apply gcc original rev. 1.59:
> * config/sh/sh.c (prepare_scc_operands): Apply force_reg to
> sh_compare_op1 when the mode is DImode and sh_compare_op1 isn't
> const0_rtx.
the patch I posted to tech-toolchain@netbsd.org as a workaround for the
GOT bug whereby the callee-saved register %a5 can be trashed in certain
optimised -fPIC functions.
This may yet need to be fixed a different way, pending input from
Andreas Schwab on his change in revison 1.33 of this file in the
FSF source...
In the meantime, I've built and installed several NetBSD shared libraries
(including libc) with my change and found no problems.
of this file in the master FSF sources; I obviously disagree
with the change, since it was apparently made to appease one
particular OS, even though previous binutils releases had been
made with ENTRY=__start, and have brought it up on
binutils@sources.redhat.com).
the target "native toolchain" if BOOTSTRAP_NEW_TOOLCHAIN is set.
This is important if you don't have any userland at all, and you're
trying to make one from which you can run toolchain2netbsd.
aligned. Corresponds to part of rev 1.21 of this file in the
master FSF sources.
Thanks to Michael Rauch <mrauch@netbsd.org> for fishing out the patch.
is in their revision 1.186, from schwab@suse.de, and said:
Don't call reg_used_between_p if the insn from BL2 is after
the insn from BL.
This fixes a cc1 SEGV on sys/dev/scsipi/ch.c rev 1.48 under -O2 or -Os,
seen on m68k and vax.
* move the code that supplies the integer and FP registers into
separate functions, rather than duplicating the code for the
ptrace and core file cases.
* Use supply_register() rather than just copying directly into
the register array and calling registers_fetched(). This way,
only the registers actually supplied are marked as valid within
the debugger.
* Add support for SSE/SSE2 registers via the PT_{GET,SET}XMMREGS
ptrace(2) request.
(Blocked on the FSF assignment clerk for feeding this back to
the master GDB sources.)
the 4.17 NetBSD support from our tree. Original port by Chris Sekiya, with
minimal tweaks and cleanups by me (mainly: build bi-endian by default, add
NetBSD kcore support back).
for files named .cc or .C. _eh gets generated into a .c file so we need
explicit rules for it's targets (.o .po and .so) to compile it correctly.
Without this exceptions just plain don't work. Nothing ever gets caught.
and place it in i386nbsd-tdep.c (new file).
- Move solib.o from TDEPFILES to NATDEPFILES (solib.c requires target
headers, namely <link_elf.h>).
Fixes build of cross-gdb pointed at i386--netbsd target.
...And while we're at it, add a profiled libgcc too.
Use the "generate .c files and let <bsd.lib.mk> sort it out" method
for compiling these libraries. Only one real divergence (-fexceptions)
existed, but exceptions are turned on for C++ code by default in gcc
2.95.3, so this option was redundant anyway.
dist .y and .c files reversed.
1. Move the .y.c and other assorted implicit rule overrides out of Makefile.inc
and into local Makefile's. The system Makefile (bsd.sys.mk) sets up .l.c and
.y.c rules so unless these come after all inclusions they just get ignored.
2. Add @true as the command for any of the rule overrides. Otherwise make
still bails complaining about not knowing how to build the requisite .c or
.h file.
This obviously wasn't tested before as it couldn't have worked as-is.
From Rafal's commit for mipseb (which applies here too):
WARNING: Binutils 2.11.2 (maybe earlier) changed the MIPS ABI, so any
shared libs built by this toolchain WILL NOT WORK without either a whack
to BFD to fix that or a patch to ld_elf.so to work around it. I need to
chase the binutils folks on this issue still.
libraries that have their dependencies recorded via NEEDED/RPATH so
that the location of the dependent libraries doesn't have to be supplied.
This closes bin/14583 by myself.
OK'd by Todd Vierling <tv@netbsd.org>
The Postfix SMTP server maintains a record of SMTP conversations
for debugging purposes. Depending on local configuration details
this record is mailed to the postmaster whenever an SMTP session
terminates with errors.
During code maintenance, a stupid error was introduced into the
code due to which the SMTP session log could grow to an unreasonable
size. This stupid error made Postfix vulnerable to a memory
exhaustion attack.
* Rewrite src/tools Make logic to work like the rest of the tree wrt
"dependall" and "install". The old "make build" hack is gone.
* Remove the MKTOOLS logic. This was linked to the "make build" hack,
and was only needed because TOOLDIR originally had no writable default.
* Redo the GNU configure/make logic to make it fit reasonably in a
BSD make wrapper. Use new ${.ALLTARGETS} variable to scan for
targets in $(srcdir), and mark them with .MADE: to prevent rebuilding.
* Only build cross tools in src/tools; remove some messy logic in
src/usr.* and src/gnu/usr.* that would do target filename rewriting
(improves consistency and readability).
* Add the ability to build cross gdb at tool build time by setting
MKCROSSGDB (default no) to "yes" in mk.conf.
* Add src/tools/groff and set up paths to work with this cross groff.
left unimplemented, and postfix daemon could become an open relay for
IPv6 connectivity. it seems that we need to torture-test IPv6 patch outside
of the tree more.
From moley@wide.ad.jp
derived from the BFD-using mdsetimage(8).
TODO:
* Use loadfile(), rather than home-grown ELF parsing code (requires
some changes to loadfile()).
* Support a.out like the gwr's dbsym(8) (this comes for free when
we switch to loadfile()).
changes to configuration stuff to (a) recognize `mipseb', and (b) build a
BE-default GCC on mipseb. gprof and gdb still not done.
WARNING: Binutils 2.11.2 (maybe earlier) changed the MIPS ABI, so any
shared libs built by this toolchain WILL NOT WORK without either a whack
to BFD to fix that or a patch to ld_elf.so to work around it. I need to
chase the binutils folks on this issue still.
That said, the new toolchain seems to work quite well once the ABI change
is worked around/fixed -- I'm committing from a machine running a user-
land built with the new compiler.
a 26-bit target), split R15 into PC and CPSR. Similarly, when storing
registers, copy CPSR back into R15 if CPSR indicates a 26-bit mode.
This makes single-stepping work correctly on arm26.
Also, when fetching registers, set arm_apcs_32 based on the mode the target's
in. This removes the need to kludge it manually. The variable should really
be lost entirely, and arm-tdep.c should look at the mode bits in CPSR, but
the ARM Linux target is horribly broken in this area, and will either need
fixing or working around.
thorpej