Commit Graph

3796 Commits

Author SHA1 Message Date
thorpej 1268b2dbc5 Add an x86_initialize_trampoline() function, similar to GCC 3.3,
change INITIALIZE_TRAMPOLINE() to use it, and apply the following
patch to it (which I have already submitted to GCC 3.3):

	* config/i386/i386.c (x86_initialize_trampoline): Emit a call
	to __enable_execute_stack with the address of the trampoline
	if TRANSFER_FROM_TRAMPOLINE is defined.

Define a TRANSFER_FROM_TRAMPOLINE suitable for NetBSD targets, to enable
stack execution on target machines which can separate exec permissions for
a region:

	* config/netbsd.h (NETBSD_ENABLE_EXECUTE_STACK): Define.
	* config/alpha/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Define
	as NETBSD_ENABLE_EXECUTE_STACK.
	* config/i386/netbsd-elf.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
	* config/i386/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
	* config/sparc/netbsd-elf-common.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
	* config/sparc/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Ditto.

(A similar patch will be submitted to GCC 3.3.)
2002-10-18 16:23:47 +00:00
thorpej 00b192ce18 Bring in the following fix from gcc-current:
2002-10-17  Jason Thorpe  <thorpej@wasabisystems.com>

        * config/alpha/alpha.c (alpha_initialize_trampoline): Use
        tramp, not addr, to pass the trampoline address to
        __enable_execute_stack.

Otherwise, incorrect behavior would result if the trampoline happend
to stradle a page boundary.
2002-10-17 21:01:03 +00:00
itojun f1215ef8af add missnig -I to gnu/dist/mrouted, which is needed by gcc 3.2
(somehow 2.95 automagically includes it into -I path)
2002-10-17 12:50:00 +00:00
christos 706c1ce906 Flip the pax switch 2002-10-16 15:23:59 +00:00
enami 9fdb1e7ba1 Restore other symlink attributes. 2002-10-13 22:58:55 +00:00
martin 14cde9f60d Add a few dependencies to make sure libtar/fnmatch.h is generated before
anything includes <fnmatch.h>. Using our native /usr/include/fnmatch.h is
no good idea, they are incompatible.

XXX - why is tar using it's own implementation?

I'm not sure this is the right way to handle the problem, but it made
a "make cleandir && make" here possible again.
2002-10-13 18:41:48 +00:00
agc 84c3d85401 If we're going to rely on a pre-generated getdate.c, we might as well rely
on one generated with the correct functionality inside it.
2002-10-11 08:58:44 +00:00
drochner 3dd7978358 add a dummy pattern rule to avoid yacc invocation
(which fails with r/o src tree)
2002-10-10 13:03:19 +00:00
itojun 4384c54c01 switch to GNU tar 1.13.25 with security fixes. 2002-10-10 08:22:22 +00:00
itojun e076df538c seemingly more correct ".." patch by Solar Designer <solar@openwall.com>,
from bugtraq posting
2002-10-09 19:58:35 +00:00
itojun 9ad5cf147a sort 2002-10-09 19:54:05 +00:00
itojun 05b137baea allow --unbzip2 for backward compat
-y is not allowed as -y is reserved by GNU tar camp, it seems.
2002-10-09 19:53:25 +00:00
itojun 1130454b1c bring in --fast-read code from freebsd. 2002-10-09 19:28:14 +00:00
itojun 34c781b6f4 sync with latest --help output. 2002-10-09 17:01:34 +00:00
itojun caaeaa46d8 patch from redhat rpm - check st_mode earlier 2002-10-09 16:40:21 +00:00
itojun 99ba27c717 vide --fast-read (no-op) 2002-10-09 16:23:27 +00:00
itojun b8621000a9 correct LDADD/DPADD 2002-10-09 16:17:45 +00:00
itojun b33ebc5522 no file in .CURDIR 2002-10-09 16:14:55 +00:00
itojun d9bffe1f20 split internal library build into subdir. 2002-10-09 16:13:07 +00:00
itojun 6dee53867d --norecurse and --unlink for backward compat 2002-10-09 12:01:29 +00:00
itojun c873f2dfba dot-dot security fix, from redhat tar-1.13.25-1.6.src.rpm 2002-10-09 12:01:11 +00:00
itojun d67562c763 new GNU tar based on gnu/dist/tar. (to replace gnu/usr.bin/tar in the
near future - hopefully)
2002-10-09 11:55:02 +00:00
itojun fab68d14ac GNU tar 1.13.25 from alpha.gnu.org 2002-10-09 10:09:12 +00:00
itojun 29ef0d0d98 more strict checking on command invocation.
http://www.sendmail.org/smrsh.adv.txt, NetBSD PR 18516
2002-10-03 03:31:46 +00:00
wiz 12d694f3c4 Fix lots of typos. Some spotted by Adrian Mrva. 2002-10-02 15:39:09 +00:00
wiz 0cddcd9152 interpreted instead of intepreted. By Adrian Mrva. 2002-10-02 11:12:41 +00:00
wiz 659da00975 hve -> have. 2002-10-02 10:00:53 +00:00
thorpej cde0a5d6aa __mcount uses jsb linkage, so there is no need to skip the (non-existent)
register save mask.  Profiling now works.

Also, use asm_fprintf() in FUNCTION_PROFILER.
2002-10-02 00:50:24 +00:00
itojun c3cbda0efb move mbone tools out of basesrc, as they use part of mrouted 2002-10-01 03:50:07 +00:00
itojun d0a9c54d7d latest license terms, via openbsd 2002-10-01 03:41:13 +00:00
itojun 9bbfbbb676 move mrouted out of basesrc, as it carries non-BSD license.
agreed w/ core.
2002-10-01 03:30:50 +00:00
grant 2f722d4d84 Include DOMAIN(generic) so that various standard configuration items are
included, notably enabling the use of /etc/mail/local-host-names.
2002-09-30 01:27:05 +00:00
thorpej e870e0c546 Pick up the following change from binutils-current:
2002-09-28  Jason Thorpe  <thorpej@wasabisystems.com>

        * elf32-vax.c (elf_vax_size_dynamic_section): Don't strip
        .got sections.
2002-09-28 23:19:01 +00:00
wiz 398c2e6b26 Change argument handling to be like in groff-1.18, thus avoiding
a possible buffer overflow.
2002-09-27 15:05:11 +00:00
wiz dcc8010de1 ANSIfy, de-__P, complete sync of usage with man page. 2002-09-27 07:28:18 +00:00
wiz 9e9685df35 Sync usage with man page. 2002-09-27 07:25:17 +00:00
wiz fefd5e68e0 Bump date for -x addition. Sort options in usage. 2002-09-27 07:25:07 +00:00
mycroft b06a300a8a Fix a bug WRT -vomit-frame-pointer -- use_return_insn() must return false if
we saved the GOT pointer register.
Note: This is already fixed in the GCC 3 sources, but GCC 3 uses a different
condition to determine whether to save and restore the register, so the patch
cannot be the same.
2002-09-26 14:21:39 +00:00
dbj 2cbb8bc3bf add the -x option to extract the filesystem image from the kernel 2002-09-26 10:40:59 +00:00
mycroft 2468728dc8 Add missing parens that caused PLT slots >=32768 to jump to the wrong place. 2002-09-25 20:05:14 +00:00
wiz 3a300f71f9 Re-add, it is still used, but one only sees it in a full build, not a tools
build.
2002-09-22 13:06:44 +00:00
wiz e751d90149 Now that the old toolchain is gone, rename gdb.new to gdb, as requested
in a comment in gdb.new's Makefile. Remove that comment.
2002-09-22 12:58:23 +00:00
wiz ed5d7e73eb Remove some survivors of the old-toolchain removal. 2002-09-22 10:31:16 +00:00
wiz 625e362cc2 Remove completely outdated maintenance history, and add proper NetBSD RCS tag. 2002-09-22 09:47:56 +00:00
thorpej 914a9c522b -D__STRONGARM__ for -mcpu=strongarm* 2002-09-21 00:26:47 +00:00
thorpej 0604df808f Do -D__XSCALE__ -D__ARM_ARCH_5TE__ for -march=xscale, as well, as
GCC 3.x does.
2002-09-21 00:19:30 +00:00
rearnsha efc53cb5f7 Pull up the following patch from FSF GCC (fixes aborts when compiling
large functions for ARMv4):
Fri Nov 12 13:31:54 1999  Bernd Schmidt  <bernds@cygnus.co.uk>
* config/arm/arm.h (LEGITIMIZE_RELOAD_ADDRESS): Treat QImode
addresses the same way GO_IF_LEGITIMATE_INDEX does.
2002-09-19 21:52:29 +00:00
lukem 0ac69ffac4 - use NETBSDSRCDIR as appropriate
- makefile lint (moving NOxx to top of file,
  include <bsd.own.mk> as necessary, ...)
2002-09-19 03:09:31 +00:00
lukem 1addf6d6b3 makefile delint and reorg 2002-09-19 01:58:06 +00:00
thorpej 1dd1ce77c1 Remove more old toolchain bits. 2002-09-17 23:24:52 +00:00