Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
213,275 Commits 606 Branches 0 Tags 3.1 GiB
ab20ab348a
Commit Graph

480 Commits

Author SHA1 Message Date
martin
6d25cb1840 Make can_trust_pointer_alignment() always return false on STRICT_ALIGNMENT 
archs - gcc 4.5 is not able to properly track alignment and backporting
the fix from 4.6/4.7 is not feasible (according to upstream).
See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46483 for details.
Fixes PR toolchain/46865.
 2012-09-02 20:58:44 +00:00
msaitoh
568a3a3a5a Don't add unwind.h stuff into BUILDSYMLINKS when MKGCC=="no". 
Fixes "Wrong number of words (1) in .for substitution list with 2 vars" problem
while mknative-ing. Approved by mrg.
 2012-08-29 03:12:08 +00:00
matt
7d57bc60df Make this to produce working shared libraries again. 
Resultant distribution was booted into single user using shared libraries.
 2012-08-27 16:50:14 +00:00
matt
7c4c75c298 Add support for setting LIB_PATH for MKCOMPAT support. 2012-08-13 02:52:34 +00:00
mrg
99a54d3479 pay attention to -pie: link crt{begin,end}S.o. 
this fixes -fpie on sparc and sparc64, and probably others.
 2012-08-12 00:01:50 +00:00
joerg
c14f2d4ef8 Deal with optional HAVE_GCC. 2012-08-10 16:05:26 +00:00
matt
a4ae934e29 Remove code that should be have excised a long time ago. 2012-08-07 01:09:13 +00:00
matt
cb8c048ec5 on earm, don't include EH code. 2012-08-06 02:34:28 +00:00
matt
e19c6b7318 Add LIB2_EHASM to SRCS 2012-08-06 01:38:29 +00:00
matt
267745d7dd More regened configury (mostly exception related). 2012-08-06 00:49:52 +00:00
matt
c73e5095cc a few more fixes (make unwinding correct for EABI) 
improve link spec to with fix-v4bx and emulations
Add bpabi-netbsd.c to resolve problem with bpabi.c conflicting with bpabi.S
 2012-08-06 00:24:13 +00:00
matt
5598cdb53e Regen again. 2012-08-06 00:21:37 +00:00
matt
93f7a012df Deal with LIB2_EHASM 2012-08-06 00:21:09 +00:00
matt
638a65b2c0 deal with LIB2_EH has ASM (.S) files. 2012-08-06 00:19:03 +00:00
matt
b626930d34 Add a few missing for earm's gcc configury 2012-08-05 17:21:47 +00:00
matt
015ce606f7 Add gdb configury for earm 2012-08-05 17:09:16 +00:00
matt
714ce8eeef gcc4.5.3 configury for earm 2012-08-05 17:03:31 +00:00
matt
cd8b5f8c57 Allow arm*-netbsd*-eabi* to be recognized. 2012-08-05 16:51:46 +00:00
matt
88a901ba99 binutils mknative run for earm 2012-08-05 16:34:05 +00:00
matt
602e0146d8 Add arm{,eb}--netbsdelf-eabi support. (still evolving) 2012-08-05 04:10:18 +00:00
matt
ad73c53594 Add arm{,eb}--netbsdelf-eabi support. 2012-08-05 04:03:03 +00:00
joerg
e7a9d9d117 Remove a bunch of broken inline qualifiers that break the -O0 build. 2012-07-20 21:18:00 +00:00
christos
edd104c76c PR/46704: Antti Kantee: Recognize tilera in config.guess 2012-07-19 19:34:26 +00:00
matt
b7c801cac8 Two fixes. Better PLT pruning (don't emit PLTs for local functions). 
If emitting a reloc against a readonly section in a shared image,
cause TEXTREL to be set.
 2012-07-17 21:37:21 +00:00
matt
ec769e4972 Warn about using a symbol as a displacement in PIC mode. 2012-07-17 20:29:44 +00:00
abs
86aebb49c7 Switch -current vax back to gcc 4.1 to match netbsd-6 - until someone has 
time to look at the gcc 4.5 codegen issues we should at least have a usable
-current for vax.
 2012-06-22 20:32:34 +00:00
martin
4d5a9c43e0 Apply upstream revision 175141: setjmp vs. unchanged register values; 
should fix PR toolchain/45837.
 2012-06-17 13:43:30 +00:00
christos
15053c8b40 add specs to automagically add the mudflap libraries when -fmudflap{,th} 2012-04-27 00:57:44 +00:00
christos
cbd0c35a2d remove code that assumed that lwpid shares pid space and can be used interchangeably. 2012-04-24 19:14:06 +00:00
christos
7006e67fd8 Bring in fix from: 
http://gcc.gnu.org/bugzilla/show_bug.cgi?format=multiple&id=43804
Allows libmudflap to compile on m68k:

/usr/src/external/gpl3/gcc/dist/libmudflap/mf-hooks1.c: In function
'__mf_wrap_alloca_indirect':
/usr/src/external/gpl3/gcc/dist/libmudflap/mf-hooks1.c:492:1: error:
insn does not satisfy its constraints:
(insn 360 142 68 13
/usr/src/external/gpl3/gcc/dist/libmudflap/mf-impl.h:257 (set (reg/f:SI
6 %d6 [159])
        (const:SI (unspec:SI [
                    (symbol_ref:SI ("__mf_state_1") [flags 0x50]
<var_decl 0x7f7ff7304d20 __mf_state_1>)
                     (const_int 1 [0x1])
                 ] 6))) 36 {*movsi_m68k}
(nil))/usr/src/external/gpl3/gcc/dist/libmudflap/mf-hooks1.c:492:1:
internal compiler error: in reload_cse_simplify_operands, at postreload.c:396
 2012-04-22 15:15:46 +00:00
christos
3fd38f84eb If we are asked to step on ptid(-1,0,0), then step on our inferior_thread 
instead, which is the main thread instead of the currently executing thread
which is not what we want, since we've been running and nother thread might
have just started.
 2012-04-21 00:16:35 +00:00
christos
49a655adee Make declarations consistent, and change unsigned int to unsigned long so 
that this works on LP64 machines. Why is the __mf_uintptr_t needed anyway?
 2012-04-19 15:52:39 +00:00
christos
9175dee1fe bad name 2012-04-18 22:42:21 +00:00
christos
6e275ac292 add the mudflap include and libraries 2012-04-18 19:22:03 +00:00
christos
6b798f149f move include to the gcc-4.5 subdir. 2012-04-18 19:21:25 +00:00
christos
f89a1d6f2b NetBSD fixes: 
1. thread_self() returns a pointer, not an unsigned integer
2. Add NetBSD to Apple and FreeBSD defines
3. Add _NETBSD_SOURCE where needed
4. Add an extra define BEGIN_PROTECTV for void functions to avoid return
   free(); where free is void.
5. Avoid weak symbol hacks to determine if we are threaded or not. We
   have a threaded copy of the library, why bother?
6. change __attribute -> __attribute__ since the former is not covered by
   our cdefs.h
 2012-04-18 02:55:05 +00:00
christos
d0800ba09f add preliminary build glue for mudflap; only the shared libraries work, 
the static ones will need us to add __real_foo aliases for a bunch of
functions in libc.
 2012-04-18 02:52:07 +00:00
christos
16918ffb5d - In the thread activation code, mutate the main thread (lid=0) to lid=1 like 
the kernel does.
- Remove - 1 hack
- Remove more ifdef notdef code
 2012-04-11 21:41:59 +00:00
christos
4ae881aba3 On NetBSD the data part of the PT_STEP request is the lid to send the signal, 
not the signal to send! Make it so.
 2012-04-11 21:39:01 +00:00
christos
defdde4f72 remove more unused SA userland thread stuff. 2012-04-11 20:21:35 +00:00
wiz
be5cf4dc86 clang complains: 
C++ default arguments are illegal in the function defininion.
christos@ ok
 2012-04-09 00:24:57 +00:00
he
692082ea92 Disable HAVE_TLS for now for gcc's libgomp, since it otherwise 
causes gcc indigestion (internal compiler error) for all m68k ports.
mrg@ says this will do for now, and committing to this generated-
by-mknative file is ok.
 2012-04-06 20:10:28 +00:00
mrg
829e755fe2 rerun mknative-gcc: 
- turn off <fenv.h> on platforms that don't have it
 2012-04-05 21:09:20 +00:00
christos
49758559c7 - remove unused thread (non-lwp) code. 
- make the get method return the value, and only add 1 in the put method.
 2012-04-04 13:33:19 +00:00
joerg
b757af438b Disable new -Wstring-plus-int warning where needed for now. 
Adjust various places that add GCC-only options to check for the active
compiler first.
 2012-04-04 10:59:44 +00:00
christos
92a9368199 only +1 for the 0'th thread (if we did not find any other) 2012-03-22 16:06:01 +00:00
christos
2f9576aacc complete the mapping of the lwp -> (lwp - 1), by adding 1 where appropriate. 2012-03-22 15:26:32 +00:00
christos
24da109f75 Use tid - 1 for the thread id, so that thread 1 matches the main thread 
before the threaded code gets loaded. fixes spurious breakpoints. Is
there a better way?
 2012-03-21 23:20:58 +00:00
christos
9b33eb8aa1 Undo previous, it was wrong: 
The problem has to do with the threaded initialization.
After we take the first breakpoint and we enter single step mode,
we set trap_expected = 1 in the thread_info structure for main <pid,0,0>.
After the threads initialize, the main thread becomes <pid,1,0>, and so
we get a new thread_info struct with trap_expected = 0, and so we break.
 2012-03-21 21:25:38 +00:00
christos
436a58b1ab Disable OS supplied single step functionality, because the code here is 
wrong.
 2012-03-21 18:01:34 +00:00