Pass in the calling (containing) Bfunction when invoking the Backend
method call_expression(), so as to handle the case where generation of
the call forces the creation of a temp var within the calling
function.
Reviewed-on: https://go-review.googlesource.com/43270
From-SVN: r247981
* system-linux-arm.ads (Memory_Size): Use Long_Integer'Size
instead of Word_Size.
Revert
2017-03-28 Andreas Schwab <schwab@suse.de>
PR ada/80117
* system-linux-aarch64-ilp32.ads: New file.
* gcc-interface/Makefile.in (LIBGNAT_TARGET_PAIRS_COMMON): Rename
from LIBGNAT_TARGET_PAIRS.
(LIBGNAT_TARGET_PAIRS_32, LIBGNAT_TARGET_PAIRS_64): Define.
(LIBGNAT_TARGET_PAIRS): Use LIBGNAT_TARGET_PAIRS_COMMON, and
LIBGNAT_TARGET_PAIRS_64 or LIBGNAT_TARGET_PAIRS_32 for -mabi=lp64
or -mabi=ilp32, resp.
From-SVN: r247979
[gcc]
2017-05-10 Will Schmidt <will_schmidt@vnet.ibm.com>
* config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
early expansion of vector divide builtins.
(builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
builtins identified as having unsigned arguments.
[gcc/testsuite]
2017-05-10 Will Schmidt <will_schmidt@vnet.ibm.com>
* gcc.target/powerpc/fold-vec-div-float.c: New.
* gcc.target/powerpc/fold-vec-div-floatdouble.c: New.
* gcc.target/powerpc/fold-vec-div-longlong.c: New.
From-SVN: r247977
[gcc]
2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
* config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
early GIMPLE expansion of vector multiplies.
From-SVN: r247974
Change memory allocation accounting for stacks to use stacks_sys,
since that seems to be what it is for.
Reviewed-on: https://go-review.googlesource.com/43297
From-SVN: r247967
2017-05-12 Richard Biener <rguenther@suse.de>
PR tree-optimization/80713
* tree-ssa-pre.c (remove_dead_inserted_code): Clear
inserted_exprs bit for not removed stmts.
* gcc.dg/torture/pr80713.c: New testcase.
From-SVN: r247963
With high -j parallelism the autofdo tests can randomly fail.
autofdo uses Linux perf to record profiling data.
Linux perf uses a locked perf buffer. By default it has
around 516k buffer per uid (/proc/sys/kernel/perf_event_mlock_kb).
An individual perf record tries to grab the full 516k,
which makes parallel perf record fail.
This patch limits the perf buffer for individual perf record to 8k.
With the default settings this allows a parallelism of the test
cases of 16, which is hopefully good enough
(if not would need to add some kind of semaphore, or ask
the user to increase the limit as root)
I also removed an unneeded -o perf.data option
Thanks to Marcin to finally spotting the problem.
Passes bootstrap and test on x86_64-linux. Ok for trunk?
gcc/testsuite/:
2017-05-12 Andi Kleen <ak@linux.intel.com>
PR testsuite/77684
* lib/target-supports.exp (profopt-perf-wrapper):
Add -m8 option to increase parallelism.
From-SVN: r247962
On MIPS, SIGABRT is defined like this:
#define SIGIOT 6
#define SIGABRT SIGIOT
This breaks addsig which tries to append __SIGIOT_ to SIGLIST. Signal
number 6 is later added to the output and go complains about a
duplicate signal number.
Fix by recursing once when obtaining the signal number from
gen-sysinfo.go if the signal is defined as an alias of another signal.
Also modify the sed expression to 's/.* = //' which is equivalent to
the original expression but is less misleading given that it might not
match a number.
Reviewed-on: https://go-review.googlesource.com/43252
From-SVN: r247948
gcc/ChangeLog:
2017-05-11 Carl Love <cel@us.ibm.com>
* config/rs6000/rs6000-c: Add support for built-in functions
vector unsigned char vec_popcnt (vector signed char)
vector unsigned char vec_popcnt (vector unsigned char)
vector unsigned short vec_popcnt (vector signed short)
vector unsigned short vec_popcnt (vector unsigned short)
vector unsigned int vec_popcnt (vector signed int)
vector unsigned int vec_popcnt (vector unsigned int)
vector unsigned long long vec_popcnt (vector signed long long)
vector unsigned long long vec_popcnt (vector unsigned long long)
vector signed long long vec_slo (vector signed long long,
vector signed char)
vector signed long long vec_slo (vector signed long long,
vector unsigned char)
vector unsigned long long vec_slo (vector unsigned long long,
vector signed char)
vector unsigned long long vec_slo (vector unsigned long long,
vector unsigned char)
* config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
* config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
* doc/extend.texi: Update the built-in documentation file for the
new built-in functions.
gcc/testsuite/ChangeLog:
2017-05-11 Carl Love <cel@us.ibm.com>
* gcc.target/powerpc/builtins-3-p8.c: Add tests for the new built-ins
to the test suite file.
* gcc.target/powerpc/builtins-3.c: Add tests for the new built-ins
to the test suite file.
From-SVN: r247933
The gc toolchain decides whether a goroutine is a system goroutine by
comparing startpc to a list of saved special PCs. In gccgo that
approach does not work as startpc is often a thunk that invokes the
real function with arguments, so the thunk address never matches the
saved special PCs.
This patch fixes gccgo's understanding of system goroutines. Since
there are only a limited number of them, we simply change each one to
mark itself as special.
This fixes stack dumps and functions like runtime.NumGoroutine to
behave more like gc. It also fixes the goprint test in the gc
testsuite.
Reviewed-on: https://go-review.googlesource.com/43156
From-SVN: r247931
2017-05-11 Jerry DeLisle <jvdelisle@gcc.gnu.org>
PR fortran/78659
* io.c (dtio_procs_present): Add new function to check for DTIO
procedures relative to I/O statement READ or WRITE.
(gfc_resolve_dt): Add namelist checks using the new function.
* resolve.c (dtio_procs_present): Remove function and related
namelist checks. (resolve_fl_namelist): Add check specific to
Fortran 95 restriction on namelist objects.
* gfortran.dg/namelist_91.f90: New test.
* gfortran.dg/namelist_92.f90: New test.
* gfortran.dg/namelist_93.f90: New test.
* gfortran.dg/namelist_94.f90: New test.
From-SVN: r247930
[gcc]
2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
PR target/80695
* config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
Account for direct move costs for vec_construct of integer
vectors.
[gcc/testsuite]
2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
PR target/80695
* gcc.target/powerpc/pr80695-p8.c: New file.
* gcc.target/powerpc/pr80695-p9.c: New file.
From-SVN: r247928
Glibc 2.17 made __secure_getenv an officially supported function, and
renamed it secure_getenv. The libgfortran configure has checked for
both of these, per
https://sourceware.org/glibc/wiki/Tips_and_Tricks/secure_getenv.
Unfortunately, while the dynamical library (libc.so) retains the
__secure_getenv symbol for backwards compatibility, the static library
(libc.a) does not. This means that a libgfortran.a compiled against an
older glibc will not work if one tries to link against a newer
libc.a. This creates problems for providing gfortran binary
distributions that work on as many target systems as possible.
Thus, retain the support for __secure_getenv but call it only via a
weak reference.
Regtested on x86_64-pc-linux-gnu.
2017-05-11 Janne Blomqvist <jb@gcc.gnu.org>
* libgfortran.h: HAVE_SECURE_GETENV: Don't check
HAVE___SECURE_GETENV.
* environ/runtime.c (secure_getenv): Use __secure_getenv via a
weak reference.
From-SVN: r247927
* cp-lang.c (get_global_decls, cxx_pushdecl, LANG_HOOK_GETDECLS)
LANG_HOOKS_PUSHDECL): Move to ...
* cp-objcp-common.c (cp_get_global_decls, cp_pushdec)
LANG_HOOK_DECLS, LANG_HOOKS_PUSHDECL): ... here.
* cp-objcp-common.h (cp_get_global_decls, cp_pushdecl): Declare.
((--This line, and those below, will be ignored--
M cp/ChangeLog
M cp/cp-lang.c
M cp/cp-objcp-common.c
M cp/cp-objcp-common.h
From-SVN: r247923
adding myself to Write After Approval list.
2017-05-11 Steven Munroe <munroesj@gcc.gnu.org>
* MAINTAINERS: Add myself to Write After Approval section.
From-SVN: r247913
