*] } { > + expr 0 Do we need to exclude those two explicitly? Why? And why eabi? And you don't need eabispe either. > + } else { > + set options "-mvsx" > + check_runtime_nocache ppc_ieee128_ok { > + int main() > + { > + __ieee128 a; > + return 0; > + } > + } $options > + } Why do you use -mvsx while we actually need a superset of that? Segher
Hi,
Thanks for the feedback. Updated below.
[v2]
- removed some explicit disables that were not necessary.
- updated options qualifier for the proc.
[V1]
Add a target_supports entry to check that the __ieee128 keyword
is understood by the target.
Also add a dg-requires check to the existing pr92796 testcase.
Sniff tested on Linux (power6,power9) and AIX.
OK for master?
[testsuite]
* lib/target-supports.exp (check_effective_target_ppc_ieee128_ok): New.
* gcc.target/powerpc/pr92796.c: Add a require-effective-target
statement for ppc_ieee128_ok.
diff --git a/gcc/testsuite/gcc.target/powerpc/pr92796.c b/gcc/testsuite/gcc.target/powerpc/pr92796.c
index aa15b2d..da4b6a4 100644
--- a/gcc/testsuite/gcc.target/powerpc/pr92796.c
+++ b/gcc/testsuite/gcc.target/powerpc/pr92796.c
@@ -1,7 +1,8 @@
/* { dg-do compile } */
/* { dg-options "-O2 -fstack-protector-strong -mcpu=power8" } */
+/* { dg-require-effective-target ppc_ieee128_ok } */
typedef union
{
__ieee128 a;
int b;
diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-supports.exp
index d3b2798..fb177c5 100644
--- a/gcc/testsuite/lib/target-supports.exp
+++ b/gcc/testsuite/lib/target-supports.exp
@@ -2248,10 +2248,29 @@ proc check_ppc_float128_hw_available { } {
} $options
}
}]
}
+# See if the __ieee128 keyword is understood.
+proc check_effective_target_ppc_ieee128_ok { } {
+ return [check_cached_effective_target ppc_ieee128_ok {
+ # disable on AIX.
+ if { [istarget *-*-aix*] } {
+ expr 0
+ } else {
+ set options "-mfloat128"
+ check_runtime_nocache ppc_ieee128_ok {
+ int main()
+ {
+ __ieee128 a;
+ return 0;
+ }
+ } $options
+ }
+ }]
+}
+
# Return 1 if the target supports executing VSX instructions, 0
# otherwise. Cache the result.
proc check_vsx_hw_available { } {
return [check_cached_effective_target vsx_hw_available {
Hi! On Tue, Feb 11, 2020 at 12:10:50PM -0600, will schmidt wrote: > +# See if the __ieee128 keyword is understood. > +proc check_effective_target_ppc_ieee128_ok { } { > + return [check_cached_effective_target ppc_ieee128_ok { > + # disable on AIX. > + if { [istarget *-*-aix*] } { > + expr 0 > + } else { > + set options "-mfloat128" > + check_runtime_nocache ppc_ieee128_ok { > + int main() > + { > + __ieee128 a; > + return 0; > + } > + } $options > + } > + }] > +} It would be better if you didn't need to special-case AIX here, if the actual test would figure out it cannot do __ieee128 on AIX. But, okay for trunk. Thanks! Segher
diff --git a/gcc/testsuite/gcc.target/powerpc/pr92796.c b/gcc/testsuite/gcc.target/powerpc/pr92796.c index aa15b2d..da4b6a4 100644 --- a/gcc/testsuite/gcc.target/powerpc/pr92796.c +++ b/gcc/testsuite/gcc.target/powerpc/pr92796.c @@ -1,7 +1,8 @@ /* { dg-do compile } */ /* { dg-options "-O2 -fstack-protector-strong -mcpu=power8" } */ +/* { dg-require-effective-target ppc_ieee128_ok } */ typedef union { __ieee128 a; int b; diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-supports.exp index d3b2798..0154457 100644 --- a/gcc/testsuite/lib/target-supports.exp +++ b/gcc/testsuite/lib/target-supports.exp @@ -2248,10 +2248,32 @@ proc check_ppc_float128_hw_available { } { } $options } }] } +# See if the __ieee128 keyword is understood. +proc check_effective_target_ppc_ieee128_ok { } { + return [check_cached_effective_target ppc_ieee128_ok { + # disable on Darwin, AIX. + if { [istarget powerpc-*-eabi] + || [istarget powerpc*-*-eabispe] + || [istarget *-*-darwin*] + || [istarget *-*-aix*] } { + expr 0 + } else { + set options "-mvsx" + check_runtime_nocache ppc_ieee128_ok { + int main() + { + __ieee128 a; + return 0; + } + } $options + } + }] +} + # Return 1 if the target supports executing VSX instructions, 0 # otherwise. Cache the result. proc check_vsx_hw_available { } { return [check_cached_effective_target vsx_hw_available {
Hi, This adds a target_supports entry to check that the _ieee128 keyword is understood by the target. Also adds that require-effective-target check to the existing pr92796 testcase. Sniff tested on Linux (power6,power9) and AIX. OK for master? Thanks -Will [testsuite] * lib/target-supports.exp (check_effective_target_ppc_ieee128_ok): New. * gcc.target/powerpc/pr92796.c: Add require-effective-target ppc64_ieee128_ok.