Commit Graph

81 Commits

Author SHA1 Message Date
mrg
3849e25709 import GCC 4.1 branch from 20061021.
includes fixes for these 109 GCC PR's:

 4520 13685 13983 17519 19505 20256 22209 22313 23454 24367 25211
25468 25519 25636 25636 26435 26504 26570 26719 26764 26881 26969
26983 26991 26991 26993 27075 27184 27226 27287 27287 27291 27334
27363 27428 27489 27490 27537 27558 27565 27566 27616 27639 27681
27697 27721 27724 27768 27793 27793 27795 27827 27878 27889 27893
28029 28075 28136 28148 28150 28162 28170 28187 28207 28207 28218
28221 28238 28243 28247 28257 28259 28267 28283 28286 28299 28386
28402 28403 28418 28473 28490 28493 28621 28634 28636 28649 28651
28677 28683 28726 28814 28825 28862 28900 28924 28946 28952 28960
28980 29006 29091 29119 29132 29154 29198 29230 29290 29323
2006-10-21 22:40:12 +00:00
mrg
622e071e10 import GCC 4.1 branch from 20061021.
includes fixes for these 109 GCC PR's:

 4520 13685 13983 17519 19505 20256 22209 22313 23454 24367 25211
25468 25519 25636 25636 26435 26504 26570 26719 26764 26881 26969
26983 26991 26991 26993 27075 27184 27226 27287 27287 27291 27334
27363 27428 27489 27490 27537 27558 27565 27566 27616 27639 27681
27697 27721 27724 27768 27793 27793 27795 27827 27878 27889 27893
28029 28075 28136 28148 28150 28162 28170 28187 28207 28207 28218
28221 28238 28243 28247 28257 28259 28267 28283 28286 28299 28386
28402 28403 28418 28473 28490 28493 28621 28634 28636 28649 28651
28677 28683 28726 28814 28825 28862 28900 28924 28946 28952 28960
28980 29006 29091 29119 29132 29154 29198 29230 29290 29323
2006-10-21 22:29:06 +00:00
mrg
2c32a05a20 import GCC 4.1 branch from 20061021.
includes fixes for these 109 GCC PR's:

 4520 13685 13983 17519 19505 20256 22209 22313 23454 24367 25211
25468 25519 25636 25636 26435 26504 26570 26719 26764 26881 26969
26983 26991 26991 26993 27075 27184 27226 27287 27287 27291 27334
27363 27428 27489 27490 27537 27558 27565 27566 27616 27639 27681
27697 27721 27724 27768 27793 27793 27795 27827 27878 27889 27893
28029 28075 28136 28148 28150 28162 28170 28187 28207 28207 28218
28221 28238 28243 28247 28257 28259 28267 28283 28286 28299 28386
28402 28403 28418 28473 28490 28493 28621 28634 28636 28649 28651
28677 28683 28726 28814 28825 28862 28900 28924 28946 28952 28960
28980 29006 29091 29119 29132 29154 29198 29230 29290 29323
2006-10-21 22:18:57 +00:00
mrg
1c6384e73d fix a merge-botch from our GCC3 fixes.
TARGET_5200 is TARGET_COLDFIRE now.
2006-10-21 05:31:41 +00:00
drochner
b7e3bcb0e5 gcc4/gdb6 work well with dwarf debug symbols (at least in the example
from PR toolchain/25094), so revert the local change which made .stabs
the default
2006-10-19 18:10:38 +00:00
apb
da8e22116c Use ${HOST_SH} instead of the host system's /bin/sh wherever possible:
* Use ${CONFIG_SHELL:-/bin/sh} instead of just /bin/sh when creating
  temporary shell scripts.
2006-09-29 20:11:07 +00:00
apb
cb575a16ce Use ${HOST_SH} instead of the host system's /bin/sh wherever possible:
* Add CONFIG_SHELL=$(SHELL) to the environment when invoking genmultilib.
  Due to substitutions made by associated configure scripts, the make
  variable $(SHELL) will actually refer to the environment variable
  ${CONFIG_SHELL}.
2006-09-29 20:10:01 +00:00
matt
b23fae2957 Fix problem with cross compiles on LP64 platforms. When splitting a DI
only compare against HOST_BITS_PER_WIDE_INT if TARGET_64BIT is true.
Otherwise, compare against 32 (as would happen on a IPL32 platform).
2006-09-23 19:07:19 +00:00
matt
4a7a5c53dd For thing that get a INTVAL result, make their type HOST_WIDE_INT. 2006-09-23 19:03:43 +00:00
matt
97da8b0449 Use HOST_WIDE_INT instead of long. 2006-09-23 19:00:03 +00:00
skrll
64001ed248 Our hppa userland space register convention isn't as flat as it needs to be
for MASK_NO_SPACE_REGS. The assumption is that sr4-sr7 == process space
id, but we have sr7 = kernel space id (0). This was inherited from
OpenBSD.

If sr4-sr7 == process space id, then

        ldw     %r1(,%r2),%r3
and
        ldw     %r2(,%r1),%r3

are equivlent for all values of %r1 and %r2.

If the right compat stuff is in place we can change the space register
convention and turn MASK_NO_SPACE_REGS back on.
2006-09-22 13:11:57 +00:00
cherry
c53d4b12a7 Fix libgcc4 build for ia64
Change #if to #ifdef
XXX: Feed upstream ?
2006-09-21 15:13:02 +00:00
scw
ddf78ea470 Check explicitly for sh5*-*-netbsd* when setting sh_multilibs. 2006-09-04 19:51:26 +00:00
mrg
56937a15b3 port the GCC3 gcc/config/ia64/netbsd.h. only change is to remove a call
to the no longer available/required NETBSD_OS_CPP_BUILTINS_LP64() macro.
2006-09-01 21:56:18 +00:00
matt
59b53ad312 More mips64 toolchain changes. With these changes, the mips64 toolchain
can build a mips64 kernel.
2006-08-26 20:01:29 +00:00
uwe
f482b86d5d gcc4 generated extremely unoptimal code for shifts of 64-bit ints on SuperH.
Pull over a fix from gcc 4.1 branch.

ok mrg@

Author: sayle
Date: Fri Jul 21 13:22:31 2006
New Revision: 115644

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115644
Log:

        PR middle-end/28283
        * expmed.c (expand_shift): Additionally check that the shift_cost
        is not MAX_COST and that INTVAL(op1) is less than MAX_BITS_PER_WORD
        before implementing a LSHIFT_EXPR as a sequence of additions.
        * config/sh/sh.c (shift_costs): Return MAX_COST to inform the
        middle-end that DImode shifts need to be synthesized by expand.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/sh/sh.c
    branches/gcc-4_1-branch/gcc/expmed.c
2006-07-24 23:36:03 +00:00
dan
33b0624dcd Change date in the nbX part of the version string to match the date
the nbX change was made.  The original gcc cvs/svn branch date is kept
separately.

After discussion with simonb@ who pointed me at rev 1.4 where he'd
corrected the same wrong assumption.
2006-07-11 07:32:07 +00:00
dan
9c66a2d708 pull in fix for gccpr 28016 from svn rev 114739:
Do not emit uninstantiated static data members.

this addresses C++ issues that broke linkage of programs using several
boost libraries, including monotone.

bump nb version accordingly

discussed with and committed by request of mrg@
2006-07-11 07:22:58 +00:00
ross
43f265ec70 Redefine RS6000_CPU_CPP_ENDIAN_BUILTINS() the same way powerpc does
to eliminate the _BIG_ENDIAN builtin which conflicts with our endian.h

XXX for now, define MASK_PROFILE_KERNEL here XXX
2006-07-10 22:58:58 +00:00
jnemeth
8021a8f195 correct NetBSD capitalisation 2006-07-10 22:51:51 +00:00
matt
8d5953af19 Define FRAME_POINTER_CFA_OFFSET since it is a constant 0 since VAX *always*
has a valid frame pointer.
2006-07-08 00:29:59 +00:00
mrg
e8f235719e also use the "config/netbsd.h" definition of NETBSD_LIB_SPEC for netbsd. 2006-07-07 00:11:06 +00:00
mrg
ad04044dda define the NetBSD start/end/link specs in terms of the ones from
"config/netbsd-elf.h", not some local definitions (that are wrong.)
2006-07-06 21:39:06 +00:00
mrg
5457f6307c do not use crtsavres 2006-07-06 21:38:14 +00:00
mrg
c3465e58a9 oops, forgot to bump the NetBSD GCC date with the last import. 2006-07-06 15:34:43 +00:00
mrg
9ffa9806b4 make this look slightly more like linux64.h. remove some unused stuff.
don't set a default rs6000_ieeequad.  now libgcc builds.
2006-07-06 10:47:37 +00:00
mrg
eb505c78cc make powerpc64-netbsd:
- not include rs6000/netbsd.h
- include biarch64.h at the start of tm_file, like powerpc64-linux does
- include the linux kernel profiling flag option (though profiling is
  currently busted in general)
2006-07-06 10:24:20 +00:00
mrg
31a7fdbc10 don't set a default rs6000_ieeequad on NetBSD just as Linux. 2006-07-06 10:21:03 +00:00
mrg
4ec585fc16 fix this for non-netbsd targets. 2006-07-06 10:13:52 +00:00
mrg
8b8880930c for armeb, TARGET_ENDIAN_DEFAULT now needs to be set to MASK_BIG_END. 2006-07-06 00:25:34 +00:00
mrg
9dc3b6767c don't need rs6000/netbsd.h twice. 2006-07-03 00:19:08 +00:00
matt
054488ff6f Use tramp.asm, not tramp.S 2006-07-02 02:18:08 +00:00
mrg
f7ccc44eb0 merge GCC 4.1-20060628. 2006-06-28 19:12:29 +00:00
mrg
3b31e3040a import GCC 4.1 branch from 20060628.
this includes new fixes for the these 40 GCC PRs:

21210 22541 25161 26223 26347 26559 27020 27082 27158 27177
27227 27421 27601 27648 27665 27666 27689 27722 27733 27781
27790 27793 27802 27806 27807 27819 27821 27830 27842 27858
27861 27884 27933 27951 27959 28045 28109 28110 28112 28151
2006-06-28 12:10:26 +00:00
mrg
e35e1713f9 allow sh-*-* as a valid sh configuration. now need_64bit_hwint=yes is
set for sh--netbsdelf, and sh3eb works.

XXX: annoyingly, this is needed in both gcc/config.gcc and
XXX: libcpp/configure*, and if just gcc/config.gcc is done, the
XXX: compiler will not work at all..
2006-06-27 02:59:09 +00:00
mrg
c82d87f94d - match sh*-netbsdelf* as well as sh-*netbsdelf*.
- test(1) does not have a == operator; replace it with =
2006-06-26 22:39:16 +00:00
mrg
62855b1d65 match 'shle-*' as being a sh platform, and try it's TLS code. with
this shle--netbsdelf configure's almost the same as sh--netbsdelf,
and the only unexpected difference remains in sh--netbsdelf files.
2006-06-26 19:56:17 +00:00
wiz
789f1d2877 Revert last chunk of previous: makes build work again.
Suggested by skrll@, ok moof.
2006-06-26 19:41:01 +00:00
dogcow
62f312bd4b revision 115002 from svn gcc; fixes PR/33816 (on my system, at least.) 2006-06-26 00:33:11 +00:00
mrg
0d6fef01af apply a patch from pa-linux.h. makes branch offsets work properly.
now hp7000 builds userland.  from skrll@.
2006-06-25 05:10:33 +00:00
mrg
0a294bff2f oops, powerpc64-netbsd should not include rs6000/netbsd.h. 2006-06-22 21:14:04 +00:00
mrg
1e5c7c79bc initial powerpc64-netbsd GCC configuration. this is only very minimally
tested and is not nearly ready for general use.

based on the 64 bit linux and 32 bit netbsd configurations.
2006-06-22 20:05:44 +00:00
mrg
6eca184761 bump NetBSD date 2006-06-03 07:40:16 +00:00
mrg
291b71d314 merge GCC 4.1.1-20060602. 2006-06-03 07:17:11 +00:00
mrg
6346714276 import GCC 4.1 branch from today. it includes these bugs fixed since
our last 4.1 branch import, plus a few other changes:
	c/27718 26242 c++/27451 c/26818 tree-optimization/26622
	target/27758 middle-end/27743 middle-end/27620
	tree-optimization/27549 tree-optimization/27283
	target/26600 c++/26757 driver/26885 tree-optimization/27603
	rtl-optimization/14261 rtl-optimization/22563 middle-end/26729
	rtl-optimization/27335 target/27421 middle-end/27384
	middle-end/27488 target/27158 bootstrap/26872 target/26545
	tree-optimization/27136 tree-optimization/27409 middle-end/27260
	tree-optimization/27151 target/26481 target/26765
	target/26481 tree-optimization/27285 optimization/25985
	tree-optimization/27364 c/25309 target/27387 target/27374
	middle-end/26565 target/26826 tree-optimization/27236
	middle-end/26869 tree-optimization/27218 rtl-optimization/26685
	tree-optimization/26865 target/26961 target/21283 c/26774 c/25875
	mudflap/26789
2006-06-03 05:31:50 +00:00
mrg
761e8f140b import GCC 4.1 branch from today. it includes these bugs fixed since
our last 4.1 branch import, plus a few other changes:
	c/27718 26242 c++/27451 c/26818 tree-optimization/26622
	target/27758 middle-end/27743 middle-end/27620
	tree-optimization/27549 tree-optimization/27283
	target/26600 c++/26757 driver/26885 tree-optimization/27603
	rtl-optimization/14261 rtl-optimization/22563 middle-end/26729
	rtl-optimization/27335 target/27421 middle-end/27384
	middle-end/27488 target/27158 bootstrap/26872 target/26545
	tree-optimization/27136 tree-optimization/27409 middle-end/27260
	tree-optimization/27151 target/26481 target/26765
	target/26481 tree-optimization/27285 optimization/25985
	tree-optimization/27364 c/25309 target/27387 target/27374
	middle-end/26565 target/26826 tree-optimization/27236
	middle-end/26869 tree-optimization/27218 rtl-optimization/26685
	tree-optimization/26865 target/26961 target/21283 c/26774 c/25875
	mudflap/26789
2006-06-03 05:21:35 +00:00
mrg
1083f0866d import GCC 4.1 branch from today. it includes these bugs fixed since
our last 4.1 branch import, plus a few other changes:
	c/27718 26242 c++/27451 c/26818 tree-optimization/26622
	target/27758 middle-end/27743 middle-end/27620
	tree-optimization/27549 tree-optimization/27283
	target/26600 c++/26757 driver/26885 tree-optimization/27603
	rtl-optimization/14261 rtl-optimization/22563 middle-end/26729
	rtl-optimization/27335 target/27421 middle-end/27384
	middle-end/27488 target/27158 bootstrap/26872 target/26545
	tree-optimization/27136 tree-optimization/27409 middle-end/27260
	tree-optimization/27151 target/26481 target/26765
	target/26481 tree-optimization/27285 optimization/25985
	tree-optimization/27364 c/25309 target/27387 target/27374
	middle-end/26565 target/26826 tree-optimization/27236
	middle-end/26869 tree-optimization/27218 rtl-optimization/26685
	tree-optimization/26865 target/26961 target/21283 c/26774 c/25875
	mudflap/26789
2006-06-03 05:11:23 +00:00
simonb
3eb98d14f3 Change date in the nbX part of the version string to match the date
the nbX change was made.  The original gcc cvs/svn branch date is kept
separately.

After discussion with phone@
2006-06-01 19:21:05 +00:00
simonb
f40854a95b Bump version to nb2 for gcc bug 26565 fix.
Revert date back to gcc branch date of 20060420.
2006-05-31 05:58:41 +00:00
simonb
e8c51d90a5 Pull up gcc svn revision 111934:
PR middle-end/26565
  * builtins.c (get_pointer_alignment): Handle component
  references for field alignment.

Fixes aligned store problems on MIPS described in:
    http://mail-index.NetBSD.org/tech-toolchain/2006/05/16/0002.html
    http://mail-index.NetBSD.org/tech-toolchain/2006/05/17/0001.html
2006-05-31 05:10:14 +00:00