GCC build failed for native with your patch on 2003-06-12T22:32:09Z.
Andrew Pinski
pinskia@physics.uc.edu
Fri Jun 13 01:21:00 GMT 2003
This is most definitely Aldy's patch:
+2003-06-12 Aldy Hernandez <aldyh@redhat.com>
+
+ * config/rs6000/rs6000.c (function_arg): Always split vectors for
+ e500 if it's a stdarg function.
+ (function_arg_advance): Advance 2 registers for vectors in a
+ stdarg function.
+ (init_cumulative_args): Initialize stdarg.
+ (rs6000_spe_function_arg): New.
+
+ * config/rs6000/rs6000.h (rs6000_args): Add stdarg.
This hunk caused it:
*************** function_arg (cum, mode, type, named)
*** 3923,3928 ****
--- 3960,3973 ----
int n_words;
int gregno = cum->sysv_gregno;
+ if (TARGET_SPE_ABI && TARGET_SPE && SPE_VECTOR_MODE (mode)
+ && !cum->stdarg
+ && cum->sysv_gregno <= GP_ARG_MAX_REG)
+ {
+ cum->sysv_gregno++;
+ return;
+ }
+
/* Aggregates and IEEE quad get passed by reference. */
if ((type && AGGREGATE_TYPE_P (type))
|| mode == TFmode)
Thanks,
Andrew Pinski
On Thursday, Jun 12, 2003, at 19:29 US/Eastern, GCC regression checker
wrote:
> With your recent patch, GCC does not compile on:
> native
> Attached is build output for those targets.
>
> The build failures are new.
>
> For more information, see http://gcc.gnu.org/regtest/.
> <ChangeLog.diff><native-log.txt>--
> Geoffrey Keating <geoffk@apple.com>
> (via an automated GCC regression-testing script.)
More information about the Gcc-regression
mailing list