Re: [PATCH] i386: Enable small loop unrolling for O2

On Wed, Oct 26, 2022 at 7:53 AM Hongyu Wang wrote: > > Hi, > > Inspired by rs6000 and s390 port changes, this patch > enables loop unrolling for small size loop at O2 by default. > The default behavior is to unroll loop with unknown trip-count and > less than 4 insns by 1 time. > > This improves 5

[PATCH] i386: Enable small loop unrolling for O2

Hi, Inspired by rs6000 and s390 port changes, this patch enables loop unrolling for small size loop at O2 by default. The default behavior is to unroll loop with unknown trip-count and less than 4 insns by 1 time. This improves 548.exchange2 by 3.5% on icelake and 6% on zen3 with 1.2% codesize in

Re: [PATCH] RISC-V: Recognized Svinval and Svnapot extensions

On Tue, Oct 25, 2022 at 9:37 PM Bernhard Reutner-Fischer via Gcc-patches wrote: > > On 25 October 2022 08:17:33 CEST, Monk Chiang wrote: > >gcc/ChangeLog: > > > > >diff --git a/gcc/config/riscv/riscv-opts.h b/gcc/config/riscv/riscv-opts.h > >index 55e0bc0a0e9..63ac56a8ca0 100644 > >--- a/gcc/conf

Re: [PATCH] rs6000: using li/lis+oris/xoris to build constants

"Kewen.Lin" writes: > Hi Jeff, > > Sorry for late review, some comments are inline. > > on 2022/8/24 16:13, Jiufu Guo via Gcc-patches wrote: >> Hi, >> >> PR106708 constaint some constants which can be support by li/lis + >> oris/xoris. >> >> For constant C: >> if ((c & 0x80008000ULL) =

Re: [PATCH] testsuite: Fix failure in test pr105586.c [PR107171]

Hi Surya, on 2022/10/14 01:02, Surya Kumari Jangala via Gcc-patches wrote: > testsuite: Fix failure in test pr105586.c [PR107171] > > The test pr105586.c fails on a big endian system when run in 32bit > mode. The failure occurs as the test case does not guard against > unsupported __int128. > I

[PATCH] [x86] Enable V4BFmode and V2BFmode.

Enable V4BFmode and V2BFmode with the same ABI as V4HFmode and V2HFmode. No real operation is supported for them except for movement. This should solve PR target/107261. Also I notice there's redundancy in VALID_AVX512FP16_REG_MODE, and remove V2BFmode remove it. Bootstrapped and regtested on x86

[PATCH][PUSHED] Don't force DWARF4 for AutoFDO tests

Support for DWARF5 was added to create_gcov in https://github.com/google/autofdo so we no longer need to force DWARF4 for AutoFDO tests. Tested on x86_64-pc-linux-gnu. gcc/testsuite/ChangeLog: * lib/profopt.exp: Don't force DWARF4 for AutoFDO tests --- gcc/testsuite/lib/profopt.exp | 2 +

[r13-3476 Regression] FAIL: gcc.dg/vect/pr100756.c scan-tree-dump-not vect "epilog loop required" on Linux/x86_64

On Linux/x86_64, 19295e8607da2f743368fe6f5708146616aafa91 is the first bad commit commit 19295e8607da2f743368fe6f5708146616aafa91 Author: Richard Biener Date: Mon Oct 24 09:51:32 2022 +0200 tree-optimization/100756 - niter analysis and folding caused FAIL: gcc.dg/vect/pr100756.c -flto -f

Re: [PATCH 1/2] Add gcc/make-unique.h

On Fri, 2022-10-21 at 12:01 -0400, David Malcolm wrote: > This patch adds gcc/make-unique.h, containing a minimal C++11 > implementation of make_unique (std::make_unique is C++14). > > The followup patch uses this in dozens of places within the analyzer. > > Successfully bootstrapped & regrtested

[PATCH][PUSHED] Start using discriminators in AutoFDO

Tested on x86_64-pc-linux-gnu. gcc/ChangeLog: * auto-profile.cc (get_combined_location): Include discriminator in the returned combined location. (read_function_instance): Read discriminators from profiles. --- gcc/auto-profile.cc | 5 +++-- 1 file changed, 3 insertions(+

Re: [PATCH] diagnostics: Allow FEs to keep customizations for middle end [PR101551, PR106274]

On Tue, Oct 25, 2022 at 7:35 AM Richard Biener wrote: > > On Thu, Oct 20, 2022 at 1:09 AM Lewis Hyatt via Gcc-patches > wrote: > > > > Currently, the ipa-free-lang-data pass resets most of the frontend's > > diagnostic customizations, such as the diagnostic_finalizer that prints > > macro > > ex

[PATCH] [PR tree-optimization/107394] Canonicalize global franges as they are read back.

[Richi/Jakub/FP experts, does this sound like the right solution, or am I missing some subtle IPA/inlining issue?] The problem here is that we're inlining a global range with NANs into a function that has been tagged with __attribute__((optimize ("-ffinite-math-only"))). As the global range is co

[PATCH] Convert flag_finite_math_only uses in frange to HONOR_*.

[As Richi, and probably Jakub, have mentioned in the past...] As mentioned earlier, we should be using HONOR_* on types rather than flag_finite_math_only. Will commit pending tests. gcc/ChangeLog: * value-range.cc (frange::set): Use HONOR_*. (frange::verify_range): Same.

c++: Adjust synthetic template parm creation

We intend to mark synthetic template parameters (coming from use of auto parms), as DECL_VIRTUAL_P. The API of process_template_parm is awkwardly confusing, and we were marking the previous template parm (unless this was the first parm). process_template_parm returns the list of parms, when mos

Re: [PATCH v3] xtensa: Prepare the transition from Reload to LRA

Hi! On Wed, 2022-10-19 17:16:24 +0900, Takayuki 'January June' Suwa via Gcc-patches wrote: > * gcc/config/xtensa/xtensa.md: Add two new split patterns: > - splits DImode immediate load into two SImode ones > - puts out-of-constraint SImode constants into the constant pool

Re: [committed] More infrastructure to avoid bogus RTL on H8

Hi Jeff! On Mon, 2022-10-17 17:47:16 -0600, Jeff Law via Gcc-patches wrote: > --- a/gcc/config/h8300/h8300.cc > +++ b/gcc/config/h8300/h8300.cc > @@ -5531,6 +5531,32 @@ h8300_ok_for_sibcall_p (tree fndecl, tree) > >return 1; > } > + > +/* Return TRUE if OP is a PRE_INC or PRE_DEC > + in

Re: c: tree: target: C2x (...) function prototypes and va_start relaxation

On Tue, 25 Oct 2022, Richard Biener wrote: > You are missing to stream the new type flag in tree-streamer-{in,out}.cc > and checking for tree merging in lto-common.cc:compare_tree_sccs_1 > > Otherwise looks reasonable. Can you add a (multi TU) runtime testcase to the > torture exercising the fea

Re: [PATCH v2] bpf: add preserve_field_info builtin

> Hi Jose, > > Thanks for your comments. I think I've addressed them all in the updated > patch below. > >>>+ get_inner_reference (src, &bitsize, &bitpos, &var_off, &mode, &unsignedp, >>>+ &reversep, &volatilep); >> >>Since the information returned by the builtin is always cons

[committed] wwwdocs: contribute: Remove

The contents of this "description" meta tag really just duplicates the title of the page and does not contribute to search engine optimzation or otherwise, so simply drop it. Pushed. Gerald --- htdocs/contribute.html | 1 - 1 file changed, 1 deletion(-) diff --git a/htdocs/contribute.html b/ht

[pushed] c++: correct fold_operand change

Tested x86_64-pc-linux-gnu, applying to trunk. -- >8 -- Still want the conversion to bool. gcc/cp/ChangeLog: * constexpr.cc (find_failing_clause_r): Re-add the call to contextual_conv_bool. --- gcc/cp/constexpr.cc | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --

Re: [PATCH] c++ modules: verify_type failure with typedef enum [PR106848]

On Mon, 24 Oct 2022, Nathan Sidwell wrote: > On 10/21/22 09:11, Patrick Palka wrote: > > On Fri, 21 Oct 2022, Nathan Sidwell wrote: > > > > > > > Thanks for the explanation, it's a situation I didn;t anticipate and your > > > fix > > > is good. Could you add a comment about why you need to prop

[PATCH] tsan: fix test for machines without pthread_cond_clockwait

Hey, As per J. Wakeley's comments on https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100655 make the call to pthread_cond_clockwait conditional. This allows compilation on machines with older glibc versions. Cheers, Michael 2022-10-25 Michael de Lang * Fix testcase for pthread_cond_clockwai

Re: [PATCH v5 0/2] IBM zSystems: Improve storing asan frame_pc

On Tue, Sep 27, 2022 at 02:23:32AM +0200, Ilya Leoshkevich wrote: > This is a resend of v4 with slightly adjusted commit messages: > > v1: https://gcc.gnu.org/pipermail/gcc-patches/2019-July/525016.html > v2: https://gcc.gnu.org/pipermail/gcc-patches/2019-July/525069.html > v3: https://gcc.gnu.org

[PATCH v2] bpf: add preserve_field_info builtin

Hi Jose, Thanks for your comments. I think I've addressed them all in the updated patch below. >>+ get_inner_reference (src, &bitsize, &bitpos, &var_off, &mode, &unsignedp, >>+&reversep, &volatilep); > >Since the information returned by the builtin is always constant >(positi

[pushed] c++: constexpr-evaluate more assumes

Tested x86_64-pc-linux-gnu, applying to trunk. -- >8 -- The initial [[assume]] support avoided evaluating assumes with TREE_SIDE_EFFECTS set, such as calls, because we don't want any side-effects that change the constexpr state. This patch allows us to evaluate expressions with that flag set by

[pushed] c++: improve failed constexpr assume diagnostic

Tested x86_64-pc-linux-gnu, applying to trunk. -- >8 -- I noticed that we were printing "the comparison reduces to (x == 42)" when we should be able to give the value of x. Fixed by doing the same evaluation in diagnose_failing_condition that we already do in find_failing_clause. gcc/cp/ChangeL

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On Tue, Oct 25, 2022 at 11:39:31AM -0400, Jason Merrill wrote: > On 10/25/22 09:14, Marek Polacek wrote: > > On Tue, Oct 25, 2022 at 12:34:50PM +0100, Jonathan Wakely wrote: > > > On Mon, 24 Oct 2022 at 18:30, Jason Merrill wrote: > > > > > > > > On 10/21/22 19:28, Marek Polacek wrote: > > > > > W

[PATCH] rs6000: Add CCANY; replace signed by

This is in preparation for adding CCFP, and maybe CCEQ, and whatever other CC mode we may want later. CCANY is used for CC mode consumers that actually can take any of the four CR field bits. Tested on p7 and p9; committing, Segher 2022-10-25 Segher Boessenkool * config/rs6000/rs6

Re: [PATCH v2] c++: Implement -Wdangling-reference [PR106393]

On 10/25/22 11:21, Marek Polacek wrote: On Mon, Oct 24, 2022 at 01:30:42PM -0400, Jason Merrill wrote: On 10/21/22 19:28, Marek Polacek wrote: This patch implements a new experimental warning (enabled by -Wextra) to detect references bound to temporaries whose lifetime has ended. The Great!

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On 10/25/22 09:14, Marek Polacek wrote: On Tue, Oct 25, 2022 at 12:34:50PM +0100, Jonathan Wakely wrote: On Mon, 24 Oct 2022 at 18:30, Jason Merrill wrote: On 10/21/22 19:28, Marek Polacek wrote: When testing a previous version of the patch, there were many FAILs in libstdc++'s 22_locale/; al

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On Tue, Oct 25, 2022 at 12:50:36PM +0100, Jonathan Wakely wrote: > On Sat, 22 Oct 2022 at 00:28, Marek Polacek wrote: > > I've run the testsuite/bootstrap with the warning enabled by default. > > There were just a few FAILs: > > * g++.dg/warn/Wdangling-pointer-2.C > > * 20_util/any/misc/any_cast.cc

Re: [PATCH zero-call-used-regs] Add leafy mode for zero-call-used-regs

Hi, Alexandre, > On Oct 24, 2022, at 10:48 PM, Alexandre Oliva wrote: > > Hello, Qing, > > It was a pleasure to meet you at the Cauldron. Me too! > > > On Oct 21, 2022, Qing Zhao wrote: > >> Hi, Alexandre, >> Could you please explain a little bit on the motivation of this patch first? > >

[PATCH v2] c++: Implement -Wdangling-reference [PR106393]

On Mon, Oct 24, 2022 at 01:30:42PM -0400, Jason Merrill wrote: > On 10/21/22 19:28, Marek Polacek wrote: > > This patch implements a new experimental warning (enabled by -Wextra) to > > detect references bound to temporaries whose lifetime has ended. The > > Great! > > > primary motivation is th

[PATCH] testsuite: Windows paths use \ and not /

Without this patch, the following error is reported on Windows: In file included from t:\build\arm-none-eabi\include\c++\11.3.1\string:54, from t:\build\arm-none-eabi\include\c++\11.3.1\bits\locale_classes.h:40, from t:\build\arm-none-eabi\include\c++\11.3.1\b

PING^3 [PATCH] testsuite: Verify that module-mapper is available

Hi, Ping, https://gcc.gnu.org/pipermail/gcc-patches/2022-October/603544.html Kind regards, Torbjörn On 2022-10-14 09:42, Torbjorn SVENSSON wrote: Hi, Ping, https://gcc.gnu.org/pipermail/gcc-patches/2022-October/602843.html Kind regards, Torbjörn On 2022-10-05 11:17, Torbjorn SVENSSON wrote:

[PATCH] c++: Use in-process client when networking is disabled

Without the patch, the output for bad-mapper-3.C would be: /src/gcc/gcc/testsuite/g++.dg/modules/bad-mapper-3.C:2:1: error: unknown Compiled Module Interface: no such module As this line is unexpected, the test case would fail. The same problem can also be seen for g++.dg/modules/bad-mapper-2.C.

[PATCH] RISC-V: Fix a mistake in previous patch.

From: Ju-Zhe Zhong I noticed that I have made a mistake in previous patch: https://patchwork.sourceware.org/project/gcc/patch/20220817071950.271762-1-juzhe.zh...@rivai.ai/ The previous statement before this patch: bool need_barrier_p = (get_frame_size () + cfun->machine->frame.arg_pointer

Re: [PATCH] RISC-V: Recognized Svinval and Svnapot extensions

On 25 October 2022 08:17:33 CEST, Monk Chiang wrote: >gcc/ChangeLog: > >diff --git a/gcc/config/riscv/riscv-opts.h b/gcc/config/riscv/riscv-opts.h >index 55e0bc0a0e9..63ac56a8ca0 100644 >--- a/gcc/config/riscv/riscv-opts.h >+++ b/gcc/config/riscv/riscv-opts.h >@@ -162,6 +162,12 @@ enum stack_prot

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On Tue, Oct 25, 2022 at 12:34:50PM +0100, Jonathan Wakely wrote: > On Mon, 24 Oct 2022 at 18:30, Jason Merrill wrote: > > > > On 10/21/22 19:28, Marek Polacek wrote: > > > When testing a previous version of the patch, there were many FAILs in > > > libstdc++'s 22_locale/; all of them because the wa

[PATCH] unswitch most profitable condition first

When doing the loop unswitching re-org we promised to followup with improvements on the cost modeling. The following makes sure we try to unswitch on the most profitable condition first. As most profitable we pick the condition leading to the edge with the highest profile count. Note the profile

Re: [PATCH 1/2] ivopts: Revert computation of address cost complexity.

Hi Richard, > I don't follow how only having BASE + OFFSET addressing prevents > calculation of complexity for other addressing modes? Can you explain? It's the valid_mem_ref_p target hook that prevents complexity calculation for other addressing modes (for Mips and RISC-V). Here's the snippet o

Re: [PATCH 2/2] ivopts: Consider number of invariants when calculating register pressure.

Hi Richard, > don't you add n_invs twice now given > > unsigned n_old = data->regs_used, n_new = n_invs + n_cands; > unsigned regs_needed = n_new + n_old, available_regs = target_avail_regs; > > ? If you are referring to the "If we have enough registers." case, correct. After c18101f, for that

Re: [PATCH] ix86: Suggest unroll factor for loop vectorization

On Tue, Oct 25, 2022 at 7:46 AM Hongtao Liu wrote: > > Any comments? +@item x86-vect-unroll-min-ldst-threshold +The vectorizer will check with target information to determine whether unroll +it. This parameter is used to limit the mininum of loads and stores in the main +loop. It's odd to "limit

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On Sat, 22 Oct 2022 at 00:28, Marek Polacek wrote: > I've run the testsuite/bootstrap with the warning enabled by default. > There were just a few FAILs: > * g++.dg/warn/Wdangling-pointer-2.C > * 20_util/any/misc/any_cast.cc > * 20_util/forward/c_neg.cc > * 20_util/forward/f_neg.cc These two are X

Re: [PATCH] diagnostics: Allow FEs to keep customizations for middle end [PR101551, PR106274]

On Thu, Oct 20, 2022 at 1:09 AM Lewis Hyatt via Gcc-patches wrote: > > Currently, the ipa-free-lang-data pass resets most of the frontend's > diagnostic customizations, such as the diagnostic_finalizer that prints macro > expansion information, which is the subject of the two PRs. In most cases, >

Re: [PATCH] c++: Implement -Wdangling-reference [PR106393]

On Mon, 24 Oct 2022 at 18:30, Jason Merrill wrote: > > On 10/21/22 19:28, Marek Polacek wrote: > > When testing a previous version of the patch, there were many FAILs in > > libstdc++'s 22_locale/; all of them because the warning triggered on > > > >const test_type& obj = std::use_facet(std::lo

Re: c: tree: target: C2x (...) function prototypes and va_start relaxation

On Sat, Oct 22, 2022 at 1:03 AM Joseph Myers wrote: > > C2x allows function prototypes to be given as (...), a prototype > meaning a variable-argument function with no named arguments. To > allow such functions to access their arguments, requirements for > va_start calls are relaxed so it ignores

Re: [PATCH 1/2] ivopts: Revert computation of address cost complexity.

On Fri, Oct 21, 2022 at 3:56 PM Dimitrije Milosevic wrote: > > From: Dimitrije Milošević > > This patch reverts the computation of address cost complexity > to the legacy one. After f9f69dd, complexity is calculated > using the valid_mem_ref_p target hook. Architectures like > Mips only allow BAS

Re: [PATCH 2/2] ivopts: Consider number of invariants when calculating register pressure.

On Fri, Oct 21, 2022 at 3:57 PM Dimitrije Milosevic wrote: > > From: Dimitrije Milošević > > This patch slightly modifies register pressure model function to consider > both the number of invariants and the number of candidates, rather than > just the number of candidates. This used to be the cas

[PATCH] tree-optimization/107176 - SCEV analysis association issue

The following fixes a wrong-code issue caused by SCEV analysis associating an addition due trying to use tail-recursion in follow_ssa_edge_expr. That causes us to apply a conversion at the wrong point and thus miscompute the scalar evolution of an induction variable. This reverts the PR66375 fix

[PATCH] IRA: Make sure array is big enough

In commit 081c96621da, the call to resize_reg_info() was moved before the call to remove_scratches() and the latter one can increase the number of regs and that would cause an out of bounds usage on the reg_renumber global array. Without this patch, the following testcase randomly fails with: duri

RE: [PATCH] [X86_64]: Enable support for next generation AMD Zen4 CPU

[Public] Hi, On Mon, Oct 24, 2022 at 4:26 PM Alexander Monakov wrote: > > > This grew insn-automata.cc from 201502 lines to 639968 lines and the > > > build of the automata (genautomata) to several minutes in my dev tree. > > > > Yeah, in my unoptimized non-bootstrapp

[PATCH] Move NOP stripping in SCEV analysis

The following moves a pair of STRIP_USELESS_TYPE_CONVERSIONS to where it belongs and adds a comment on why we handle GENERIC there at all. Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed. * tree-scalar-evolution.cc (follow_ssa_edge_expr): Move STRIP_USELESS_TYPE_CONVER

Patch ping

Hi! On Mon, Oct 24, 2022 at 10:28:34AM -0600, Jeff Law wrote: > On 10/21/22 09:42, Jakub Jelinek wrote: > > On top of the pending > > https://gcc.gnu.org/pipermail/gcc-patches/2022-October/603665.html > > https://gcc.gnu.org/pipermail/gcc-patches/2022-October/604080.html > > the following patch ad

[committed] gimplify: Fix comment typos

Hi! While looking at gimple_boolify for PR107368, I've noticed 2 comment typos. Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk. 2022-10-25 Jakub Jelinek * gimplify.cc (gimple_boolify): Fix comment typos, prduce -> produce and There -> These. --- gc

[committed] gimplify: Call gimple_boolify on IFN_ASSUME argument [PR107368]

Hi! The following testcase ICEs in C, because assume attribute condition has int type rather than bool and the gimplification into GIMPLE_ASSUME assigns it into a bool variable. Fixed by calling gimple_boolify. Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk. 2022-10-2

[committed] gimplify: Don't add GIMPLE_ASSUME if errors were seen [PR107369]

Hi! The FEs emit errors about jumps into assume attribute conditions, but when we add GIMPLE_ASSUME for the condition which is reachable through those jumps, we can run into cfg verification diagnostics. Fixed by throwing the IFN_ASSUME away during gimplification if seen_error () - like we alread

Re: [PATCH] Add -gcodeview option

On Tue, 25 Oct 2022, Mark Harmstone wrote: On 24/10/22 12:08, Martin Storsjö wrote: Hmm, what does this end up passing to the linker in the end - does it just pass "-pdb="? (What does the "*" parameter do here?) If that's the case - that sounds reasonable - assuming that if a user passes an ex

[PATCH] tree-optimization/100756 - niter analysis and folding

niter analysis, specifically the part trying to simplify the computed maybe_zero condition against the loop header copying condition, is confused by us now simplifying _15 = n_8(D) * 4; if (_15 > 0) to _15 = n_8(D) * 4; if (n_8(D) > 0) which is perfectly sound at the point we do this tr