[PATCH], PING: Fix illegal cast to rtx (*insn_gen_fn) (rtx, ...)

On Fri, Jul 26, 2013 at 10:08 AM, Stefan Kristiansson wrote: > On Fri, Jul 26, 2013 at 08:51:41AM +0200, Uros Bizjak wrote: >> >> Stefan, can you resubmit an updated patch (with proposed update from [1])? >> >> I would really like to see this patch in the mainline. >> > > Attached is the updated p

Re: [PATCH, AArch64] Support NEG in vector registers for DI and SI mode

On 23 July 2013 13:35, Ian Bolton wrote: > 2013-07-23 Ian Bolton > > gcc/ > * config/aarch64/aarch64-simd.md (neg2): Offer alternative > that uses vector r> > Cheers, > Ian OK /Marcus

[C++ testcase, committed] PR 57101

Hi, I'm adding the testcase and keeping the PR open (the issue seems still present in 4_8-branch). Thanks, Paolo. // 2013-07-26 Paolo Carlini PR c++/57101 * g++.dg/cpp0x/pr57101.C: New. Index: g++.dg/cpp0x/pr57101.C =

[PATCH, AArch64] Skip gcc.dg/lower-subreg-1.c

Hi, This patch changes to skip gcc.dg/lower-subreg-1.c for aarch64*-*-*. The word mode in aarch64 is 64-bit so the lower-subreg pass won't happen in this test case. The test is currently skipped on aarch64 with lp64 due to the directive of "dg-require-effective-target ilp32", but fails when

Re: [c++-concepts] requires expressions

Fixed and committed, but I have a small follow-up related to parameter packs in requires clauses. The checking for unexpanded parameter packs treats the parameter declarations like regular PARM_DECL references and gives errors that they are unexpanded packs. 2013-07-26 Andrew Sutton * gcc/c

Re: Request to merge Undefined Behavior Sanitizer in

On Thu, Jul 25, 2013 at 10:33:46PM -0700, Andrew Pinski wrote: > What does it mean by "unsigned-integer-overflow"? Unsigned integers > never overflow. > Or maybe I misread the documentation here. Well, clang can sanitize even when unsigned int wraps. But this feature seems doubtful, I don't think

[c++-concepts] __is_same_as

This patch implements a new trait __is_same_as. This is foundational for future work on concepts in that it provides a mechanism for reasoning about type equivalences. It also provides the correct preconditions for __is_convertible_to as required in meta.rel. 2013-07-26 Andrew Sutton * gcc

Re: [c++-concepts] __is_same_as

Hi, On 07/26/2013 02:11 PM, Andrew Sutton wrote: This patch implements a new trait __is_same_as. This is foundational for future work on concepts in that it provides a mechanism for reasoning about type equivalences. Isn't the name a little misleading? I immediately wondered what was wrong with

Re: [testsuite, android] Disabling thread_local4.C and thread_local4g.C for Android.

Maxim, thank you for your input! That's for sure a better solution. __BIONIC__ is not defined in features.h so I had to change the include to ctype.h. (__BIONIC__ is defined in sys/ctypes.h, but it's not safe to include that directly..) tested on x86_64_unknow_linux and on android device. Is it

Re: [c++-concepts] __is_same_as

> Isn't the name a little misleading? I immediately wondered what was wrong > with std::is_same. IMHO something a little longer/technical clarifying that > the trait isn't just about comparing types is in order... Sure. First, it means we don't have to instantiate any class templates in order to

Re: [c++-concepts] __is_same_as

Hi, On 07/26/2013 03:23 PM, Andrew Sutton wrote: Isn't the name a little misleading? I immediately wondered what was wrong with std::is_same. IMHO something a little longer/technical clarifying that the trait isn't just about comparing types is in order... Sure. First, it means we don't have t

Re: [PATCH 00/18] resurrect automatic dependency tracking

> "Joseph" == Joseph S Myers writes: Joseph> We have a reliable reproducer for the bug, at least in the form Joseph> in which it appeared with the old patch: Joseph> - even Joseph> though it was never clear exactly what the make bug wa

Re: [c++-concepts] __is_same_as

> Thanks a lot. Now I'm afraid that some of these nice clarifications, > delicate technical details included, may get lost. Do you think they exist > already in some of your design documents, papers, etc. Then a reference in > the code would do. Otherwise, please consider adding some of the above t

[PATCH][ARM] Fix gcc.target/arm/minmax_minus.c

Hi all, The minmaxsi_minus.c test in gcc.target/arm/ was added to confirm that we generate two conditional subtractions instead of two conditional moves and an unconditional subtraction. It tests that by scanning for two conditional rsb instructions. But now, the arm backend generates sub instruct

Re: [PATCH][ARM] Fix gcc.target/arm/minmax_minus.c

On 26/07/13 15:44, Kyrylo Tkachov wrote: Hi all, The minmaxsi_minus.c test in gcc.target/arm/ was added to confirm that we generate two conditional subtractions instead of two conditional moves and an unconditional subtraction. It tests that by scanning for two conditional rsb instructions. But

[PATCH 00/11] Rewrite of pass management

The following patch series eliminates the mutable global variables representing GCC's passes, allowing for multiple compilation contexts in one process, potentially with different combinations of passes (e.g. JIT-compilation of JavaScript in one thread, JIT-compilation of OpenGL shader programs in

[PATCH 04/11] Automated conversion of passes to C++ classes

This is the automated part of the conversion of passes from C structs to C++ classes. It is generated by the refactor_passes.py script in https://github.com/davidmalcolm/gcc-refactoring-scripts The script has its own test suite: test_refactor_passes.py, and you can get an idea of the behavior of

[PATCH 01/11] Introduce beginnings of a pipeline class.

This patch introduces a gcc::pipeline class and moves various non-GTY globals relating to pass management into it. The gcc::context gains its first field: a pointer to the gcc::pipeline instance. It was previously sent as: http://gcc.gnu.org/ml/gcc-patches/2013-07/msg01090.html as part of: ht

[PATCH 05/11] Add -fno-rtti when building plugins.

With the conversion of passes to C++ classes, plugins that add custom passes must create them by creating their own derived classes of the relevant subclass of opt_pass. gcc itself is built with -fno-rtti, hence there is no RTTI available for the opt_pass class hierarchy. Hence plugins that creat

[PATCH 03/11] Handwritten part of conversion of passes to C++ classes

This patch is the hand-written part of the conversion of passes from C structs to C++ classes. It does not work without the subsequent autogenerated part, which is huge. Given that the autogenerated part of the conversion is very large (500k), for the sake of human comprehension I have kept the c

[PATCH 06/11] Rewrite how instances of passes are cloned

gcc/ Rewrite how instances of passes are cloned to remove assumptions about their sizes (thus allowing pass subclasses to have additional data fields, albeit non-GC-managed ones at this point). * passes.c (make_pass_instance): Now that passes have clone met

[PATCH 07/11] Introduce virtual functions in testsuite/gcc.dg/plugin/one_time_plugin.c

This is an example of converting the "gate" and "execute" functions of a pass into C++ virtual functions, so that in the next patch we can move a variable into member data of the opt_pass subclass. gcc/testsuite/ * gcc.dg/plugin/one_time_plugin.c: (one_pass_gate): convert to membe

[PATCH 08/11] Example of converting global state to per-pass state

gcc/testsuite/ Example of converting global state to per-pass state. * gcc.dg/plugin/one_time_plugin.c (one_pass::execute): Convert global state "static int counter" to... (one_pass::counter): ...this instance data. --- gcc/testsuite/gcc.dg/plugin/one_time_plugin.

[PATCH 10/11] Make gcc::context be GC-managed

This patch makes gcc::context instances be allocated within the GC-heap, and adds traversal hooks for GC/PCH so that a gcc::context can own refs to other GC-allocated objects. gcc/ * Makefile.in (GTFILES): Add context.h. * context.c (gcc::context::operator new): New. (gcc::

[PATCH 09/11] Support "gcc" namespace in gengtype

This patch adds enough special-casing to gengtype to allow it to cope with types that are within the gcc namespace. gcc/ * gengtype.c (type_for_name): Add special-case support for locating types within the "gcc::" namespace. (open_base_files): Emit a "using namespace gcc"

[PATCH 11/11] Make opt_pass and gcc::pipeline be GC-managed

This patch makes gcc::pipeline and opt_pass instances be allocated within the GC-heap, and adds traversal hooks for GC/PCH, so that passes can own refs to other GC-allocated objects. gcc/ Make opt_pass and gcc::pipeline be GC-managed, so that pass instances can own GC refs.

[PATCH 02/11] Generate pass-instances.def

Introduce a new gen-pass-instances.awk script, and use it at build time to make a pass-instances.def from passes.def. An example of the result can be seen at: http://dmalcolm.fedorapeople.org/gcc/2013-07-25/pass-instances.def The generated pass-instances.def contains similar content to passes.

[PATCH 02/11] Generate pass-instances.def

Introduce a new gen-pass-instances.awk script, and use it at build time to make a pass-instances.def from passes.def. An example of the result can be seen at: http://dmalcolm.fedorapeople.org/gcc/2013-07-25/pass-instances.def The generated pass-instances.def contains similar content to passes.

committed: Skip tests for epiphany that make invalid struct layout assumptions

Quoting Richard Guenther : On Sun, Sep 25, 2011 at 2:36 PM, Joern Rennecke wrote: This patch has not been reviewed for eight weeks. - Forwarded message from amyl...@spamcop.net -    Date: Mon, 01 Aug 2011 01:01:13 -0400    From: Joern Rennecke Reply-To: Joern Rennecke  Subject: RFA:

Re: [PATCH 00/18] resurrect automatic dependency tracking

Tom> Good idea. I'll dig up make 3.81 and give it a try soon, if not Tom> tomorrow, then early next week. I did this today. I downloaded and built GNU make 3.81 and put it in my PATH. Then I did "make -j2" builds in the gcc directory of each revision. I also ran "touch Makefile.in" and then "mak

[google/gcc-4_8] Fix problem with type signatures and template instantiations.

This patch is for the google/gcc-4_8 branch. I will also submit this to trunk separately. This fixes a problem with -fdebug-types-section where some types that have contain nested type templates or member function templates are given different type signatures in different compilation units, depend

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

Ok for google branches. Many changes in coverage.c (such as get_coverage_counts) and value-prof.c need to be in trunk too. David On Thu, Jul 25, 2013 at 10:02 AM, Teresa Johnson wrote: > This patch ports the remaining -fopt-info messages that had been added > to google/gcc-4_7 using the original

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

Thanks. I'll work on a trunk patch to send next week. Teresa On Fri, Jul 26, 2013 at 10:05 AM, Xinliang David Li wrote: > Ok for google branches. Many changes in coverage.c (such as > get_coverage_counts) and value-prof.c need to be in trunk too. > > David > > On Thu, Jul 25, 2013 at 10:02 AM, Te

[PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

Compiling the test program: #include enum x { a, b }; std::atomic v; bool test_strong() { x expected = a; return v.compare_exchange_strong(expected, b, std::memory_order_acq_rel); } bool test_weak() { x expected = a; return v.compare_exchange_weak(expected, b, std::memory_order_acq_re

Re: [PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

Hi, Nathan Froyd ha scritto: >Compiling the test program: > >#include > >enum x { a, b }; > >std::atomic v; > >bool test_strong() >{ > x expected = a; >return v.compare_exchange_strong(expected, b, >std::memory_order_acq_rel); >} > >bool test_weak() >{ > x expected = a; >return v.compare_exc

PS: committed: Skip tests for epiphany that make invalid struct layout assumptions

2013-07-26 Joern Rennecke * gcc.c-torture/execute/builtins/stpcpy-chk.x: New file. Index: gcc.c-torture/execute/builtins/stpncpy-chk.x === --- gcc.c-torture/execute/builtins/stpncpy-chk.x(revision 0) +++ gcc.c-tor

testsuite, committed: use -mshort-calls for epiphany when expecting function name but once

2013-07-26 Joern Rennecke * gcc.dg/pr27095.c: For Epiphany, add -mshort-calls. * gcc.dg/tree-ssa/loop-1.c: Likewise. Index: gcc.dg/pr27095.c === --- gcc.dg/pr27095.c(revision 201267) +++ gcc.dg/pr27095.c(

testsuite,committed: Disable gcc.dg/torture/pr37868.c / enable gcc.dg/sibcall-6.c for epiphany.

2013-07-26 Joern Rennecke * gcc.dg/torture/pr37868.c: Disable for epiphany. * gcc.dg/sibcall-6.c: Enable for epiphany. Index: gcc.dg/sibcall-6.c === --- gcc.dg/sibcall-6.c (revision 201267) +++ gcc.dg/sibcall-6.c

Re: [PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

Sure, I can do that. For maximum effectiveness, it'd be good to have it check the specializations for atomic<>, too. Is there something in the libstdc++ testsuite for iterating template instantiations over a list of types, or do I have to roll the list myself? Thanks, -Nathan - Original

Re: [PATCH] Add atomic type qualifier

Andrew MacLeod writes: > > What it doesn't do: > * It doesn't implement the C11 expression expansion into atomic > built-ins. ie, you can't write: > _Atomic int x; > x = 0; > and have the result be an atomic operation calling > __atomic_store (&x, 0). How would this work if you want

Re: Request to merge Undefined Behavior Sanitizer in

On 07/26/2013 01:33 AM, Andrew Pinski wrote: What does it mean by "unsigned-integer-overflow"? Unsigned integers never overflow. This is a common misconception. Conversion to an unsigned integer type never has undefined behavior, but unsigned integer arithmetic does. C11 says, If an excep

Re: [c++-concepts] requires expressions

Makes sense. Jason

Re: [PATCH] Add atomic type qualifier

On 07/26/2013 03:01 PM, Andi Kleen wrote: Andrew MacLeod writes: What it doesn't do: * It doesn't implement the C11 expression expansion into atomic built-ins. ie, you can't write: _Atomic int x; x = 0; and have the result be an atomic operation calling __atomic_store (&x, 0). How

Re: [PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

Hi, On 07/26/2013 08:42 PM, Nathan Froyd wrote: Sure, I can do that. For maximum effectiveness, it'd be good to have it check the specializations for atomic<>, too. Is there something in the libstdc++ testsuite for iterating template instantiations over a list of types, or do I have to roll

[patch] Fix problem with -fdebug-types-section and template instantiations.

I'm committing this patch to fix a problem with -fdebug-types-section where some types that have contain nested type templates or member function templates are given different type signatures in different compilation units, depending on which instantiations of those members are contained in the com

Re: Request to merge Undefined Behavior Sanitizer in

On Fri, 26 Jul 2013, Jason Merrill wrote: On 07/26/2013 01:33 AM, Andrew Pinski wrote: What does it mean by "unsigned-integer-overflow"? Unsigned integers never overflow. This is a common misconception. Conversion to an unsigned integer type never has undefined behavior, but unsigned intege

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

After converting so many more messages to the new dump framework with my previous change, I noticed a bug where the dumper was not always starting a dump message on a new line. The issue is that the dump framework does not always emit the source position info, but that was where the new line was be

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

is it better to hoist the newline dump before all the branches? David On Fri, Jul 26, 2013 at 1:06 PM, Teresa Johnson wrote: > After converting so many more messages to the new dump framework with my > previous change, I noticed a bug where the dumper was not always starting > a dump message on

Re: [PATCH] Add atomic type qualifier

On Fri, 26 Jul 2013, Andrew MacLeod wrote: > This patch adds an atomic type qualifier to GCC. It can be accessed via > __attribute__((atomic)) or in C11 mode via the _Atomic keyword. > HP, you might want to give this a try and see if you can get the alignment > correct for the cris port finally

Re: [testsuite, android] Disabling thread_local4.C and thread_local4g.C for Android.

On 27/07/2013, at 12:17 AM, Alexander Ivchenko wrote: > Maxim, thank you for your input! That's for sure a better solution. > > __BIONIC__ is not defined in features.h so I had to change the include > to ctype.h. (__BIONIC__ is defined in sys/ctypes.h, but it's not safe > to include that directly

Re: [Patch, Fortran] PR57530 (Part 2 of 3) Support TYPE => CLASS

Hi Tobias, > This patch is a follow up to the resolve part, which permits TYPE=>CLASS. > That approved patch is at > http://gcc.gnu.org/ml/fortran/2013-06/msg00049.html (I didn't want to > commit it without this trans*.c patch.) > > The attached patch adds support for: >TYPE => CLASS > additi

[Patch, Fortran, OOP] PR 57306: ICE on valid with class pointer initialization

Hi all, here is a fix for class pointer initialization. The problem was that 'gfc_get_symbol_decl' did not recognize class pointers properly. The first version of the patch posted in the PR induced problems with NULL initialization of class pointers, which are fixed by the attached second version.

Re: [PATCH] Add atomic type qualifier

Andrew MacLeod writes: >> > The way the standard is defined, any implicit operation like that is > seq_cst. If you want something other than seq-cst, you have to > explicitly call atomic_store (&x, 0, model). Thanks. This doesn't sound like a good default for x86. seq_cst requires somewhat exp

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

Yes, and in fact I moved it inside the check for "if (dump_kind)" since we wouldn't be emitting any message if that is null (in practice all the callers of dump_loc check this too). I don't really like the fact that the newline is being emitted for all messages emitted via the "_loc" versions of t

Re: [GOOGLE] Port remaining -fopt-info messages from google/4_7 (plus dump infrastructure enhancement)

ok. thanks, David On Fri, Jul 26, 2013 at 2:14 PM, Teresa Johnson wrote: > Yes, and in fact I moved it inside the check for "if (dump_kind)" > since we wouldn't be emitting any message if that is null (in practice > all the callers of dump_loc check this too). > > I don't really like the fact t

Re: [PATCH] Add atomic type qualifier

On 07/26/2013 05:13 PM, Andi Kleen wrote: Andrew MacLeod writes: The way the standard is defined, any implicit operation like that is seq_cst. If you want something other than seq-cst, you have to explicitly call atomic_store (&x, 0, model). Thanks. This doesn't sound like a good default for

Re: Request to merge Undefined Behavior Sanitizer in

On 07/26/2013 04:00 PM, Marc Glisse wrote: C11 also says: "A computation involving unsigned operands can never overflow, because a result that cannot be represented by the resulting unsigned integer type is reduced modulo the number that is one greater than the largest value that can be represen

Re: [PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

- Original Message - > On 07/26/2013 08:42 PM, Nathan Froyd wrote: > > Sure, I can do that. For maximum effectiveness, it'd be good to have it > > check the specializations for atomic<>, too. Is there something in the > > libstdc++ testsuite for iterating template instantiations over a li

Re: [patch] implement Cilk Plus simd loops on trunk

On 07/26/2013 11:46 AM, Aldy Hernandez wrote: How's this? What branch is this supposed to apply to? Jason

Re: [PATCH] Add atomic type qualifier

On Fri, 26 Jul 2013, Andrew MacLeod wrote: > This patch adds an atomic type qualifier to GCC. It can be accessed via > __attribute__((atomic)) or in C11 mode via the _Atomic keyword. Why the attribute - why not just the keyword? I'll review the patch in detail later (which will involve checkin

[PATCH v4] MIPS: IEEE 754-2008 features support

On Fri, 19 Jul 2013, Richard Sandiford wrote: > > while the 2008 > > NaN encoding has its shortcomings, e.g. unlike with the legacy encoding > > there's no single sNaN bit pattern to preset FPRs or variables with to > > catch uninitialised use that would work across all the floating-point > >

Re: [patch] implement Cilk Plus simd loops on trunk

On 07/26/2013 11:46 AM, Aldy Hernandez wrote: + Upon return, *INITP will contain the initialization expression, + *DECLP will contain the induction variable. + + *REAL_DECLP + *FOR_BLOCK */ DECLP is a variable declared in the for-init, if any. REAL_DECLP is the induction v

Re: [PATCH] fix generic std::atomic::compare_exchange_{weak,strong}

Hi, On 07/27/2013 01:07 AM, Nathan Froyd wrote: - Original Message - On 07/26/2013 08:42 PM, Nathan Froyd wrote: Sure, I can do that. For maximum effectiveness, it'd be good to have it check the specializations for atomic<>, too. Is there something in the libstdc++ testsuite for iter

Re: [PATCH] Add atomic type qualifier

On 07/26/2013 07:21 PM, Joseph S. Myers wrote: On Fri, 26 Jul 2013, Andrew MacLeod wrote: This patch adds an atomic type qualifier to GCC. It can be accessed via __attribute__((atomic)) or in C11 mode via the _Atomic keyword. Why the attribute - why not just the keyword? 2 reasons: 1 - we

Re: [PATCH] Fix the bug to check if lookup_stmt_eh_lp returns positive instead of non-zero

ping^2 Thanks, Dehao On Tue, Jul 16, 2013 at 5:40 PM, Dehao Chen wrote: > ping... > > Thanks, > Dehao > > On Mon, Jul 8, 2013 at 5:55 PM, Dehao Chen wrote: >> In lookup_stmt_eh_lp, negative return value indicates a MUST_NOT_THROW >> region index. In this case, we should *not* add an EH edge dur

Re: [c++-concepts] requires expressions

Andrew Sutton writes: | Fixed and committed, but I have a small follow-up related to parameter | packs in requires clauses. The checking for unexpanded parameter packs | treats the parameter declarations like regular PARM_DECL references | and gives errors that they are unexpanded packs. | | 201

Testsuite tweaks for the SPARC

This fixes the failures that recently appeared in the vectorization testsuite on the SPARC: - gcc.dg/vect/pr57705.c: the last loop uses a truncation so we can vectorize only the first 2 loops, - gcc.dg/vect/pr57741-2.c: the testcase requires vectors of floats, - gcc.dg/vect/pr57741-3.c: likewi