This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
fix alpha ld nop fill
- From: Richard Henderson <rth at redhat dot com>
- To: binutils at sources dot redhat dot com
- Date: Fri, 15 Feb 2002 15:43:27 -0800
- Subject: fix alpha ld nop fill
I'd previously missed the fact that the fill string is always
big-endian, and thus the existing definition was incorrect.
Also take advantage of Alan's new arbitrary fill length to
emit unop+nop pairs, which runs slightly better on EV4s. Not
that there are too many of those still running, but hey, it
doesn't hurt.
I applied a similar patch to the 2.12 branch, since even with
the 4 byte definition it was wrong there too.
r~
* emulparams/elf64alpha.sh (NOP): Adjust for big-endian
definition. Emit a unop+nop pair.
Index: emulparams/elf64alpha.sh
===================================================================
RCS file: /cvs/src/src/ld/emulparams/elf64alpha.sh,v
retrieving revision 1.9
diff -c -p -d -r1.9 elf64alpha.sh
*** elf64alpha.sh 2002/02/12 14:50:07 1.9
--- elf64alpha.sh 2002/02/15 23:38:07
*************** ARCH=alpha
*** 11,17 ****
MACHINE=
GENERATE_SHLIB_SCRIPT=yes
DATA_PLT=
! NOP=0x2ffe0000 # unop
OTHER_READONLY_SECTIONS="
.reginfo ${RELOCATING-0} : { *(.reginfo) }"
--- 11,19 ----
MACHINE=
GENERATE_SHLIB_SCRIPT=yes
DATA_PLT=
! # Note that the number is always big-endian, thus we have to
! # reverse the digit string.
! NOP=0x0000fe2f1f04ff47 # unop; nop
OTHER_READONLY_SECTIONS="
.reginfo ${RELOCATING-0} : { *(.reginfo) }"