word) to be written back to the now initialized PLT entry thus
destroying a portion of the initialization that had occurred only a
short time ago. When execution continued, the zero word would be
- executed as an instruction an an illegal instruction trap was
+ executed as an instruction an illegal instruction trap was
generated instead. (0 is not a legal instruction.)
The fix for this problem was fairly straightforward. The function
program modified the code on us, so it is wrong to put back the
old value. */
if (val == 0 && memcmp (bp, old_contents, bplen) == 0)
- val = target_write_memory (addr, bp_tgt->shadow_contents, bplen);
+ val = target_write_raw_memory (addr, bp_tgt->shadow_contents, bplen);
do_cleanups (cleanup);
return val;
gdb_byte *buf;
buf = alloca (register_size (gdbarch, regnum));
- regcache_cooked_read (cache->regcache, regnum, buf);
+
+ if (regnum < gdbarch_num_regs (gdbarch))
+ regcache_raw_read (cache->regcache, regnum, buf);
+ else
+ gdbarch_pseudo_register_read (gdbarch, cache->regcache, regnum, buf);
+
return frame_unwind_got_bytes (this_frame, regnum, buf);
}
else if (regnum == SPU_PC_REGNUM)
store_unsigned_integer (buf, 4, byte_order, data->npc);
else
- return 0;
+ return REG_UNAVAILABLE;
- return 1;
+ return REG_VALID;
}
static int
static const struct frame_unwind ppu2spu_unwind = {
ARCH_FRAME,
+ default_frame_unwind_stop_reason,
ppu2spu_this_id,
ppu2spu_prev_register,
NULL,
/* Until November 2001, gcc did not comply with the 32 bit SysV
R4 ABI requirement that structures less than or equal to 8
bytes should be returned in registers. Instead GCC was using
- the the AIX/PowerOpen ABI - everything returned in memory
+ the AIX/PowerOpen ABI - everything returned in memory
(well ignoring vectors that is). When this was corrected, it
wasn't fixed for GNU/Linux native platform. Use the
PowerOpen struct convention. */