The basic problem is that some mainstream programs cannot deal with the way
clang optimizes tail calls, as in this example:
int foo(void);
int bar(void) {
return foo();
}
where the call is transformed to:
calll .L0$pb
.L0$pb:
popl %eax
.Ltmp0:
addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp0-.L0$pb), %eax
movl foo@GOT(%eax), %eax
popl %ebp
jmpl *%eax # TAILCALL
However, the GOT references must all be resolved at dlopen() time, and so this
approach cannot be used with lazy dynamic linking (e.g. using RTLD_LAZY), which
usually populates the PLT with stubs that perform the actual resolving.
This patch changes X86TargetLowering::LowerCall() to skip tail call
optimization, if the called function is a global or external symbol.
This fixes problems with loading X.org driver modules, which could occur
when X.org was compiled on i386 with tailcall optimization on, for which
ports r312583 was committed as a workaround. After this change, the
workaround can be removed.
MFC after: 3 days
comparing a directory to an mtree file more compatible with fmtree when
FreeBSD 9 compatiblity mode is on. This output is clearly intended for
humans not computers, but some tools such as mergemaster's -U option rely
on it.
MFC after: 3 days
Bring The following revisions from the gcc43 branch[1]:
118360, 118361, 118363, 118576, 119820,
123906, 125246, and 125721.
They all have in common that the were merged long ago
into Apple's gcc and should help improve the general
quality of the compiler and make it easier to bring
new features from Apple's gcc42.
For details please review the additions to the files:
gcc/ChangeLog.gcc43
gcc/cp/ChangeLog.gcc43 (new, adds previous revisions)
Reference:
[1] http://gcc.gnu.org/viewcvs/gcc/trunk/?pathrev=126700
Obtained from: gcc pre4.3 (GPLv2) branch
MFC after: 3 weeks
Otherwise, a formatted string with a strlen equal to the remaining
buffer space would have the last character omitted (because vsnprintf
always null-terminates), and later the assert in backtrace_symbols_fmt
would fail.
MFC after: 3 days
Sponsored by: DARPA, AFRL
CaptureTracking: Plug a loophole in the "too many uses" heuristic.
The heuristic was added to avoid spending too much compile time in a
specially crafted test case (PR17461, PR16474) with many uses on a
select or bitcast instruction can still trigger the slow case. Add a
check for that case.
This only affects compile time, don't have a good way to test it.
This fixes the excessive compile time spent on a specific file of the
graphics/rawtherapee port.
Reported by: mandree
MFC after: 3 days
This is because the atf vendor branch now includes a verbatim copy of
the distfile sources. As a result, the list of files to-be-removed from
the contrib/ directory is now more aggressive (and different) and the
upgrade notes now only describe stuff that is specific to the atf import
and is not documented in the Subversion Primer.
Approved by: rpaulo (mentor)
* lex.c (_cpp_clean_line): Add uses of __builtin_expect. Don't
look backward at the end of the line unless we saw a backslash.
* internal.h (struct cpp_reader): Add new fields:
nonexistent_file_hash and nonexistent_file_ob.
* files.c: Include "obstack.h".
(find_file_in_dir): Before trying to open the file, look up the
path name in the hash table of nonexistent files. After failing
to open the file, add the path name to the hash table.
(_cpp_find_file): Cache the results of looking up the file name
starting with the quote and bracket chain heads, if we can.
(nonexistent_file_hash_eq): New static function.
(_cpp_init_files): Initialize pfile->nonexistent_file_hash and
pfile->nonexistent_file_ob.
(_cpp_cleanup_files): Free pfile->nonexistent_file_hash and
pfile->nonexistent_file_ob.
Obtained from: gcc 4.3 (rev. 120263, 124929 ; GPLv2)
MFC after: 3 weeks
Fix missed conversion from / to >> (GCC PR32521)
Note that vrp-tree is currently disabled by default in
FreeBSD's gcc due many bugs.
While here fix a revision number in the 4.3 Changelog.
Obtained from: gcc 4.3 (rev. 122831 - partial; GPLv2)
MFC after: 3 weeks
ludes minor changes relative to upstream, for compatibility with
FreeBSD's in-tree LLVM 3.3:
- Reverted LLDB r191806, restoring use of previous API.
- Reverted part of LLDB r189317, restoring previous enum names.
- Work around missing LLVM r192504, using previous registerEHFrames API
(limited functionality).
- Removed PlatformWindows header include and init/terminate calls.
Sponsored by: DARPA, AFRL
MFV of 257651, tzdata2013h
tzdata2013f
- Jordan goes to winter time on the last Friday in October.
- Tocantins in Brazil will not go into summer time in October.
- Indonesian time zones renames.
- Lots of cleanups in with regarding to links and historical data.
tzdata2013h
- Libya didn't go back to DST.
- Fix Morocco 2038 issue.
- Brazil/Acre and Western Amazonas are chaning timezones.
this call with the same #if defined (PREFER_STDARG) directive as in display.c
Using -E to compile display.c/search.c shows that this is the code chosen by
the build when we create libreadline
yp_get_default_domain is defined in workaround.c but is not declared
in any header file. Tie the declaration to the same #define conditional
used when the function is called, NETGROUP
1. define the CODE * as const
2. restructure function to eliminate warning about exiting with no return.
severity_map() never returns when it can't find an appropriate sysylog
facility, and it longjmp()'s away into error code handling. Keep this
behavior by stashing the facility value found during our search and
checking for -1 if found.
radiotap.h in upstream wpa originally came from the FreeBSD CVS
repository. The copy in vendor/ matched the upstream one; with this
change the contrib copy now matches too.
Discussed with: jhb@
1. missing explicit includes for string.h, stdio.h, etc
2. missing explicit declaration for some common functions
I have been unable to contact the upstream maintainer for this patch,
http://www.inner.net/opie appears to be the source of truth but it
unreachable
in net, to avoid compatibility breakage for no sake.
The future plan is to split most of non-kernel parts of
pfvar.h into pf.h, and then make pfvar.h a kernel only
include breaking compatibility.
Discussed with: bz
- Provide pf_altq.h that has only stuff needed for ALTQ.
- Start pf.h, that would have all constant values and
eventually non-kernel structures.
- Build ALTQ w/o pfvar.h, include if_var.h, that before
came via pollution.
- Build tcpdump w/o pfvar.h.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
mode as c++ instead of defaulting to c for the binary names CC and clang-CC.
This fixes builds that use cmake, which automatically sets CXX to
/usr/bin/CC by default.
PR: bin/182442
Reviewed by: dwhite, wca
MFC after: 2 days
of string format in a call so syslog
/usr/src/gnu/lib/libssp/../../../contrib/gcclibs/libssp/ssp.c:137:23:
warning: format string is not a string literal (potentially insecure)
[-Wformat-security]
syslog (LOG_CRIT, msg1);
^~~~
Reviewed by: dim@
64-bit debug data is only necessary for objects with greater than 4GB of
debug data, and is not used on other 64-bit FreeBSD targets.
Sponsored by: DARPA, AFRL
Fix the VCVT instruction. It must round towards zero when converting from
a floating-point to an integer value. This was not the case causing issues
when printing certain values.
There is a VCVTR instruction that will round depending on the current
rounding mode. We don't yet support this instruction, or setting the
rounding mode.
The header_length field is the number of bytes following the field to
the first byte of the line number program. The hard-coded constants
previously here (4 + 2 + 4) were correct only for 32-bit DWARF.
Sponsored by: DARPA, AFRL
re-removes the printing of size for non-regular files, supports
type-less root (.) entries as seen in a couple ports, and corrects a bug
in -N that caused expanded Subversion $FreeBSD$ strings in comments to
be processed as /etc/group entries.
Includes one trivial portability fix (including <stdint.h> in three
files) that has been submitted upstream but not yet committed.
MFC after: 3 days
runtime by the dynamic linker, check for their equality in libcxxrt by
not only comparing the typeinfo's name pointers, but also comparing the
full names, if necessary. (This is similar to what GNU libstdc++ does
in its default configuration.) The 'deep' check can be turned off again
by defining LIBCXXRT_MERGED_TYPEINFO, and recompiling libcxxrt.
Reviewed by: theraven
MFC after: 3 days
X86: Don't fold spills into SSE operations if the stack is unaligned.
Regalloc can emit unaligned spills nowadays, but we can't fold the
spills into SSE ops if we can't guarantee alignment. PR12250.
This fixes unaligned SSE accesses (leading to a SIGBUS) which could
occur in the ffmpeg ports.
Approved by: re (kib)
Reported by: tijl
MFC after: 3 days
Add ms_abi and sysv_abi attribute handling.
Based on a patch by Benno Rice!
This will help to develop EFI support.
Approved by: re (kib)
Verified by: benno
MFC after: 1 week
Remove invalid assert in DAGTypeLegalizer::RemapValue
There is a comment at the top of DAGTypeLegalizer::PerformExpensiveChecks
which, in part, says:
// Note that these invariants may not hold momentarily when processing a node:
// the node being processed may be put in a map before being marked Processed.
Unfortunately, this assert would be valid only if the above-mentioned invariant
held unconditionally. This was causing llc to assert when, in fact,
everything was fine.
Thanks to Richard Sandiford for investigating this issue!
Fixes PR16562.
This fixes assertions which could occur in the multimedia/ffmpeg1 and
multimedia/ffmpeg2 ports.
Approved by: re (hrs)
Reported by: Matthias Apitz <guru@unixarea.de>
MFC after: 3 days
On MIPS .dynamic is read-only and so a special section .rld_map is used
to store the pointer to the rtld information for debuggers. This
section had a hard coded size of 4 bytes which is not correct for
mips64. (Note that FreeBSD's rtld does not yet populate .rld_map.)
Sponsored by: DARPA, AFRL
Approved by: re (delphij)
The X86FixupLEAs pass for Intel Atom must not call
convertToThreeAddress on ADD16rr opcodes, if src1 != src, since that
would cause convertToThreeAddress to try to create a virtual register.
This is not permitted after register allocation, which is when the
X86FixupLEAs pass runs.
This patch fixes PR16785.
Pull in r191715 from upstream llvm trunk:
Forgot to add a break statement.
This should enable building the x11-toolskits/libXaw port with
CPUTYPE=atom.
Approved by: re (gjb)
Reported by: Kenta Suzumoto <kentas@hush.com>
MFC after: 3 days
linking of a shared library, leading to corrupt indexes in the dynamic
symbol table. This should fix the multimedia/ffmpegthumbnailer port.
Approved by: re (kib)
Reported by: swills
MFC after: 2 weeks
for. This is useful for software needing to know which architecture a
binary is built for as arm and armv6 have slight differences meaning only
some binaries build for one will work as expected on the other. It is
expected pkgng will be able to make use of this to simplify the logic to
determine which package ABI to use.
Approved by: re (kib)
ISelDAG: spot chain cycles involving MachineNodes
Previously, the DAGISel function WalkChainUsers was spotting that it
had entered already-selected territory by whether a node was a
MachineNode (amongst other things). Since it's fairly common practice
to insert MachineNodes during ISelLowering, this was not the correct
check.
Looking around, it seems that other nodes get their NodeId set to -1
upon selection, so this makes sure the same thing happens to all
MachineNodes and uses that characteristic to determine whether we
should stop looking for a loop during selection.
This should fix PR15840.
Specifically, this fixes the long-standing assertion failure when
compiling the multimedia/gstreamer port on i386. Thanks to Tijl
Coosemans for his help in getting upstream to fix it.
Approved by: re (marius)
not its header file, resulting in a mismatch between the lexer and parser
and strange errors when reading the configuration file.
Approved by: re (blanket)
