[PATCH 1/3] rs6000: rs6000_dbx_register_number for fp/ap/mq
Segher Boessenkool
segher@kernel.crashing.org
Mon May 6 21:56:00 GMT 2019
The frame pointer and the argument pointer aren't real registers. MQ
was a register on old POWER. All three are still used as arguments to
rs6000_dbx_register_number during initialisation. If we handle them
explicitly we can do a gcc_unreachable to catch other unexpected
registers.
Tested on powerpc64-linux {-m32,-m64} and on powerpc64le-linux;
installing on trunk.
Segher
2019-05-06 Segher Boessenkool <segher@kernel.crashing.org>
* config/rs6000/rs6000.c (rs6000_dbx_register_number): Handle
FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and 64 (which was MQ).
--
gcc/config/rs6000/rs6000.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/gcc/config/rs6000/rs6000.c b/gcc/config/rs6000/rs6000.c
index c75fd86..6f46fdd 100644
--- a/gcc/config/rs6000/rs6000.c
+++ b/gcc/config/rs6000/rs6000.c
@@ -36305,7 +36305,15 @@ rs6000_dbx_register_number (unsigned int regno, unsigned int format)
if (regno == TEXASR_REGNO)
return 230;
- return regno;
+ /* These do not make much sense. */
+ if (regno == FRAME_POINTER_REGNUM)
+ return 111;
+ if (regno == ARG_POINTER_REGNUM)
+ return 67;
+ if (regno == 64)
+ return 100;
+
+ gcc_unreachable ();
#endif
}
@@ -36337,7 +36345,14 @@ rs6000_dbx_register_number (unsigned int regno, unsigned int format)
if (regno == TEXASR_REGNO)
return 116;
- return regno;
+ if (regno == FRAME_POINTER_REGNUM)
+ return 111;
+ if (regno == ARG_POINTER_REGNUM)
+ return 67;
+ if (regno == 64)
+ return 64;
+
+ gcc_unreachable ();
}
/* target hook eh_return_filter_mode */
--
1.8.3.1
More information about the Gcc-patches
mailing list