matt
9d4b91ab87
Keep the unsigned, just add HOST_WIDE_INT.
2003-10-16 22:08:01 +00:00
matt
6bcdd1cd1e
Cast using HOST_WIDE_INT, not unsigned.
2003-10-16 22:02:25 +00:00
lukem
56528942aa
"anti-timestamp-skew" fixes for gengtype-yacc.c and gengtype-lex.c
2003-10-16 04:35:17 +00:00
skrll
fd905830f3
Don't pull in libgcc_pic.a for -shared - it was added as a workaround for
...
gcc 2.95.3. Removing this fixes EH for gcc 3.3 by using the .eh.section
support in libcrt{begin,end}
Closes PR 22673.
2003-10-13 13:19:41 +00:00
lukem
d47cc49cff
Add rules to prevent c-parse.c and objc-parse.c from being rebuilt if the
...
respective .y file has a newer timestamp (usually due to cvs checkout issues);
the code isn't .OBJDIR friendly, and it's unnecessary in our build envrionment.
2003-10-10 15:25:17 +00:00
lukem
83590dd4eb
Change method used to disable ``maintainer-mode'' style rules to one
...
that is more "[cvs] diff" friendly.
Method suggested by Chris Demetriou on source-changes@ on July 25.
2003-10-10 15:14:16 +00:00
matt
1a3bbf394a
Override TARGET_DEFAULTS to include MASK_STRICT_ALIGN (for IBM403).
...
It still includes (MASK_POWERPC | MASK_NEW_MNEMONICS) as well.
Change STACK_BOUNDARY to 128 to make Altivec'ed code easier to deal with.
2003-10-10 01:58:55 +00:00
matt
9fbbfac1b2
Pull in ld_vers / ld_date methods from gcc-current. (avoid GNU sed'isms).
...
This allows configure to correctly determine that gas supports .hidden.
2003-10-07 02:13:49 +00:00
matt
39caecb211
Add in-tree toolchain support for HPPA.
2003-10-06 05:10:13 +00:00
martin
b82411d0b6
Ooops, added previous note to the wrong ChangeLog
2003-10-03 09:06:14 +00:00
rearnsha
f539610cfa
When allocating a common object, ALIGN is in bits, but common expects
...
bytes. So divide by BITS_PER_UNIT. Fixes over-allocation of
common objects and thus reduces the BSS by approx 75%.
2003-09-29 12:40:35 +00:00
martin
d1faf5f973
Make gcc default to the medlow code model.
2003-09-27 11:08:41 +00:00
jwise
869320cb24
NetBSD includes many manual pages. It does not, however, include a gpl(7),
...
a gfdl(7), or (particularly) an fsf-funding(7).
So don't cross-reference them.
2003-09-26 17:59:10 +00:00
skrll
b9c8a77380
Backout previous - sparc* and i386 have switched to gdb 5.3.
2003-09-26 13:39:49 +00:00
matt
3edfe4aa79
NetBSD doesn't use crtsavres.o
2003-09-25 22:08:52 +00:00
matt
c9f03a720f
Redefine WCHAR/WINT defines since rs6000/svr4.h redefined them wrongly for
...
NetBSD.
2003-09-23 18:52:15 +00:00
matt
e426fdd061
Move #include to the beginning-of-line so -traditional-cpp can find it.
2003-09-23 14:46:05 +00:00
mrg
f4fc0bf370
make i386 & sparc* output DBX_DEBUG format ("-gstabs") by default for now.
...
XXX: back this out when these platforms switch to GDB 5.3.
2003-09-19 08:43:12 +00:00
martin
29f773f184
Add ChangeLog entry (from gcc-current) for recent sparc change.
2003-09-10 21:08:28 +00:00
martin
f1870485d5
Avoid outputing illegal assembler code for shifts by constants out of
...
bounds. Gcc bugzilla ticket 11965; update previous patch to the final
version accepted by the gcc people.
2003-09-09 21:26:20 +00:00
mrg
70ab3cd1bf
make "sh64--netbsdelf" use 32 bit by default. per advice from thorpej.
2003-09-04 08:04:18 +00:00
martin
c3181b5741
Shifts with >= 32 bit used to generate illegal code for 64bit targets,
...
fix by using for example "sllx" instead of "sll" in this cases.
Submitted to the gcc maintainers as attachment to [Bug c/11965] in their
bugzilla.
2003-08-23 10:19:02 +00:00
mrg
ee74cc4c6b
find ppc-asm.h via "rs6000/ppc-asm.h".
2003-08-22 02:46:17 +00:00
mrg
a827e79b49
patch from <sdegler@degler.net> to add /usr/include/g++/backward to the
...
default C++ include path so that one can find old headers without any
special help. (using them still generates warnings though.)
2003-08-22 02:41:15 +00:00
uwe
5b9717919b
In sh.h, ASM_OUTPUT_LABELREF - const'itfy lname.
...
Pull the fix for RETURN_ADDR_RTX from gcc-current:
sh-protos.h - 1.45; sh.h - 1.217 (partial); sh.c - 1.230 (partial).
Pertinent parts of the gcc changelog follows.
prologue / epilogue / warning patches:
2003-07-16 J"orn Rennecke <joern.rennecke@superh.com>
Con Bradley <con.bradley@superh.com>
* sh-protos.h (sh_get_pr_initial_val): Declare.
[...]
* sh.h [...]
(RETURN_ADDR_RTX): Use sh_get_pr_initial_val.
2003-08-11 00:45:34 +00:00
mrg
62340ef2af
document -isystem-cxx.
2003-08-10 01:57:50 +00:00
mrg
7fb34d4e7b
merge gcc3.3.1
2003-08-09 12:59:00 +00:00
mrg
adab0762e5
initial import of (final) GCC 3.3.1 sources
2003-08-09 11:11:09 +00:00
uwe
cb445c2c44
#define NO_PROFILE_COUNTERS as we don't need them. Otherwise both the
...
code to emit profile counters and the FUNCTION_PROFILER macro in this
file emit/define the same label. For gcc 2.95.3 it used to work
because FUNCTION_PROFILER used local numeric labels instead of using
LABELNO, so it caused no conflict.
This makes -pg code compilable.
2003-08-04 00:52:43 +00:00
mrg
774d0597a5
new (old) -isystem-cxx flag to set a C++ system include directory.
...
* c-opts.c: Document -isystem-cxx.
* cppinit.c (new_pending_directive): Add -isystem-cxx entry.
* gcc.h. (DEFAULT_WORD_SWITCH_TAKES_ARG): add -isystem.
XXX: possibly can use $CPLUS_INCLUDE_DIR.
2003-08-01 13:37:28 +00:00
mrg
952f507549
pull across a patch from .../toolchain/gcc/gcc.c that removes a whole
...
bunch of unrequired (and problem-causing) paths from various lists.
set (new) TOOLDIR_BASE_PREFIX and STANDARD_BINDIR_PREFIX defines in the
#ifdef NETBSD_NATIVE case as well as several other *_PREFIX defines.
2003-07-29 07:20:21 +00:00
mrg
127fd99434
avoid redefining a symbol differently
2003-07-27 15:13:39 +00:00
mrg
e70774a5af
oops, we need the generated-manpages target still.
2003-07-25 18:38:18 +00:00
mrg
5e7d44a059
comment some more rules that update read-only source trees
2003-07-25 18:33:19 +00:00
mrg
80b751809f
- comment the rules to rebuild .info files into $srcdir.
...
- don't try to link the (not yet built!) target libiberty.a for host programs(!)
2003-07-25 18:06:03 +00:00
mrg
20c6cdf6dc
make `cpp -gcc' the default on NetBSD.
2003-07-25 17:57:41 +00:00
mrg
731484776c
merge import conflicts.
2003-07-25 09:04:06 +00:00
mrg
1da79fd671
initial import of GCC 3.3 sources. (this is the latest GCC 3.3.1
...
prerelease snapshot, 3.3.1-20030720, we will update to 3.3.1 when
it becomes available.)
2003-07-23 02:40:42 +00:00
thorpej
1bdb20e71b
Remove the old egcs compiler.
2002-09-16 16:53:47 +00:00
thorpej
0b58bd1716
Remove the old egcs compiler.
2002-09-16 16:50:59 +00:00
thorpej
88edc7e11d
Remove the old egcs compiler.
2002-09-16 16:40:58 +00:00
thorpej
f6abb51010
Remove the old egcs compiler.
2002-09-16 16:39:52 +00:00
msaitoh
54d5b8ac24
apply gcc original rev. 1.84
...
> 2001-01-11 Alexandre Oliva <aoliva@redhat.com>
>
> * config/sh/sh.c (barrier_align): Recognize branch around far
> branch and redundant insn.
2002-04-05 18:06:29 +00:00
msaitoh
e029db8b1c
apply gcc original rev. 1.51
...
> 2000-03-06 Clinton Popetz <cpopetz@cygnus.com>
>
> * config/sh/sh.c: (barrier_align): Handle a delay slot that is
> filled with an insn from the jump target.
2002-04-05 17:58:28 +00:00
msaitoh
bf44c5577f
apply a part of gcc's original rev. 1.96
...
ADD MISSING "()"
> 2001-06-03 Alexandre Oliva <aoliva@redhat.com>
>
> * config/sh/sh.c (barrier_align): Add parentheses in initial
> credit computation.
2002-04-05 17:34:27 +00:00
msaitoh
ffef804eee
add missing "return 0;"
2002-04-05 17:26:00 +00:00
msaitoh
83dcb72379
apply gcc's original rev. 1.81
...
> 2000-12-30 Alexandre Oliva <aoliva@redhat.com>
>
> * config/sh/sh.c (split_branches): Don't dereference re-computed
> `beyond' before checking it's non-NULL.
2002-04-05 17:23:20 +00:00
msaitoh
f8de406a0c
apply a part of gcc's original rev. 1.67
...
> Mon Sep 18 20:26:50 2000 J"orn Rennecke <amylaar@redhat.co.uk>
>
> * sh.c (find_barrier): For Sh1, take alignment after jumps into account.
2002-04-05 17:10:24 +00:00
msaitoh
689473c860
apply gcc original rev. 1.28
...
> Fix pcrel too far problems due to upping CACHE_LOG on SH2:
> * sh.c (barrier_align): Don't return early for normal branch/barrier
> when optimizing for SH2.
2002-04-05 16:59:16 +00:00
msaitoh
e19e995528
apply gcc original rev. 1.144
...
> 2000-09-19 Bernd Schmidt <bernds@redhat.co.uk>
>
> * final.c (insn_current_reference_address): Use INSN_SHUID of seq
> rather than that of branch.
> (shorten_branches): Don't increment insn_current_address twice.
2002-04-05 16:21:38 +00:00