X86: Disable cmov-memory patterns on subtargets without cmov.
Fixes PR15115.
For the i386 arch, this should enable cmov instructions only on
-march=pentiumpro and higher. Since our default CPU is i486, cmov
instructions will now be disabled by default.
MFC after: 1 week
Refactor the x86 CPU name logic in the driver and pass -march and -mcpu
flag information down from the Clang driver into the Gold linker plugin
for LTO. This allows specifying -march on the linker commandline and
should hopefully have it pass all the way through to the LTO optimizer.
Fixes PR14697.
Pull in r175919 from upstream clang trunk:
Driver: Pass down the -march setting down to -cc1as on x86 too.
The assembler historically didn't make use of any target features, but this has
changed when support for old CPUs that don't support long nops was added.
This should fix the long nops that still occurred in crt*.o, and
possibly other object files, if the system was compiled for a CPU that
does not support those, such as Geode.
Note that gcc on i386 also does not pass through any -march, -mcpu or
-mtune setting to gas, but this has not caused any trouble yet, because
gas defaults to i386.
Reported by: lev
MFC after: 1 week
This fixes the problem on amd64 miscompiling mpboot.s causing boot
issues... We are still using gas for a few files in the kernel...
Submitted by: kib
MFC after: 1 month
Thanks to Mike Belopuhov for the pointer to the OpenBSD patch, though
OpenBSD's gcc is very different that it only helped w/ where to modify,
not how... Thanks to jhb for some early reviews...
Reviewed by: imp, kib
MFC after: 1 month
MCParser: Reject .balign with non-pow2 alignments.
GNU as rejects them and there are configure scripts in the wild that
check if the assembler rejects ".align 3" to determine whether the
alignment is in bytes or powers of two.
MFC after: 3 days
From: Jouni Malinen <j@w1.fi>
Date: Fri, 15 Jul 2011 13:42:06 +0300
Subject: [PATCH] MD5: Fix clearing of temporary stack memory to use correct length
sizeof of the structure instead of the pointer was supposed to be used
here. Fix this to clear the full structure at the end of MD5Final().
Found by: clang ToT
Reviewed by: rpaulo
MFC after: 3 days
X86: Disable generation of rep;movsl when %esi is used as a base pointer.
This happens when there is both stack realignment and a dynamic alloca in the
function. If we overwrite %esi (rep;movsl uses fixed registers) we'll lose the
base pointer and the next register spill will write into oblivion.
Fixes PR15249 and unbreaks firefox on i386/freebsd. Mozilla uses dynamic allocas
and freebsd a 4 byte stack alignment.
MFC after: 1 week
_Unwind_ForcedUnwind in libgcc_s takes as one of it's parameters a stop
function to tell it when to stop unwinding. One of the stop function's
parameters is a _Unwind_Exception_Class. On most architectures this is an
int64_t, however on ARM EABI the gcc developers have made this a char array
with 8 items. While both of these take the same space they are passed into
the stop function differently, an int64_t is passed in in registers r2 and
r3, while the char[8] is passed in as a pointer to the first item in
register r2.
Because libthr expects the value to be an int64_t we would get incorrect
results when it passes a function that take an int64_t but libgcc passes in
a pointer to a char array including crashing.
The fix is to update libgcc_s to make it pass an int64_t to the stop
function and to libstdc++ as it expects _Unwind_Exception_Class to be an
array.
.note.ABI-tag section.
This helps on ARM EABI where the OS/ABI field is zero. It would be better
to use the NOTES program header however this would require a more invasive
change.
Dont use/link ARCMT, StaticAnalyzer and Rewriter to clang when the user
specifies not to. Dont build ASTMatchers with Rewriter disabled and
StaticAnalyzer when it's disabled.
Without all those three, the clang binary shrinks (x86_64) from ~36MB
to ~32MB (unstripped).
To disable these clang components, and get a smaller clang binary built
and installed, set WITHOUT_CLANG_FULL in src.conf(5). During the
initial stages of buildworld, those extra components are already
disabled automatically, to save some build time.
MFC after: 1 week
Since ARP and routing are separated, "proxy only" entries
don't have any meaning, thus we don't need additional field
in sockaddr to pass SIN_PROXY flag.
New kernel is binary compatible with old tools, since sizes
of sockaddr_inarp and sockaddr_in match, and sa_family are
filled with same value.
The structure declaration is left for compatibility with
third party software, but in tree code no longer use it.
Reviewed by: ru, andre, net@
an exception when you haven't caught one. This is largely a cosmetic fix, as
(unless you have a very unusual terminate handler installed) it will print a
nice error and then abort, rather than just aborting.
MFC after: 7 days
ARM EABI support is disabled by default and can be enabled by setting
WITH_ARM_EABI when building, however only the kernel-toolchain target will
work with this flag until the rest of the support is added.
- Don't treat pointers to members as pointers in catch blocks (they're usually
fat pointers).
- Correctly catch foreign exceptions in catchalls.
- Ensure that a happens-before relationship is established when setting
terminate handlers in one thread and calling them in another.
runtime on amd64, when it is compiled by clang. Some versions of clang
don't save and restore all callee registers, if a __builtin_eh_return()
intrinsic is used in a function. This is particularly bad on amd64.
Until the problem gets fixed by upstream, use an asm statement to force
clang to assume the registers in question are clobbered, when invoking
__builtin_eh_return(), so it will emit code to save and restore them.
This should fix the crashes reported on -current with some C++ programs,
particularly those that throw exceptions over multiple function
boundaries.
Reported by: stefanf
MFC after: 3 days
break mixed form dialogs in conjunction with the FreeBSD termcap, making
the bsdinstall partition editor Add dialog, among other things, completely
nonfunctional. This restores dialog 20110707.
There is a bug in gcc (GCC/35998) where dwarf reports
sizes of unsigned -1 (0xffffffff).
On NetBSD this generated a faulty CTF entry which then
caused a segfault in ctfmerge. The issue was worked
around in NetBSD's Dtrace but since the issue originated
in gcc, it seems reasonable to fix it here.
Upstream gcc has been slow to react to this issue and
the author that submitted the patch is not interested
in licensing the change to us, so I did an independent
workaround for the issue.
MFC after: 1 week
