linaro-toolchain

linaro-toolchain@lists.linaro.org

9 participants
5482 discussions

[ACTIVITY] week ending 11 Jul 2021

by Richard Henderson

[ UM-2 ] * Sent v2 vdso + signal page patch set. * Fix #457 (tb_phys_invalidate_cont) * Fix #187 (arm on s390x) * Sent new rev for i386 fcs:fip patch set. * Sent v3 translator_use_goto_tb patch set. * Sent bswap cleanup. * Sent tcg_constant cleanups: alpha, hppa, openrisc, riscv. r~

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-release-aarch64-check_cross - Build # 27 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *binutils* in CI configuration tcwg_cross/gnu-release-aarch64-check_cross. So far, this commit has regressed CI configurations: - tcwg_cross/gnu-release-aarch64-check_cross Culprit: <cut> commit abe2a28aaa7a2bfd0f3061c72a98eb898976b721 Author: Nick Clifton <nickc(a)redhat.com> Date: Fri Jul 9 16:39:06 2021 +0100 Backportpatch to fix parsing of DWARF fuction and variable tags. PR 27484 bfd * dwarf2.c (scan_unit_for_symbols): Scan twice, once to accumulate function and variable tags and a second time to resolve their attributes. </cut> Results regressed to (for first_bad == abe2a28aaa7a2bfd0f3061c72a98eb898976b721) # reset_artifacts: -10 # build_abe binutils: -8 # build_abe stage1: -7 # build_abe linux: -6 # build_abe glibc: -5 # build_abe stage2: -4 # build_abe qemu: -3 # build_abe dejagnu: 0 # build_abe check_gcc --: 1 # Getting actual results from build directory /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/libitm.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/libgomp.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/libstdc++.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/g++.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/gcc.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/gfortran.sum # /home/tcwg-buildslave/workspace/tcwg_gnu_1/artifacts/build-abe2a28aaa7a2bfd0f3061c72a98eb898976b721/sumfiles/libatomic.sum # Manifest: gcc-compare-results/contrib/testsuite-management/flaky.xfail # Getting actual results from build directory base-artifacts/sumfiles # base-artifacts/sumfiles/libitm.sum # base-artifacts/sumfiles/libgomp.sum # base-artifacts/sumfiles/libstdc++.sum # base-artifacts/sumfiles/g++.sum # base-artifacts/sumfiles/gcc.sum # base-artifacts/sumfiles/gfortran.sum # base-artifacts/sumfiles/libatomic.sum # # SUCCESS: No unexpected failures. from (for last_good == 814d9ecf9b08018aa8ae2cd768ad8628945fd6aa) # reset_artifacts: -10 # build_abe binutils: -8 # build_abe stage1: -7 # build_abe linux: -6 # build_abe glibc: -5 # build_abe stage2: -4 # build_abe qemu: -3 # build_abe dejagnu: 0 # build_abe check_gcc --: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… Build top page/logs: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… Configuration details: Reproduce builds: <cut> mkdir investigate-binutils-abe2a28aaa7a2bfd0f3061c72a98eb898976b721 cd investigate-binutils-abe2a28aaa7a2bfd0f3061c72a98eb898976b721 git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_gnu-build.sh @@ artifacts/manifests/build-baseline.sh cd binutils # Reproduce first_bad build git checkout --detach abe2a28aaa7a2bfd0f3061c72a98eb898976b721 ../artifacts/test.sh # Reproduce last_good build git checkout --detach 814d9ecf9b08018aa8ae2cd768ad8628945fd6aa ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… Build log: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-release-aarch64-check_cross… Full commit (up to 1000 lines): <cut> commit abe2a28aaa7a2bfd0f3061c72a98eb898976b721 Author: Nick Clifton <nickc(a)redhat.com> Date: Fri Jul 9 16:39:06 2021 +0100 Backportpatch to fix parsing of DWARF fuction and variable tags. PR 27484 bfd * dwarf2.c (scan_unit_for_symbols): Scan twice, once to accumulate function and variable tags and a second time to resolve their attributes. --- bfd/ChangeLog | 10 +++++ bfd/dwarf2.c | 134 ++++++++++++++++++++++++++++++++++++++++++++++------------ 2 files changed, 117 insertions(+), 27 deletions(-) diff --git a/bfd/ChangeLog b/bfd/ChangeLog index a8b7415fadc..cd6f3706fe2 100644 --- a/bfd/ChangeLog +++ b/bfd/ChangeLog @@ -1,3 +1,13 @@ +2021-07-09 Nick Clifton <nickc(a)redhat.com> + + Backport this patch from the mainline: + 2021-03-02 Nick Clifton <nickc(a)redhat.com> + + PR 27484 + * dwarf2.c (scan_unit_for_symbols): Scan twice, once to accumulate + function and variable tags and a second time to resolve their + attributes. + 2021-05-03 Alan Modra <amodra(a)gmail.com> PR 27755 diff --git a/bfd/dwarf2.c b/bfd/dwarf2.c index 5651696c0f7..02618154a21 100644 --- a/bfd/dwarf2.c +++ b/bfd/dwarf2.c @@ -1484,6 +1484,8 @@ struct funcinfo struct arange arange; /* Where the symbol is defined. */ asection * sec; + /* The offset of the funcinfo from the start of the unit. */ + bfd_uint64_t unit_offset; }; struct lookup_funcinfo @@ -3304,6 +3306,15 @@ read_rangelist (struct comp_unit *unit, struct arange *arange, return read_rnglists (unit, arange, offset); } +static struct funcinfo * +lookup_func_by_offset (bfd_uint64_t offset, struct funcinfo * table) +{ + for (; table != NULL; table = table->prev_func) + if (table->unit_offset == offset) + return table; + return NULL; +} + static struct varinfo * lookup_var_by_offset (bfd_uint64_t offset, struct varinfo * table) { @@ -3317,7 +3328,6 @@ lookup_var_by_offset (bfd_uint64_t offset, struct varinfo * table) return NULL; } - /* DWARF2 Compilation unit functions. */ /* Scan over each die in a comp. unit looking for functions to add @@ -3330,7 +3340,8 @@ scan_unit_for_symbols (struct comp_unit *unit) bfd_byte *info_ptr = unit->first_child_die_ptr; bfd_byte *info_ptr_end = unit->end_ptr; int nesting_level = 0; - struct nest_funcinfo { + struct nest_funcinfo + { struct funcinfo *func; } *nested_funcs; int nested_funcs_size; @@ -3344,16 +3355,16 @@ scan_unit_for_symbols (struct comp_unit *unit) return FALSE; nested_funcs[nesting_level].func = 0; + /* PR 27484: We must scan the DIEs twice. The first time we look for + function and variable tags and accumulate them into their respective + tables. The second time through we process the attributes of the + functions/variables and augment the table entries. */ while (nesting_level >= 0) { unsigned int abbrev_number, bytes_read, i; struct abbrev_info *abbrev; - struct attribute attr; struct funcinfo *func; struct varinfo *var; - bfd_vma low_pc = 0; - bfd_vma high_pc = 0; - bfd_boolean high_pc_relative = FALSE; bfd_uint64_t current_offset; /* PR 17512: file: 9f405d9d. */ @@ -3365,7 +3376,7 @@ scan_unit_for_symbols (struct comp_unit *unit) FALSE, info_ptr_end); info_ptr += bytes_read; - if (! abbrev_number) + if (abbrev_number == 0) { nesting_level--; continue; @@ -3400,6 +3411,7 @@ scan_unit_for_symbols (struct comp_unit *unit) goto fail; func->tag = abbrev->tag; func->prev_func = unit->function_table; + func->unit_offset = current_offset; unit->function_table = func; unit->number_of_functions++; BFD_ASSERT (!unit->cached); @@ -3420,6 +3432,7 @@ scan_unit_for_symbols (struct comp_unit *unit) || abbrev->tag == DW_TAG_member) { size_t amt = sizeof (struct varinfo); + var = (struct varinfo *) bfd_zalloc (abfd, amt); if (var == NULL) goto fail; @@ -3438,6 +3451,89 @@ scan_unit_for_symbols (struct comp_unit *unit) nested_funcs[nesting_level].func = 0; } + for (i = 0; i < abbrev->num_attrs; ++i) + { + struct attribute attr; + + info_ptr = read_attribute (&attr, &abbrev->attrs[i], + unit, info_ptr, info_ptr_end); + if (info_ptr == NULL) + goto fail; + } + + if (abbrev->has_children) + { + nesting_level++; + + if (nesting_level >= nested_funcs_size) + { + struct nest_funcinfo *tmp; + + nested_funcs_size *= 2; + tmp = (struct nest_funcinfo *) + bfd_realloc (nested_funcs, + nested_funcs_size * sizeof (*nested_funcs)); + if (tmp == NULL) + goto fail; + nested_funcs = tmp; + } + nested_funcs[nesting_level].func = 0; + } + } + + /* This is the second pass over the abbrevs. */ + info_ptr = unit->first_child_die_ptr; + nesting_level = 0; + + while (nesting_level >= 0) + { + unsigned int abbrev_number, bytes_read, i; + struct abbrev_info *abbrev; + struct attribute attr; + struct funcinfo *func; + struct varinfo *var; + bfd_vma low_pc = 0; + bfd_vma high_pc = 0; + bfd_boolean high_pc_relative = FALSE; + bfd_uint64_t current_offset; + + /* PR 17512: file: 9f405d9d. */ + if (info_ptr >= info_ptr_end) + goto fail; + + current_offset = info_ptr - unit->info_ptr_unit; + abbrev_number = _bfd_safe_read_leb128 (abfd, info_ptr, &bytes_read, + FALSE, info_ptr_end); + info_ptr += bytes_read; + + if (! abbrev_number) + { + nesting_level--; + continue; + } + + abbrev = lookup_abbrev (abbrev_number, unit->abbrevs); + /* This should have been handled above. */ + BFD_ASSERT (abbrev != NULL); + + func = NULL; + var = NULL; + if (abbrev->tag == DW_TAG_subprogram + || abbrev->tag == DW_TAG_entry_point + || abbrev->tag == DW_TAG_inlined_subroutine) + { + func = lookup_func_by_offset (current_offset, unit->function_table); + if (func == NULL) + goto fail; + } + else if (abbrev->tag == DW_TAG_variable + || abbrev->tag == DW_TAG_member) + { + var = lookup_var_by_offset (current_offset, unit->variable_table); + if (var == NULL) + goto fail; + } + for (i = 0; i < abbrev->num_attrs; ++i) { info_ptr = read_attribute (&attr, &abbrev->attrs[i], @@ -3532,7 +3628,7 @@ scan_unit_for_symbols (struct comp_unit *unit) { _bfd_error_handler (_("DWARF error: could not find " "variable specification " - "at offset %lx"), + "at offset 0x%lx"), (unsigned long) attr.u.val); break; } @@ -3604,6 +3700,9 @@ scan_unit_for_symbols (struct comp_unit *unit) } } + if (abbrev->has_children) + nesting_level++; + if (high_pc_relative) high_pc += low_pc; @@ -3612,25 +3711,6 @@ scan_unit_for_symbols (struct comp_unit *unit) if (!arange_add (unit, &func->arange, low_pc, high_pc)) goto fail; } - - if (abbrev->has_children) - { - nesting_level++; - - if (nesting_level >= nested_funcs_size) - { - struct nest_funcinfo *tmp; - - nested_funcs_size *= 2; - tmp = (struct nest_funcinfo *) - bfd_realloc (nested_funcs, - nested_funcs_size * sizeof (*nested_funcs)); - if (tmp == NULL) - goto fail; - nested_funcs = tmp; - } - nested_funcs[nesting_level].func = 0; - } } free (nested_funcs); </cut>

3 years, 4 months

[ACTIVITY] report week ending 9 Jul

by Peter Maydell

Progress: * UM-2 [QEMU upstream maintainership] + Code review: - RTH's translator_use_goto_tb() series - Shashi's v5 ITS series + Arm pullreq * QEMU-406 [QEMU support for MVE (M-profile Vector Extension; Helium)] + Good progress with implementing more insns: VMULL poly, VIDUP, VIWDUP, VDDUP, VDWDUP, VCMP, VPT, VPSEL, VMLAS, VSHL T2, VRSHL T2, VQSHL T2, VQRSHL T2, VMAXV, VMINV, VMAXAV, VMINAV, VABAV, VMOVN, VQMOVUN, VQMOVN, VMLADAV, VMLSDAV, VMLA, VQDMLAH, VQRDMLAH, VQDMLASH, VQRDMLASH, VQABS, VQNEG, VMAXA, VMINA, VMOV-to-2gp, VMOV-from-2gp, VPNOT + Remaining is mostly floating point, and some loads and stores + Progress: 161/210 (76%) thanks -- PMM

3 years, 4 months

[ACTIVITY] week ending Jul. 11 2021

by Alex Bennée

QEMU Upstream Work ([UM-2]) =========================== - spent some time pulling together various maintainer trees for final 6.1 PRs - hopefully [this is the last iteration before Monday] - posted [PATCH for 6.1 v2 0/5] custom device configs Message-Id: <20210707131744.26027-1-alex.bennee(a)linaro.org> - chips away at the mega ARM re-factor patch set - posted [PATCH v1 0/2] some device emulation documentation cleanup Message-Id: <20210707184549.24436-1-alex.bennee(a)linaro.org> [UM-2] <https://projects.linaro.org/browse/UM-2> [this is the last iteration before Monday] <https://patchew.org/QEMU/20210709143005.1554-1-alex.bennee@linaro.org/> Other ===== - ARM CCA Announcement Event - Rust Training - Bunch of meetings with various members talking about future for QEMU - now the JIRA is migrated there are some things to add - wrote and abstract for KVM forum and [experimented with some visualisation] [experimented with some visualisation] <https://mastodon.org.uk/web/statuses/106535665831189241> Announced QEMU 6.1 on Linux Open Discussion Message-Id: <87zgv0tt7a.fsf(a)linaro.org> Completed Reviews [5/5] ======================= [PATCH v6 0/4] GitLab Custom Runners and Jobs (was: QEMU Gating CI) Message-Id: <20210608031425.833536-1-crosa(a)redhat.com> [PATCH v4 00/22] tests/docker: start using libvirt-ci's "lcitool" for dockerfiles Message-Id: <20210623142245.307776-1-berrange(a)redhat.com> [PATCH v4 0/5] plugins: New TCG plugin for cache modelling Message-Id: <20210623125458.450462-1-ma.mandourr(a)gmail.com> [PATCH v3 0/2] execlog TCG plugin to log instructions Message-Id: <20210702081307.1653644-1-erdnaxe(a)crans.org> [PATCH 0/3] ci: use cirrus-run to utilize Cirrus CI from GitLab CI Message-Id: <20210625172211.451010-1-berrange(a)redhat.com> Absences ======== Current Review Queue ==================== TODO [PATCH v5 00/10] GICv3 LPI and ITS feature implementation Message-Id: <20210630153156.9421-1-shashi.mallela(a)linaro.org> =========================================================================================================================== TODO [PATCH 00/17] tcg: breakpoint reorg Message-Id: <20210701152537.3330420-1-richard.henderson(a)linaro.org> =========================================================================================================== TODO [PATCH v2 00/28] accel/tcg: Introduce translator_use_goto_tb Message-Id: <20210630183226.3290849-1-richard.henderson(a)linaro.org> ==================================================================================================================================== TODO [PATCH v3 00/29] tcg: bswap improvements Message-Id: <20210626063631.2411938-1-richard.henderson(a)linaro.org> ================================================================================================================ -- Alex Bennée

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tk1/llvm-master-arm-spec2k6-Oz_LTO - Build # 10 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *llvm* in CI configuration tcwg_bmk_llvm_tk1/llvm-master-arm-spec2k6-Oz_LTO. So far, this commit has regressed CI configurations: - tcwg_bmk_llvm_tk1/llvm-master-arm-spec2k6-Oz_LTO Culprit: <cut> commit 51d334a845a082338735b0fdfc620a4b15fa26fe Author: Max Kazantsev <mkazantsev(a)azul.com> Date: Thu May 27 13:20:57 2021 +0700 [NFCI] Lazily evaluate SCEVs of PHIs Eager evaluation has cost of compile time. Only query them if they are required for proving predicates. </cut> Results regressed to (for first_bad == 51d334a845a082338735b0fdfc620a4b15fa26fe) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=thumb --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=thumb --set gcc_override_configure=--disable-libsanitizer: -5 # build_llvm true: -3 # true: 0 # benchmark -Oz_LTO_mthumb -- artifacts/build-51d334a845a082338735b0fdfc620a4b15fa26fe/results_id: 1 # 450.soplex,soplex_base.default regressed by 509909 # 473.astar,astar_base.default regressed by 4232401 # 453.povray,povray_base.default regressed by 166915 # 403.gcc,gcc_base.default regressed by 50755 from (for last_good == 59d938e649e62db0cef4903d495e838fbc6a6eb8) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=thumb --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=thumb --set gcc_override_configure=--disable-libsanitizer: -5 # build_llvm true: -3 # true: 0 # benchmark -Oz_LTO_mthumb -- artifacts/build-59d938e649e62db0cef4903d495e838fbc6a6eb8/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… Results ID of last_good: tk1_32/tcwg_bmk_llvm_tk1/bisect-llvm-master-arm-spec2k6-Oz_LTO/1235 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… Results ID of first_bad: tk1_32/tcwg_bmk_llvm_tk1/bisect-llvm-master-arm-spec2k6-Oz_LTO/1241 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… Configuration details: Reproduce builds: <cut> mkdir investigate-llvm-51d334a845a082338735b0fdfc620a4b15fa26fe cd investigate-llvm-51d334a845a082338735b0fdfc620a4b15fa26fe git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd llvm # Reproduce first_bad build git checkout --detach 51d334a845a082338735b0fdfc620a4b15fa26fe ../artifacts/test.sh # Reproduce last_good build git checkout --detach 59d938e649e62db0cef4903d495e838fbc6a6eb8 ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tk1-llvm-master-… Full commit (up to 1000 lines): <cut> commit 51d334a845a082338735b0fdfc620a4b15fa26fe Author: Max Kazantsev <mkazantsev(a)azul.com> Date: Thu May 27 13:20:57 2021 +0700 [NFCI] Lazily evaluate SCEVs of PHIs Eager evaluation has cost of compile time. Only query them if they are required for proving predicates. --- llvm/lib/Transforms/Scalar/LoopDeletion.cpp | 115 +++++++++++++++++----------- 1 file changed, 72 insertions(+), 43 deletions(-) diff --git a/llvm/lib/Transforms/Scalar/LoopDeletion.cpp b/llvm/lib/Transforms/Scalar/LoopDeletion.cpp index cd2a3fc48e3b..ed6b078fcf8e 100644 --- a/llvm/lib/Transforms/Scalar/LoopDeletion.cpp +++ b/llvm/lib/Transforms/Scalar/LoopDeletion.cpp @@ -136,30 +136,88 @@ static bool isLoopNeverExecuted(Loop *L) { return true; } +BasicBlock * +getSolePredecessorOnFirstIteration(BasicBlock *BB, Loop *L, + const DenseSet<BasicBlockEdge> &LiveEdges) { + if (BB == L->getHeader()) + return L->getLoopPredecessor(); + BasicBlock *OnlyPred = nullptr; + for (auto *Pred : predecessors(BB)) + if (OnlyPred != Pred && LiveEdges.count({ Pred, BB })) { + // 2 live preds. + if (OnlyPred) + return nullptr; + OnlyPred = Pred; + } + + assert(OnlyPred && "No live predecessors?"); + return OnlyPred; +} + +// Forward declaration. +static const SCEV * +getSCEVOnFirstIteration(Value *V, Loop *L, DominatorTree &DT, + ScalarEvolution &SE, + DenseMap<Value *, const SCEV *> &FirstIterSCEV, + const DenseSet<BasicBlockEdge> &LiveEdges); + static const SCEV * -getSCEVOnFirstIteration(Value *V, Loop *L, ScalarEvolution &SE, - DenseMap<Value *, const SCEV *> &FirstIterSCEV) { +getPHISCEVOnFirstIteration(PHINode *PN, Loop *L, DominatorTree &DT, + ScalarEvolution &SE, + DenseMap<Value *, const SCEV *> &FirstIterSCEV, + const DenseSet<BasicBlockEdge> &LiveEdges) { + auto *BB = PN->getParent(); + if (!L->contains(PN)) + return SE.getSCEV(PN); + // If this block has only one live pred, map its phis onto their SCEVs. + // Check if there is only one predecessor on 1st iteration. Note that because + // we iterate in RPOT, we have already visited all its (non-latch) + // predecessors. + auto *OnlyPred = getSolePredecessorOnFirstIteration(BB, L, LiveEdges); + if (!OnlyPred) + return SE.getSCEV(PN); + auto *Incoming = PN->getIncomingValueForBlock(OnlyPred); + if (DT.dominates(Incoming, BB->getTerminator())) + return getSCEVOnFirstIteration(Incoming, L, DT, SE, FirstIterSCEV, + LiveEdges); + return SE.getSCEV(PN); +} + +static const SCEV * +getSCEVOnFirstIteration(Value *V, Loop *L, DominatorTree &DT, + ScalarEvolution &SE, + DenseMap<Value *, const SCEV *> &FirstIterSCEV, + const DenseSet<BasicBlockEdge> &LiveEdges) { // Fist, check in cache. auto Existing = FirstIterSCEV.find(V); if (Existing != FirstIterSCEV.end()) return Existing->second; + const SCEV *S = nullptr; // TODO: Once ScalarEvolution supports getValueOnNthIteration for anything // else but AddRecs, it's a good use case for it. So far, just consider some // simple cases, like arithmetic operations. Value *LHS, *RHS; using namespace PatternMatch; - if (match(V, m_Add(m_Value(LHS), m_Value(RHS)))) { - const SCEV *LHSS = getSCEVOnFirstIteration(LHS, L, SE, FirstIterSCEV); - const SCEV *RHSS = getSCEVOnFirstIteration(RHS, L, SE, FirstIterSCEV); + if (auto *PN = dyn_cast<PHINode>(V)) { + S = getPHISCEVOnFirstIteration(PN, L, DT, SE, FirstIterSCEV, LiveEdges); + } else if (match(V, m_Add(m_Value(LHS), m_Value(RHS)))) { + const SCEV *LHSS = + getSCEVOnFirstIteration(LHS, L, DT, SE, FirstIterSCEV, LiveEdges); + const SCEV *RHSS = + getSCEVOnFirstIteration(RHS, L, DT, SE, FirstIterSCEV, LiveEdges); S = SE.getAddExpr(LHSS, RHSS); } else if (match(V, m_Sub(m_Value(LHS), m_Value(RHS)))) { - const SCEV *LHSS = getSCEVOnFirstIteration(LHS, L, SE, FirstIterSCEV); - const SCEV *RHSS = getSCEVOnFirstIteration(RHS, L, SE, FirstIterSCEV); + const SCEV *LHSS = + getSCEVOnFirstIteration(LHS, L, DT, SE, FirstIterSCEV, LiveEdges); + const SCEV *RHSS = + getSCEVOnFirstIteration(RHS, L, DT, SE, FirstIterSCEV, LiveEdges); S = SE.getMinusSCEV(LHSS, RHSS); } else if (match(V, m_Mul(m_Value(LHS), m_Value(RHS)))) { - const SCEV *LHSS = getSCEVOnFirstIteration(LHS, L, SE, FirstIterSCEV); - const SCEV *RHSS = getSCEVOnFirstIteration(RHS, L, SE, FirstIterSCEV); + const SCEV *LHSS = + getSCEVOnFirstIteration(LHS, L, DT, SE, FirstIterSCEV, LiveEdges); + const SCEV *RHSS = + getSCEVOnFirstIteration(RHS, L, DT, SE, FirstIterSCEV, LiveEdges); S = SE.getMulExpr(LHSS, RHSS); } else S = SE.getSCEV(V); @@ -185,7 +243,7 @@ static bool canProveExitOnFirstIteration(Loop *L, DominatorTree &DT, SmallPtrSet<BasicBlock *, 4> LiveBlocks; // Edges that are reachable on the 1st iteration. DenseSet<BasicBlockEdge> LiveEdges; - LiveBlocks.insert(L->getHeader()); + LiveBlocks.insert(Header); auto MarkLiveEdge = [&](BasicBlock *From, BasicBlock *To) { assert(LiveBlocks.count(From) && "Must be live!"); @@ -198,24 +256,6 @@ static bool canProveExitOnFirstIteration(Loop *L, DominatorTree &DT, MarkLiveEdge(BB, Succ); }; - // Check if there is only one predecessor on 1st iteration. Note that because - // we iterate in RPOT, we have already visited all its (non-latch) - // predecessors. - auto GetSolePredecessorOnFirstIteration = [&](BasicBlock * BB)->BasicBlock * { - if (BB == Header) - return L->getLoopPredecessor(); - BasicBlock *OnlyPred = nullptr; - for (auto *Pred : predecessors(BB)) - if (OnlyPred != Pred && LiveEdges.count({ Pred, BB })) { - // 2 live preds. - if (OnlyPred) - return nullptr; - OnlyPred = Pred; - } - - assert(OnlyPred && "No live predecessors?"); - return OnlyPred; - }; DenseMap<Value *, const SCEV *> FirstIterSCEV; SmallPtrSet<BasicBlock *, 4> Visited; @@ -250,19 +290,6 @@ static bool canProveExitOnFirstIteration(Loop *L, DominatorTree &DT, if (!Visited.count(Pred)) return false; - // If this block has only one live pred, map its phis onto their SCEVs. - if (auto *OnlyPred = GetSolePredecessorOnFirstIteration(BB)) - for (auto &PN : BB->phis()) { - if (!SE.isSCEVable(PN.getType())) - continue; - auto *Incoming = PN.getIncomingValueForBlock(OnlyPred); - if (DT.dominates(Incoming, BB->getTerminator())) { - const SCEV *IncSCEV = - getSCEVOnFirstIteration(Incoming, L, SE, FirstIterSCEV); - FirstIterSCEV[&PN] = IncSCEV; - } - } - using namespace PatternMatch; ICmpInst::Predicate Pred; Value *LHS, *RHS; @@ -281,8 +308,10 @@ static bool canProveExitOnFirstIteration(Loop *L, DominatorTree &DT, } // Can we prove constant true or false for this condition? - const SCEV *LHSS = getSCEVOnFirstIteration(LHS, L, SE, FirstIterSCEV); - const SCEV *RHSS = getSCEVOnFirstIteration(RHS, L, SE, FirstIterSCEV); + const SCEV *LHSS = + getSCEVOnFirstIteration(LHS, L, DT, SE, FirstIterSCEV, LiveEdges); + const SCEV *RHSS = + getSCEVOnFirstIteration(RHS, L, DT, SE, FirstIterSCEV, LiveEdges); // Only query for liveness of in-loop edge if another successor is also // in-loop. // TODO: isKnownPredicateAt is more powerful, but it's too compile time </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tk1/gnu-master-arm-spec2k6-O3_LTO - Build # 28 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *glibc* in CI configuration tcwg_bmk_gnu_tk1/gnu-master-arm-spec2k6-O3_LTO. So far, this commit has regressed CI configurations: - tcwg_bmk_gnu_tk1/gnu-master-arm-spec2k6-O3_LTO Culprit: <cut> commit 52a5fe70a2c77935afe807fb6e904e512ddd894e Author: Adhemerval Zanella <adhemerval.zanella(a)linaro.org> Date: Tue Mar 2 17:06:02 2021 -0300 Use 64 bit time_t stat internally For the legacy ABI with supports 32-bit time_t it calls the 64-bit time directly, since the LFS symbols calls the 64-bit time_t ones internally. Checked on i686-linux-gnu and x86_64-linux-gnu. Reviewed-by: Lukasz Majewski <lukma(a)denx.de> </cut> Results regressed to (for first_bad == 52a5fe70a2c77935afe807fb6e904e512ddd894e) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_LTO_marm -- artifacts/build-52a5fe70a2c77935afe807fb6e904e512ddd894e/results_id: 1 # 465.tonto,tonto_base.default regressed by 104 from (for last_good == a318262bc0081ab83e3f3c90e50462f99148605e) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_LTO_marm -- artifacts/build-a318262bc0081ab83e3f3c90e50462f99148605e/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… Results ID of last_good: tk1_32/tcwg_bmk_gnu_tk1/bisect-gnu-master-arm-spec2k6-O3_LTO/1228 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… Results ID of first_bad: tk1_32/tcwg_bmk_gnu_tk1/bisect-gnu-master-arm-spec2k6-O3_LTO/1189 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… Configuration details: Reproduce builds: <cut> mkdir investigate-glibc-52a5fe70a2c77935afe807fb6e904e512ddd894e cd investigate-glibc-52a5fe70a2c77935afe807fb6e904e512ddd894e git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd glibc # Reproduce first_bad build git checkout --detach 52a5fe70a2c77935afe807fb6e904e512ddd894e ../artifacts/test.sh # Reproduce last_good build git checkout --detach a318262bc0081ab83e3f3c90e50462f99148605e ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-master-ar… Full commit (up to 1000 lines): <cut> commit 52a5fe70a2c77935afe807fb6e904e512ddd894e Author: Adhemerval Zanella <adhemerval.zanella(a)linaro.org> Date: Tue Mar 2 17:06:02 2021 -0300 Use 64 bit time_t stat internally For the legacy ABI with supports 32-bit time_t it calls the 64-bit time directly, since the LFS symbols calls the 64-bit time_t ones internally. Checked on i686-linux-gnu and x86_64-linux-gnu. Reviewed-by: Lukasz Majewski <lukma(a)denx.de> --- csu/check_fds.c | 8 ++++---- elf/dl-load.c | 8 ++++---- elf/dl-misc.c | 4 ++-- elf/dl-profile.c | 4 ++-- iconv/gconv_cache.c | 4 ++-- include/dirent.h | 2 +- include/file_change_detection.h | 6 +++--- inet/rcmd.c | 6 +++--- intl/loadmsgcat.c | 4 ++-- io/Makefile | 4 ++-- io/file_change_detection.c | 10 +++++----- io/getdirname.c | 6 +++--- libio/filedoalloc.c | 2 +- libio/fileops.c | 8 ++++---- libio/oldfileops.c | 2 +- libio/wfileops.c | 2 +- locale/loadarchive.c | 8 ++++---- locale/loadlocale.c | 6 +++--- nscd/nscd_helper.c | 4 ++-- nss/nss_database.c | 4 ++-- sysdeps/posix/dl-fileid.h | 4 ++-- sysdeps/posix/euidaccess.c | 4 ++-- sysdeps/posix/getaddrinfo.c | 18 +++++++++--------- sysdeps/posix/getcwd.c | 15 ++++++++------- sysdeps/posix/pathconf.c | 4 ++-- sysdeps/posix/sysconf.c | 4 ++-- sysdeps/posix/tempname.c | 8 ++++---- sysdeps/unix/sysv/linux/fdopendir.c | 4 ++-- sysdeps/unix/sysv/linux/fexecve.c | 4 ++-- sysdeps/unix/sysv/linux/opendir.c | 7 ++++--- sysdeps/unix/sysv/linux/pathconf.c | 5 +++-- sysdeps/unix/sysv/linux/ttyname.h | 10 ++-------- sysdeps/unix/sysv/linux/ttyname_r.c | 16 ++++++++-------- sysvipc/ftok.c | 4 ++-- time/tzfile.c | 6 +++--- 35 files changed, 106 insertions(+), 109 deletions(-) diff --git a/csu/check_fds.c b/csu/check_fds.c index fbab28c284..6dc67334af 100644 --- a/csu/check_fds.c +++ b/csu/check_fds.c @@ -69,10 +69,10 @@ check_one_fd (int fd, int mode) Note that the following code assumes that STDIN_FILENO, STDOUT_FILENO, STDERR_FILENO are the three lowest file decsriptor numbers, in this order. */ - struct stat64 st; - if (__builtin_expect (nullfd != fd, 0) - || __builtin_expect (__fstat64 (fd, &st), 0) != 0 - || __builtin_expect (S_ISCHR (st.st_mode), 1) == 0 + struct __stat64_t64 st; + if (__glibc_unlikely (nullfd != fd) + || __glibc_likely (__fstat64_time64 (fd, &st) != 0) + || __glibc_unlikely (S_ISCHR (st.st_mode) == 0) || st.st_rdev != dev) /* We cannot even give an error message here since it would run into the same problems. */ diff --git a/elf/dl-load.c b/elf/dl-load.c index 918ec7546c..a08df001af 100644 --- a/elf/dl-load.c +++ b/elf/dl-load.c @@ -1946,11 +1946,11 @@ open_path (const char *name, size_t namelen, int mode, { /* We failed to open machine dependent library. Let's test whether there is any directory at all. */ - struct stat64 st; + struct __stat64_t64 st; buf[buflen - namelen - 1] = '\0'; - if (__stat64 (buf, &st) != 0 + if (__stat64_time64 (buf, &st) != 0 || ! S_ISDIR (st.st_mode)) /* The directory does not exist or it is no directory. */ this_dir->status[cnt] = nonexisting; @@ -1968,9 +1968,9 @@ open_path (const char *name, size_t namelen, int mode, /* This is an extra security effort to make sure nobody can preload broken shared objects which are in the trusted directories and so exploit the bugs. */ - struct stat64 st; + struct __stat64_t64 st; - if (__fstat64 (fd, &st) != 0 + if (__fstat64_time64 (fd, &st) != 0 || (st.st_mode & S_ISUID) == 0) { /* The shared object cannot be tested for being SUID diff --git a/elf/dl-misc.c b/elf/dl-misc.c index d4803bba4e..b256d792c6 100644 --- a/elf/dl-misc.c +++ b/elf/dl-misc.c @@ -43,11 +43,11 @@ void * _dl_sysdep_read_whole_file (const char *file, size_t *sizep, int prot) { void *result = MAP_FAILED; - struct stat64 st; + struct __stat64_t64 st; int fd = __open64_nocancel (file, O_RDONLY | O_CLOEXEC); if (fd >= 0) { - if (__fstat64 (fd, &st) >= 0) + if (__fstat64_time64 (fd, &st) >= 0) { *sizep = st.st_size; diff --git a/elf/dl-profile.c b/elf/dl-profile.c index 35b03959bc..088c023d95 100644 --- a/elf/dl-profile.c +++ b/elf/dl-profile.c @@ -185,7 +185,7 @@ _dl_start_profile (void) { char *filename; int fd; - struct stat64 st; + struct __stat64_t64 st; const ElfW(Phdr) *ph; ElfW(Addr) mapstart = ~((ElfW(Addr)) 0); ElfW(Addr) mapend = 0; @@ -342,7 +342,7 @@ _dl_start_profile (void) return; } - if (__fstat64 (fd, &st) < 0 || !S_ISREG (st.st_mode)) + if (__fstat64_time64 (fd, &st) < 0 || !S_ISREG (st.st_mode)) { /* Not stat'able or not a regular file => don't use it. */ errstr = "%s: cannot stat file: %s\n"; diff --git a/iconv/gconv_cache.c b/iconv/gconv_cache.c index 642cc731eb..68d6386d01 100644 --- a/iconv/gconv_cache.c +++ b/iconv/gconv_cache.c @@ -48,7 +48,7 @@ int __gconv_load_cache (void) { int fd; - struct stat64 st; + struct __stat64_t64 st; struct gconvcache_header *header; /* We cannot use the cache if the GCONV_PATH environment variable is @@ -64,7 +64,7 @@ __gconv_load_cache (void) return -1; /* Get information about the file. */ - if (__builtin_expect (__fstat64 (fd, &st), 0) < 0 + if (__glibc_unlikely (__fstat64_time64 (fd, &st) < 0) /* We do not have to start looking at the file if it cannot contain at least the cache header. */ || (size_t) st.st_size < sizeof (struct gconvcache_header)) diff --git a/include/dirent.h b/include/dirent.h index f1d7e8359c..d7567f5e86 100644 --- a/include/dirent.h +++ b/include/dirent.h @@ -49,7 +49,7 @@ extern int __versionsort64 (const struct dirent64 **a, const struct dirent64 **b) __attribute_pure__; extern DIR *__alloc_dir (int fd, bool close_fd, int flags, - const struct stat64 *statp) + const struct __stat64_t64 *statp) __nonnull (4) attribute_hidden; extern __typeof (rewinddir) __rewinddir; extern __typeof (seekdir) __seekdir; diff --git a/include/file_change_detection.h b/include/file_change_detection.h index 9de2bd0b79..953de7c7fd 100644 --- a/include/file_change_detection.h +++ b/include/file_change_detection.h @@ -33,8 +33,8 @@ struct file_change_detection off64_t size; ino64_t ino; - struct timespec mtime; - struct timespec ctime; + struct __timespec64 mtime; + struct __timespec64 ctime; }; /* Returns true if *LEFT and *RIGHT describe the same version of the @@ -45,7 +45,7 @@ bool __file_is_unchanged (const struct file_change_detection *left, /* Extract file change information to *FILE from the stat buffer *ST. */ void __file_change_detection_for_stat (struct file_change_detection *file, - const struct stat64 *st); + const struct __stat64_t64 *st); /* Writes file change information for PATH to *FILE. Returns true on success. For benign errors, *FILE is cleared, and true is diff --git a/inet/rcmd.c b/inet/rcmd.c index 0cfdaee15e..0d6b595572 100644 --- a/inet/rcmd.c +++ b/inet/rcmd.c @@ -468,14 +468,14 @@ ruserok (const char *rhost, int superuser, const char *ruser, static FILE * iruserfopen (const char *file, uid_t okuser) { - struct stat64 st; + struct __stat64_t64 st; char *cp = NULL; FILE *res = NULL; /* If not a regular file, if owned by someone other than user or root, if writeable by anyone but the owner, or if hardlinked anywhere, quit. */ - if (__lstat64 (file, &st)) + if (__lstat64_time64 (file, &st)) cp = _("lstat failed"); else if (!S_ISREG (st.st_mode)) cp = _("not regular file"); @@ -484,7 +484,7 @@ iruserfopen (const char *file, uid_t okuser) res = fopen (file, "rce"); if (!res) cp = _("cannot open"); - else if (__fstat64 (fileno (res), &st) < 0) + else if (__fstat64_time64 (fileno (res), &st) < 0) cp = _("fstat failed"); else if (st.st_uid && st.st_uid != okuser) cp = _("bad owner"); diff --git a/intl/loadmsgcat.c b/intl/loadmsgcat.c index 4fda9b0219..9b10697dd9 100644 --- a/intl/loadmsgcat.c +++ b/intl/loadmsgcat.c @@ -756,7 +756,7 @@ _nl_load_domain (struct loaded_l10nfile *domain_file, int fd = -1; size_t size; #ifdef _LIBC - struct stat64 st; + struct __stat64_t64 st; #else struct stat st; #endif @@ -804,7 +804,7 @@ _nl_load_domain (struct loaded_l10nfile *domain_file, /* We must know about the size of the file. */ if ( #ifdef _LIBC - __builtin_expect (__fstat64 (fd, &st) != 0, 0) + __glibc_unlikely (__fstat64_time64 (fd, &st) != 0) #else __builtin_expect (fstat (fd, &st) != 0, 0) #endif diff --git a/io/Makefile b/io/Makefile index ba8bd37355..1a16990205 100644 --- a/io/Makefile +++ b/io/Makefile @@ -69,7 +69,7 @@ tests := test-utime test-stat test-stat2 test-lfs tst-getcwd \ tst-posix_fallocate tst-posix_fallocate64 \ tst-fts tst-fts-lfs tst-open-tmpfile \ tst-copy_file_range tst-getcwd-abspath tst-lockf \ - tst-ftw-lnk tst-file_change_detection tst-lchmod \ + tst-ftw-lnk tst-lchmod \ tst-ftw-bz26353 tst-stat tst-stat-lfs \ tst-utime \ tst-utimes \ @@ -90,7 +90,7 @@ tests-time64 := \ tst-utimes-time64 \ # Likewise for statx, but we do not need static linking here. -tests-internal += tst-statx +tests-internal += tst-statx tst-file_change_detection tests-static += tst-statx ifeq ($(run-built-tests),yes) diff --git a/io/file_change_detection.c b/io/file_change_detection.c index 9871bc9166..fad0907223 100644 --- a/io/file_change_detection.c +++ b/io/file_change_detection.c @@ -44,7 +44,7 @@ libc_hidden_def (__file_is_unchanged) void __file_change_detection_for_stat (struct file_change_detection *file, - const struct stat64 *st) + const struct __stat64_t64 *st) { if (S_ISDIR (st->st_mode)) /* Treat as empty file. */ @@ -66,8 +66,8 @@ bool __file_change_detection_for_path (struct file_change_detection *file, const char *path) { - struct stat64 st; - if (__stat64 (path, &st) != 0) + struct __stat64_t64 st; + if (__stat64_time64 (path, &st) != 0) switch (errno) { case EACCES: @@ -104,8 +104,8 @@ __file_change_detection_for_fp (struct file_change_detection *file, } else { - struct stat64 st; - if (__fstat64 (__fileno (fp), &st) != 0) + struct __stat64_t64 st; + if (__fstat64_time64 (__fileno (fp), &st) != 0) /* If we already have a file descriptor, all errors are fatal. */ return false; else diff --git a/io/getdirname.c b/io/getdirname.c index 32b5ab4b6c..f29e910b37 100644 --- a/io/getdirname.c +++ b/io/getdirname.c @@ -28,12 +28,12 @@ char * get_current_dir_name (void) { char *pwd; - struct stat64 dotstat, pwdstat; + struct __stat64_t64 dotstat, pwdstat; pwd = getenv ("PWD"); if (pwd != NULL - && __stat64 (".", &dotstat) == 0 - && __stat64 (pwd, &pwdstat) == 0 + && __stat64_time64 (".", &dotstat) == 0 + && __stat64_time64 (pwd, &pwdstat) == 0 && pwdstat.st_dev == dotstat.st_dev && pwdstat.st_ino == dotstat.st_ino) /* The PWD value is correct. Use it. */ diff --git a/libio/filedoalloc.c b/libio/filedoalloc.c index 7afa3cb683..c792eec88f 100644 --- a/libio/filedoalloc.c +++ b/libio/filedoalloc.c @@ -78,7 +78,7 @@ _IO_file_doallocate (FILE *fp) { size_t size; char *p; - struct stat64 st; + struct __stat64_t64 st; size = BUFSIZ; if (fp->_fileno >= 0 && __builtin_expect (_IO_SYSSTAT (fp, &st), 0) >= 0) diff --git a/libio/fileops.c b/libio/fileops.c index a71a567547..e8ee374821 100644 --- a/libio/fileops.c +++ b/libio/fileops.c @@ -545,7 +545,7 @@ libc_hidden_ver (_IO_new_file_underflow, _IO_file_underflow) static int mmap_remap_check (FILE *fp) { - struct stat64 st; + struct __stat64_t64 st; if (_IO_SYSSTAT (fp, &st) == 0 && S_ISREG (st.st_mode) && st.st_size != 0 @@ -663,7 +663,7 @@ decide_maybe_mmap (FILE *fp) file descriptors are for mmap-able objects and on 32-bit machines we don't want to map files which are too large since this would require too much virtual memory. */ - struct stat64 st; + struct __stat64_t64 st; if (_IO_SYSSTAT (fp, &st) == 0 && S_ISREG (st.st_mode) && st.st_size != 0 @@ -962,7 +962,7 @@ _IO_new_file_seekoff (FILE *fp, off64_t offset, int dir, int mode) break; case _IO_seek_end: { - struct stat64 st; + struct __stat64_t64 st; if (_IO_SYSSTAT (fp, &st) == 0 && S_ISREG (st.st_mode)) { offset += st.st_size; @@ -1145,7 +1145,7 @@ libc_hidden_def (_IO_file_seek) int _IO_file_stat (FILE *fp, void *st) { - return __fstat64 (fp->_fileno, (struct stat64 *) st); + return __fstat64_time64 (fp->_fileno, (struct __stat64_t64 *) st); } libc_hidden_def (_IO_file_stat) diff --git a/libio/oldfileops.c b/libio/oldfileops.c index ed235360ad..ed3c98bd6d 100644 --- a/libio/oldfileops.c +++ b/libio/oldfileops.c @@ -487,7 +487,7 @@ _IO_old_file_seekoff (FILE *fp, off64_t offset, int dir, int mode) break; case _IO_seek_end: { - struct stat64 st; + struct __stat64_t64 st; if (_IO_SYSSTAT (fp, &st) == 0 && S_ISREG (st.st_mode)) { offset += st.st_size; diff --git a/libio/wfileops.c b/libio/wfileops.c index 5d23566171..37f44780f8 100644 --- a/libio/wfileops.c +++ b/libio/wfileops.c @@ -840,7 +840,7 @@ _IO_wfile_seekoff (FILE *fp, off64_t offset, int dir, int mode) break; case _IO_seek_end: { - struct stat64 st; + struct __stat64_t64 st; if (_IO_SYSSTAT (fp, &st) == 0 && S_ISREG (st.st_mode)) { offset += st.st_size; diff --git a/locale/loadarchive.c b/locale/loadarchive.c index 4177fc8972..512769eaec 100644 --- a/locale/loadarchive.c +++ b/locale/loadarchive.c @@ -78,7 +78,7 @@ static struct archmapped *archmapped; ARCHMAPPED points to this; if mapping the archive header failed, then headmap.ptr is null. */ static struct archmapped headmap; -static struct stat64 archive_stat; /* stat of archive when header mapped. */ +static struct __stat64_t64 archive_stat; /* stat of archive when header mapped. */ /* Record of locales that we have already loaded from the archive. */ struct locale_in_archive @@ -207,7 +207,7 @@ _nl_load_locale_from_archive (int category, const char **namep) /* Cannot open the archive, for whatever reason. */ return NULL; - if (__fstat64 (fd, &archive_stat) == -1) + if (__fstat64_time64 (fd, &archive_stat) == -1) { /* stat failed, very strange. */ close_and_out: @@ -396,7 +396,7 @@ _nl_load_locale_from_archive (int category, const char **namep) /* Open the file if it hasn't happened yet. */ if (fd == -1) { - struct stat64 st; + struct __stat64_t64 st; fd = __open_nocancel (archfname, O_RDONLY|O_LARGEFILE|O_CLOEXEC); if (fd == -1) @@ -405,7 +405,7 @@ _nl_load_locale_from_archive (int category, const char **namep) /* Now verify we think this is really the same archive file we opened before. If it has been changed we cannot trust the header we read previously. */ - if (__fstat64 (fd, &st) < 0 + if (__fstat64_time64 (fd, &st) < 0 || st.st_size != archive_stat.st_size || st.st_mtime != archive_stat.st_mtime || st.st_dev != archive_stat.st_dev diff --git a/locale/loadlocale.c b/locale/loadlocale.c index 9c03490883..f4e6cc9fc2 100644 --- a/locale/loadlocale.c +++ b/locale/loadlocale.c @@ -167,7 +167,7 @@ _nl_load_locale (struct loaded_l10nfile *file, int category) { int fd; void *filedata; - struct stat64 st; + struct __stat64_t64 st; struct __locale_data *newdata; int save_err; int alloc = ld_mapped; @@ -180,7 +180,7 @@ _nl_load_locale (struct loaded_l10nfile *file, int category) /* Cannot open the file. */ return; - if (__builtin_expect (__fstat64 (fd, &st), 0) < 0) + if (__glibc_unlikely (__fstat64_time64 (fd, &st) < 0)) { puntfd: __close_nocancel_nostatus (fd); @@ -206,7 +206,7 @@ _nl_load_locale (struct loaded_l10nfile *file, int category) if (__builtin_expect (fd, 0) < 0) return; - if (__builtin_expect (__fstat64 (fd, &st), 0) < 0) + if (__glibc_unlikely (__fstat64_time64 (fd, &st) < 0)) goto puntfd; } diff --git a/nscd/nscd_helper.c b/nscd/nscd_helper.c index 462504d828..06ba7476e5 100644 --- a/nscd/nscd_helper.c +++ b/nscd/nscd_helper.c @@ -324,8 +324,8 @@ __nscd_get_mapping (request_type type, const char *key, if (__glibc_unlikely (n == keylen)) { - struct stat64 st; - if (__builtin_expect (__fstat64 (mapfd, &st) != 0, 0) + struct __stat64_t64 st; + if (__glibc_unlikely (__fstat64_time64 (mapfd, &st) != 0) || __builtin_expect (st.st_size < sizeof (struct database_pers_head), 0)) goto out_close; diff --git a/nss/nss_database.c b/nss/nss_database.c index 1e11294406..ab121cb371 100644 --- a/nss/nss_database.c +++ b/nss/nss_database.c @@ -394,7 +394,7 @@ nss_database_check_reload_and_get (struct nss_database_state *local, nss_action_list *result, enum nss_database database_index) { - struct stat64 str; + struct __stat64_t64 str; /* Acquire MO is needed because the thread that sets reload_disabled may have loaded the configuration first, so synchronize with the @@ -424,7 +424,7 @@ nss_database_check_reload_and_get (struct nss_database_state *local, errors here are very unlikely, but the chance that we're entering a container is also very unlikely, so we err on the side of both very unlikely things not happening at the same time. */ - if (__stat64 ("/", &str) != 0 + if (__stat64_time64 ("/", &str) != 0 || (local->root_ino != 0 && (str.st_ino != local->root_ino || str.st_dev != local->root_dev))) diff --git a/sysdeps/posix/dl-fileid.h b/sysdeps/posix/dl-fileid.h index 0bb6794dbe..bf437f3d71 100644 --- a/sysdeps/posix/dl-fileid.h +++ b/sysdeps/posix/dl-fileid.h @@ -32,9 +32,9 @@ struct r_file_id static inline bool _dl_get_file_id (int fd, struct r_file_id *id) { - struct stat64 st; + struct __stat64_t64 st; - if (__glibc_unlikely (__fstat64 (fd, &st) < 0)) + if (__glibc_unlikely (__fstat64_time64 (fd, &st) < 0)) return false; id->dev = st.st_dev; diff --git a/sysdeps/posix/euidaccess.c b/sysdeps/posix/euidaccess.c index 26ebb432a2..86f3285471 100644 --- a/sysdeps/posix/euidaccess.c +++ b/sysdeps/posix/euidaccess.c @@ -119,7 +119,7 @@ int group_member (); int euidaccess (const char *path, int mode) { - struct stat64 stats; + struct __stat64_t64 stats; int granted; #ifdef _LIBC @@ -140,7 +140,7 @@ euidaccess (const char *path, int mode) return access (path, mode); #endif - if (__stat64 (path, &stats)) + if (__stat64_time64 (path, &stats)) return -1; mode &= (X_OK | W_OK | R_OK); /* Clear any bogus bits. */ diff --git a/sysdeps/posix/getaddrinfo.c b/sysdeps/posix/getaddrinfo.c index b7e1aee80f..f7f19f1d7c 100644 --- a/sysdeps/posix/getaddrinfo.c +++ b/sysdeps/posix/getaddrinfo.c @@ -1636,16 +1636,16 @@ static int gaiconf_reload_flag_ever_set; /* Last modification time. */ #ifdef _STATBUF_ST_NSEC -static struct timespec gaiconf_mtime; +static struct __timespec64 gaiconf_mtime; static inline void -save_gaiconf_mtime (const struct stat64 *st) +save_gaiconf_mtime (const struct __stat64_t64 *st) { gaiconf_mtime = st->st_mtim; } static inline bool -check_gaiconf_mtime (const struct stat64 *st) +check_gaiconf_mtime (const struct __stat64_t64 *st) { return (st->st_mtim.tv_sec == gaiconf_mtime.tv_sec && st->st_mtim.tv_nsec == gaiconf_mtime.tv_nsec); @@ -1656,13 +1656,13 @@ check_gaiconf_mtime (const struct stat64 *st) static time_t gaiconf_mtime; static inline void -save_gaiconf_mtime (const struct stat64 *st) +save_gaiconf_mtime (const struct __stat64_t64 *st) { gaiconf_mtime = st->st_mtime; } static inline bool -check_gaiconf_mtime (const struct stat64 *st) +check_gaiconf_mtime (const struct __stat64_t64 *st) { return st->st_mtime == gaiconf_mtime; } @@ -1777,8 +1777,8 @@ gaiconf_init (void) FILE *fp = fopen (GAICONF_FNAME, "rce"); if (fp != NULL) { - struct stat64 st; - if (__fstat64 (fileno (fp), &st) != 0) + struct __stat64_t64 st; + if (__fstat64_time64 (fileno (fp), &st) != 0) { fclose (fp); goto no_file; @@ -2130,8 +2130,8 @@ gaiconf_init (void) static void gaiconf_reload (void) { - struct stat64 st; - if (__stat64 (GAICONF_FNAME, &st) != 0 + struct __stat64_t64 st; + if (__stat64_time64 (GAICONF_FNAME, &st) != 0 || !check_gaiconf_mtime (&st)) gaiconf_init (); } diff --git a/sysdeps/posix/getcwd.c b/sysdeps/posix/getcwd.c index f11644aae7..13680026ff 100644 --- a/sysdeps/posix/getcwd.c +++ b/sysdeps/posix/getcwd.c @@ -183,7 +183,7 @@ __getcwd_generic (char *buf, size_t size) ino_t rootino, thisino; char *dir; register char *dirp; - struct stat64 st; + struct __stat64_t64 st; size_t allocated = size; size_t used; @@ -249,12 +249,12 @@ __getcwd_generic (char *buf, size_t size) dirp = dir + allocated; *--dirp = '\0'; - if (__lstat64 (".", &st) < 0) + if (__lstat64_time64 (".", &st) < 0) goto lose; thisdev = st.st_dev; thisino = st.st_ino; - if (__lstat64 ("/", &st) < 0) + if (__lstat64_time64 ("/", &st) < 0) goto lose; rootdev = st.st_dev; rootino = st.st_ino; @@ -276,12 +276,12 @@ __getcwd_generic (char *buf, size_t size) if (fd < 0) goto lose; fd_needs_closing = true; - parent_status = __fstat64 (fd, &st); + parent_status = __fstat64_time64 (fd, &st); #else dotlist[dotlen++] = '.'; dotlist[dotlen++] = '.'; dotlist[dotlen] = '\0'; - parent_status = __lstat64 (dotlist, &st); + parent_status = __lstat64_time64 (dotlist, &st); #endif if (parent_status != 0) goto lose; @@ -353,7 +353,8 @@ __getcwd_generic (char *buf, size_t size) { int entry_status; #if HAVE_OPENAT_SUPPORT - entry_status = __fstatat64 (fd, d->d_name, &st, AT_SYMLINK_NOFOLLOW); + entry_status = __fstatat64_time64 (fd, d->d_name, &st, + AT_SYMLINK_NOFOLLOW); #else /* Compute size needed for this file name, or for the file name ".." in the same directory, whichever is larger. @@ -390,7 +391,7 @@ __getcwd_generic (char *buf, size_t size) } memcpy (dotlist + dotlen, d->d_name, _D_ALLOC_NAMLEN (d)); - entry_status = __lstat64 (dotlist, &st); + entry_status = __lstat64_time64 (dotlist, &st); #endif /* We don't fail here if we cannot stat() a directory entry. This can happen when (network) file systems fail. If this diff --git a/sysdeps/posix/pathconf.c b/sysdeps/posix/pathconf.c index 9eb17d1536..89fcfad49d 100644 --- a/sysdeps/posix/pathconf.c +++ b/sysdeps/posix/pathconf.c @@ -129,9 +129,9 @@ __pathconf (const char *path, int name) #ifdef _POSIX_ASYNC_IO { /* AIO is only allowed on regular files and block devices. */ - struct stat64 st; + struct __stat64_t64 st; - if (__stat64 (path, &st) < 0 + if (__stat64_time64 (path, &st) < 0 || (! S_ISREG (st.st_mode) && ! S_ISBLK (st.st_mode))) return -1; else diff --git a/sysdeps/posix/sysconf.c b/sysdeps/posix/sysconf.c index 54d6046d00..3e8ec5cd51 100644 --- a/sysdeps/posix/sysconf.c +++ b/sysdeps/posix/sysconf.c @@ -1215,8 +1215,8 @@ __sysconf_check_spec (const char *spec) "/POSIX_V6_", sizeof ("/POSIX_V6_") - 1), spec, speclen + 1); - struct stat64 st; - long int ret = __stat64 (name, &st) >= 0 ? 1 : -1; + struct __stat64_t64 st; + long int ret = __stat64_time64 (name, &st) >= 0 ? 1 : -1; __set_errno (save_errno); return ret; diff --git a/sysdeps/posix/tempname.c b/sysdeps/posix/tempname.c index 5f804b38d7..ff38f1e31f 100644 --- a/sysdeps/posix/tempname.c +++ b/sysdeps/posix/tempname.c @@ -55,14 +55,14 @@ #include <time.h> #if _LIBC -# define struct_stat64 struct stat64 +# define struct_stat64 struct __stat64_t64 # define __secure_getenv __libc_secure_getenv #else # define struct_stat64 struct stat # define __gen_tempname gen_tempname # define __mkdir mkdir # define __open open -# define __lstat64(file, buf) lstat (file, buf) +# define __lstat64_time64(file, buf) lstat (file, buf) # define __stat64(file, buf) stat (file, buf) # define __getrandom getrandom # define __clock_gettime64 clock_gettime @@ -99,7 +99,7 @@ static int direxists (const char *dir) { struct_stat64 buf; - return __stat64 (dir, &buf) == 0 && S_ISDIR (buf.st_mode); + return __stat64_time64 (dir, &buf) == 0 && S_ISDIR (buf.st_mode); } /* Path search algorithm, for tmpnam, tmpfile, etc. If DIR is @@ -191,7 +191,7 @@ try_nocreate (char *tmpl, void *flags _GL_UNUSED) { struct_stat64 st; - if (__lstat64 (tmpl, &st) == 0 || errno == EOVERFLOW) + if (__lstat64_time64 (tmpl, &st) == 0 || errno == EOVERFLOW) __set_errno (EEXIST); return errno == ENOENT ? 0 : -1; } diff --git a/sysdeps/unix/sysv/linux/fdopendir.c b/sysdeps/unix/sysv/linux/fdopendir.c index ede43f2485..32ec10e206 100644 --- a/sysdeps/unix/sysv/linux/fdopendir.c +++ b/sysdeps/unix/sysv/linux/fdopendir.c @@ -27,9 +27,9 @@ DIR * __fdopendir (int fd) { - struct stat64 statbuf; + struct __stat64_t64 statbuf; - if (__builtin_expect (__fstat64 (fd, &statbuf), 0) < 0) + if (__glibc_unlikely (__fstat64_time64 (fd, &statbuf) < 0)) return NULL; if (__glibc_unlikely (! S_ISDIR (statbuf.st_mode))) { diff --git a/sysdeps/unix/sysv/linux/fexecve.c b/sysdeps/unix/sysv/linux/fexecve.c index df25c2acb8..4dfcaeedc1 100644 --- a/sysdeps/unix/sysv/linux/fexecve.c +++ b/sysdeps/unix/sysv/linux/fexecve.c @@ -58,8 +58,8 @@ fexecve (int fd, char *const argv[], char *const envp[]) /* We come here only if the 'execve' call fails. Determine whether /proc is mounted. If not we return ENOSYS. */ - struct stat64 st; - if (__stat64 ("/proc/self/fd", &st) != 0 && errno == ENOENT) + struct __stat64_t64 st; + if (__stat64_time64 ("/proc/self/fd", &st) != 0 && errno == ENOENT) save = ENOSYS; __set_errno (save); diff --git a/sysdeps/unix/sysv/linux/opendir.c b/sysdeps/unix/sysv/linux/opendir.c index 4020a826f9..48f254d169 100644 --- a/sysdeps/unix/sysv/linux/opendir.c +++ b/sysdeps/unix/sysv/linux/opendir.c @@ -49,8 +49,8 @@ opendir_tail (int fd) /* Now make sure this really is a directory and nothing changed since the `stat' call. The S_ISDIR check is superfluous if O_DIRECTORY works, but it's cheap and we need the stat call for st_blksize anyway. */ - struct stat64 statbuf; - if (__glibc_unlikely (__fstat64 (fd, &statbuf) < 0)) + struct __stat64_t64 statbuf; + if (__glibc_unlikely (__fstat64_time64 (fd, &statbuf) < 0)) goto lose; if (__glibc_unlikely (! S_ISDIR (statbuf.st_mode))) { @@ -88,7 +88,8 @@ __opendir (const char *name) weak_alias (__opendir, opendir) DIR * -__alloc_dir (int fd, bool close_fd, int flags, const struct stat64 *statp) +__alloc_dir (int fd, bool close_fd, int flags, + const struct __stat64_t64 *statp) { /* We have to set the close-on-exit flag if the user provided the file descriptor. */ diff --git a/sysdeps/unix/sysv/linux/pathconf.c b/sysdeps/unix/sysv/linux/pathconf.c index f37e8aaf68..b599a66c93 100644 --- a/sysdeps/unix/sysv/linux/pathconf.c +++ b/sysdeps/unix/sysv/linux/pathconf.c @@ -65,9 +65,10 @@ distinguish_extX (const struct statfs *fsbuf, const char *file, int fd) { char buf[64]; char path[PATH_MAX]; - struct stat64 st; + struct __stat64_t64 st; - if ((file == NULL ? __fstat64 (fd, &st) : __stat64 (file, &st)) != 0) + if ((file == NULL ? __fstat64_time64 (fd, &st) + : __stat64_time64 (file, &st)) != 0) /* Strange. The statfd call worked, but stat fails. Default to the more pessimistic value. */ return EXT2_LINK_MAX; diff --git a/sysdeps/unix/sysv/linux/ttyname.h b/sysdeps/unix/sysv/linux/ttyname.h index 0a0048cc02..5dcbfef973 100644 --- a/sysdeps/unix/sysv/linux/ttyname.h +++ b/sysdeps/unix/sysv/linux/ttyname.h @@ -25,24 +25,18 @@ linux/Documentation/devices.txt (on linux < 4.10) or linux/Documentation/admin-guide/devices.txt (on linux >= 4.10). */ static inline bool -is_pty (struct stat64 *sb) +is_pty (struct __stat64_t64 *sb) { -#ifdef _STATBUF_ST_RDEV int m = __gnu_dev_major (sb->st_rdev); return (136 <= m && m <= 143); -#else - return false; -#endif } static inline bool -is_mytty (const struct stat64 *mytty, const struct stat64 *maybe) +is_mytty (const struct __stat64_t64 *mytty, const struct __stat64_t64 *maybe) { return (maybe->st_ino == mytty->st_ino && maybe->st_dev == mytty->st_dev -#ifdef _STATBUF_ST_RDEV && S_ISCHR (maybe->st_mode) && maybe->st_rdev == mytty->st_rdev -#endif ); } diff --git a/sysdeps/unix/sysv/linux/ttyname_r.c b/sysdeps/unix/sysv/linux/ttyname_r.c index 899a851a83..9ef9f42883 100644 --- a/sysdeps/unix/sysv/linux/ttyname_r.c +++ b/sysdeps/unix/sysv/linux/ttyname_r.c @@ -31,15 +31,15 @@ #include "ttyname.h" static int getttyname_r (char *buf, size_t buflen, - const struct stat64 *mytty, int save, + const struct __stat64_t64 *mytty, int save, int *dostat); static int attribute_compat_text_section -getttyname_r (char *buf, size_t buflen, const struct stat64 *mytty, +getttyname_r (char *buf, size_t buflen, const struct __stat64_t64 *mytty, int save, int *dostat) { - struct stat64 st; + struct __stat64_t64 st; DIR *dirstream; struct dirent64 *d; size_t devlen = strlen (buf); @@ -71,7 +71,7 @@ getttyname_r (char *buf, size_t buflen, const struct stat64 *mytty, cp = __stpncpy (buf + devlen, d->d_name, needed); cp[0] = '\0'; - if (__stat64 (buf, &st) == 0 + if (__stat64_time64 (buf, &st) == 0 && is_mytty (mytty, &st)) { (void) __closedir (dirstream); @@ -93,7 +93,7 @@ int __ttyname_r (int fd, char *buf, size_t buflen) { struct fd_to_filename filename; - struct stat64 st, st1; + struct __stat64_t64 st, st1; int dostat = 0; int doispty = 0; int save = errno; @@ -118,7 +118,7 @@ __ttyname_r (int fd, char *buf, size_t buflen) if (__glibc_unlikely (__tcgetattr (fd, &term) < 0)) return errno; - if (__fstat64 (fd, &st) < 0) + if (__fstat64_time64 (fd, &st) < 0) return errno; /* We try using the /proc filesystem. */ @@ -144,7 +144,7 @@ __ttyname_r (int fd, char *buf, size_t buflen) /* Verify readlink result, fall back on iterating through devices. */ if (buf[0] == '/' - && __stat64 (buf, &st1) == 0 + && __stat64_time64 (buf, &st1) == 0 && is_mytty (&st, &st1)) return 0; @@ -155,7 +155,7 @@ __ttyname_r (int fd, char *buf, size_t buflen) memcpy (buf, "/dev/pts/", sizeof ("/dev/pts/")); buflen -= sizeof ("/dev/pts/") - 1; - if (__stat64 (buf, &st1) == 0 && S_ISDIR (st1.st_mode)) + if (__stat64_time64 (buf, &st1) == 0 && S_ISDIR (st1.st_mode)) { ret = getttyname_r (buf, buflen, &st, save, &dostat); diff --git a/sysvipc/ftok.c b/sysvipc/ftok.c index 2e39c74415..bd633bd395 100644 --- a/sysvipc/ftok.c +++ b/sysvipc/ftok.c @@ -22,10 +22,10 @@ key_t ftok (const char *pathname, int proj_id) { - struct stat64 st; + struct __stat64_t64 st; key_t key; - if (__stat64 (pathname, &st) < 0) + if (__stat64_time64 (pathname, &st) < 0) return (key_t) -1; key = ((st.st_ino & 0xffff) | ((st.st_dev & 0xff) << 16) diff --git a/time/tzfile.c b/time/tzfile.c index 040a5e341f..4377018a55 100644 --- a/time/tzfile.c +++ b/time/tzfile.c @@ -150,9 +150,9 @@ __tzfile_read (const char *file, size_t extra, char **extrap) } /* If we were already using tzfile, check whether the file changed. */ - struct stat64 st; + struct __stat64_t64 st; if (was_using_tzfile - && __stat64 (file, &st) == 0 + && __stat64_time64 (file, &st) == 0 && tzfile_ino == st.st_ino && tzfile_dev == st.st_dev && tzfile_mtime == st.st_mtime) goto done; /* Nothing to do. */ @@ -164,7 +164,7 @@ __tzfile_read (const char *file, size_t extra, char **extrap) goto ret_free_transitions; /* Get information about the file we are actually using. */ - if (__fstat64 (__fileno (f), &st) != 0) + if (__fstat64_time64 (__fileno (f), &st) != 0) goto lose; free ((void *) transitions); </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tx1/gnu-release-aarch64-spec2k6-O2 - Build # 14 - Fixed!

by ci_notify＠linaro.org

Successfully identified regression in *gcc* in CI configuration tcwg_bmk_gnu_tx1/gnu-release-aarch64-spec2k6-O2. So far, this commit has regressed CI configurations: - tcwg_bmk_gnu_tx1/gnu-release-aarch64-spec2k6-O2 Culprit: <cut> commit 9593e8e5e391e77bb065d4689b7511bed6a640a3 Author: Richard Biener <rguenther(a)suse.de> Date: Thu Sep 26 13:52:45 2019 +0000 tree-vect-loop.c (vect_analyze_loop_operations): Also call vectorizable_reduction for vect_double_reduction_def. 2019-09-26 Richard Biener <rguenther(a)suse.de> * tree-vect-loop.c (vect_analyze_loop_operations): Also call vectorizable_reduction for vect_double_reduction_def. (vect_transform_loop): Likewise. (vect_create_epilog_for_reduction): Move double-reduction PHI creation and preheader argument setting of PHIs ... (vectorizable_reduction): ... here. Also process vect_double_reduction_def PHIs, creating the vectorized PHI nodes, remembering the scalar adjustment computed for the epilogue in STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT. Remember the original reduction code in STMT_VINFO_REDUC_CODE. * tree-vectorizer.c (vec_info::new_stmt_vec_info): Initialize STMT_VINFO_REDUC_CODE. * tree-vectorizer.h (_stmt_vec_info::reduc_epilogue_adjustment): New. (_stmt_vec_info::reduc_code): Likewise. (STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT): Likewise. (STMT_VINFO_REDUC_CODE): Likewise. From-SVN: r276150 </cut> Results regressed to (for first_bad == 9593e8e5e391e77bb065d4689b7511bed6a640a3) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O2 -- artifacts/build-9593e8e5e391e77bb065d4689b7511bed6a640a3/results_id: 1 # 482.sphinx3,[.] subvq_mgau_shortlist regressed by 112 from (for last_good == 5fdd1d33524c94783eadbf9dcd2c900d65baee72) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O2 -- artifacts/build-5fdd1d33524c94783eadbf9dcd2c900d65baee72/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… Results ID of last_good: tx1_64/tcwg_bmk_gnu_tx1/bisect-gnu-release-aarch64-spec2k6-O2/1200 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… Results ID of first_bad: tx1_64/tcwg_bmk_gnu_tx1/bisect-gnu-release-aarch64-spec2k6-O2/1186 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… Configuration details: Reproduce builds: <cut> mkdir investigate-gcc-9593e8e5e391e77bb065d4689b7511bed6a640a3 cd investigate-gcc-9593e8e5e391e77bb065d4689b7511bed6a640a3 git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd gcc # Reproduce first_bad build git checkout --detach 9593e8e5e391e77bb065d4689b7511bed6a640a3 ../artifacts/test.sh # Reproduce last_good build git checkout --detach 5fdd1d33524c94783eadbf9dcd2c900d65baee72 ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-release-a… Full commit (up to 1000 lines): <cut> commit 9593e8e5e391e77bb065d4689b7511bed6a640a3 Author: Richard Biener <rguenther(a)suse.de> Date: Thu Sep 26 13:52:45 2019 +0000 tree-vect-loop.c (vect_analyze_loop_operations): Also call vectorizable_reduction for vect_double_reduction_def. 2019-09-26 Richard Biener <rguenther(a)suse.de> * tree-vect-loop.c (vect_analyze_loop_operations): Also call vectorizable_reduction for vect_double_reduction_def. (vect_transform_loop): Likewise. (vect_create_epilog_for_reduction): Move double-reduction PHI creation and preheader argument setting of PHIs ... (vectorizable_reduction): ... here. Also process vect_double_reduction_def PHIs, creating the vectorized PHI nodes, remembering the scalar adjustment computed for the epilogue in STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT. Remember the original reduction code in STMT_VINFO_REDUC_CODE. * tree-vectorizer.c (vec_info::new_stmt_vec_info): Initialize STMT_VINFO_REDUC_CODE. * tree-vectorizer.h (_stmt_vec_info::reduc_epilogue_adjustment): New. (_stmt_vec_info::reduc_code): Likewise. (STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT): Likewise. (STMT_VINFO_REDUC_CODE): Likewise. From-SVN: r276150 --- gcc/ChangeLog | 19 +++ gcc/tree-vect-loop.c | 319 +++++++++++++++++++++++++++----------------------- gcc/tree-vectorizer.c | 1 + gcc/tree-vectorizer.h | 8 ++ 4 files changed, 198 insertions(+), 149 deletions(-) diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 98fa0ffe56b..19892af07c0 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,22 @@ +2019-09-26 Richard Biener <rguenther(a)suse.de> + + * tree-vect-loop.c (vect_analyze_loop_operations): Also call + vectorizable_reduction for vect_double_reduction_def. + (vect_transform_loop): Likewise. + (vect_create_epilog_for_reduction): Move double-reduction + PHI creation and preheader argument setting of PHIs ... + (vectorizable_reduction): ... here. Also process + vect_double_reduction_def PHIs, creating the vectorized + PHI nodes, remembering the scalar adjustment computed for + the epilogue in STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT. + Remember the original reduction code in STMT_VINFO_REDUC_CODE. + * tree-vectorizer.c (vec_info::new_stmt_vec_info): + Initialize STMT_VINFO_REDUC_CODE. + * tree-vectorizer.h (_stmt_vec_info::reduc_epilogue_adjustment): New. + (_stmt_vec_info::reduc_code): Likewise. + (STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT): Likewise. + (STMT_VINFO_REDUC_CODE): Likewise. + 2019-09-26 Matt Turner <mattst88(a)gmail.com> PR driver/69471 diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c index ec00912e67d..1a561f9d16f 100644 --- a/gcc/tree-vect-loop.c +++ b/gcc/tree-vect-loop.c @@ -1548,6 +1548,8 @@ vect_analyze_loop_operations (loop_vec_info loop_vinfo) ok = vectorizable_induction (stmt_info, NULL, NULL, NULL, &cost_vec); else if ((STMT_VINFO_DEF_TYPE (stmt_info) == vect_reduction_def + || (STMT_VINFO_DEF_TYPE (stmt_info) + == vect_double_reduction_def) || STMT_VINFO_DEF_TYPE (stmt_info) == vect_nested_cycle) && ! PURE_SLP_STMT (stmt_info)) ok = vectorizable_reduction (stmt_info, NULL, NULL, NULL, NULL, @@ -4299,20 +4301,17 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, gimple *exit_phi; tree bitsize; tree adjustment_def = NULL; - tree vec_initial_def = NULL; tree expr, def, initial_def = NULL; tree orig_name, scalar_result; imm_use_iterator imm_iter, phi_imm_iter; use_operand_p use_p, phi_use_p; gimple *use_stmt; - stmt_vec_info reduction_phi_info = NULL; bool nested_in_vect_loop = false; auto_vec<gimple *> new_phis; auto_vec<stmt_vec_info> inner_phis; int j, i; auto_vec<tree> scalar_results; unsigned int group_size = 1, k, ratio; - auto_vec<tree> vec_initial_defs; auto_vec<gimple *> phis; bool slp_reduc = false; bool direct_slp_reduc; @@ -4336,10 +4335,10 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, mode = TYPE_MODE (vectype); /* 1. Create the reduction def-use cycle: - Set the arguments of REDUCTION_PHIS, i.e., transform + Set the backedge argument of REDUCTION_PHIS, i.e., transform loop: - vec_def = phi <null, null> # REDUCTION_PHI + vec_def = phi <vec_init, null> # REDUCTION_PHI VECT_DEF = vector_stmt # vectorized form of STMT ... @@ -4350,18 +4349,10 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, VECT_DEF = vector_stmt # vectorized form of STMT ... - (in case of SLP, do it for all the phis). */ + (in case of SLP, do it for all the phis). */ - /* Get the loop-entry arguments. */ if (slp_node) - { - unsigned vec_num = SLP_TREE_NUMBER_OF_VEC_STMTS (slp_node); - vec_initial_defs.reserve (vec_num); - get_initial_defs_for_reduction (slp_node_instance->reduc_phis, - &vec_initial_defs, vec_num, - REDUC_GROUP_FIRST_ELEMENT (stmt_info), - neutral_op); - } + ; else { /* Get at the scalar def before the loop, that defines the initial value @@ -4373,64 +4364,29 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, for REDUC_MIN and initial_def larger than the base. */ if (STMT_VINFO_VEC_REDUCTION_TYPE (stmt_info) == INTEGER_INDUC_COND_REDUCTION) - { - induc_val = STMT_VINFO_VEC_INDUC_COND_INITIAL_VAL (stmt_info); - if (TREE_CODE (initial_def) == INTEGER_CST - && (STMT_VINFO_VEC_REDUCTION_TYPE (stmt_info) - == INTEGER_INDUC_COND_REDUCTION) - && !integer_zerop (induc_val) - && ((code == MAX_EXPR - && tree_int_cst_lt (initial_def, induc_val)) - || (code == MIN_EXPR - && tree_int_cst_lt (induc_val, initial_def)))) - induc_val = initial_def; - vec_initial_def = build_vector_from_val (vectype, induc_val); - } + induc_val = STMT_VINFO_VEC_INDUC_COND_INITIAL_VAL (stmt_info); else if (double_reduc) - /* In case of double reduction we only create a vector variable - to be put in the reduction phi node. The actual statement - creation is done later in this function. */ - vec_initial_def = vect_create_destination_var (initial_def, vectype); + ; else if (nested_in_vect_loop) - { - /* Do not use an adjustment def as that case is not supported - correctly if ncopies is not one. */ - vec_initial_def = vect_get_vec_def_for_operand (initial_def, - stmt_info); - } + ; else - vec_initial_def - = get_initial_def_for_reduction (stmt_info, code, initial_def, - &adjustment_def); - vec_initial_defs.create (1); - vec_initial_defs.quick_push (vec_initial_def); + adjustment_def = STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT (stmt_info); } - /* Set phi nodes arguments. */ + /* Set phi nodes latch arguments. */ FOR_EACH_VEC_ELT (reduction_phis, i, phi_info) { - tree vec_init_def = vec_initial_defs[i]; tree def = vect_defs[i]; for (j = 0; j < ncopies; j++) { + /* Set the loop-latch arg for the reduction-phi. */ if (j != 0) { phi_info = STMT_VINFO_RELATED_STMT (phi_info); - if (nested_in_vect_loop) - vec_init_def - = vect_get_vec_def_for_stmt_copy (loop_vinfo, vec_init_def); + def = vect_get_vec_def_for_stmt_copy (loop_vinfo, def); } - /* Set the loop-entry arg of the reduction-phi. */ - gphi *phi = as_a <gphi *> (phi_info->stmt); - add_phi_arg (phi, vec_init_def, loop_preheader_edge (loop), - UNKNOWN_LOCATION); - - /* Set the loop-latch arg for the reduction-phi. */ - if (j > 0) - def = vect_get_vec_def_for_stmt_copy (loop_vinfo, def); - add_phi_arg (phi, def, loop_latch_edge (loop), UNKNOWN_LOCATION); if (dump_enabled_p ()) @@ -4934,7 +4890,7 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, if ((STMT_VINFO_VEC_REDUCTION_TYPE (stmt_info) == INTEGER_INDUC_COND_REDUCTION) - && !operand_equal_p (initial_def, induc_val, 0)) + && induc_val) { /* Earlier we set the initial value to be a vector if induc_val values. Check the result and if it is induc_val then replace @@ -5297,7 +5253,7 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, if ((STMT_VINFO_VEC_REDUCTION_TYPE (stmt_info) == INTEGER_INDUC_COND_REDUCTION) - && !operand_equal_p (initial_def, induc_val, 0)) + && induc_val) { /* Earlier we set the initial value to be a vector if induc_val values. Check the result and if it is induc_val then replace @@ -5420,7 +5376,6 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, if (k % ratio == 0) { epilog_stmt_info = loop_vinfo->lookup_stmt (new_phis[k / ratio]); - reduction_phi_info = reduction_phis[k / ratio]; if (double_reduc) inner_phi = inner_phis[k / ratio]; } @@ -5435,6 +5390,8 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, scalar_dest = gimple_assign_lhs (scalar_stmt_info->stmt); } + if (outer_loop) + { phis.create (3); /* Find the loop-closed-use at the loop exit of the original scalar result. (The reduction result is expected to have two immediate uses - @@ -5449,8 +5406,6 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, FOR_EACH_VEC_ELT (phis, i, exit_phi) { - if (outer_loop) - { stmt_vec_info exit_phi_vinfo = loop_vinfo->lookup_stmt (exit_phi); gphi *vect_phi; @@ -5482,7 +5437,6 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, FOR_EACH_IMM_USE_STMT (use_stmt, imm_iter, orig_name) { stmt_vec_info use_stmt_vinfo; - tree vect_phi_init, preheader_arg, vect_phi_res; basic_block bb = gimple_bb (use_stmt); /* Check that USE_STMT is really double reduction phi @@ -5497,53 +5451,23 @@ vect_create_epilog_for_reduction (vec<tree> vect_defs, != vect_double_reduction_def) continue; - /* Create vector phi node for double reduction: - vs1 = phi <vs0, vs2> - vs1 was created previously in this function by a call to - vect_get_vec_def_for_operand and is stored in - vec_initial_def; - vs2 is defined by INNER_PHI, the vectorized EXIT_PHI; - vs0 is created here. */ - - /* Create vector phi node. */ - vect_phi = create_phi_node (vec_initial_def, bb); - loop_vec_info_for_loop (outer_loop)->add_stmt (vect_phi); - - /* Create vs0 - initial def of the double reduction phi. */ - preheader_arg = PHI_ARG_DEF_FROM_EDGE (use_stmt, - loop_preheader_edge (outer_loop)); - vect_phi_init = get_initial_def_for_reduction - (stmt_info, code, preheader_arg, NULL); - - /* Update phi node arguments with vs0 and vs2. */ - add_phi_arg (vect_phi, vect_phi_init, - loop_preheader_edge (outer_loop), - UNKNOWN_LOCATION); + /* Set the outer loop vector phi args. The PHI node + itself was created in vectorizable_reduction. */ + vect_phi = as_a <gphi *> + (STMT_VINFO_VEC_STMT (use_stmt_vinfo)->stmt); + + /* Update phi node argument with vs2. */ add_phi_arg (vect_phi, PHI_RESULT (inner_phi->stmt), loop_latch_edge (outer_loop), UNKNOWN_LOCATION); if (dump_enabled_p ()) dump_printf_loc (MSG_NOTE, vect_location, "created double reduction phi node: %G", vect_phi); - - vect_phi_res = PHI_RESULT (vect_phi); - - /* Replace the use, i.e., set the correct vs1 in the regular - reduction phi node. FORNOW, NCOPIES is always 1, so the - loop is redundant. */ - stmt_vec_info use_info = reduction_phi_info; - for (j = 0; j < ncopies; j++) - { - edge pr_edge = loop_preheader_edge (loop); - SET_PHI_ARG_DEF (as_a <gphi *> (use_info->stmt), - pr_edge->dest_idx, vect_phi_res); - use_info = STMT_VINFO_RELATED_STMT (use_info); - } } - } } - phis.release (); + } + if (nested_in_vect_loop) { if (double_reduc) @@ -6027,6 +5951,7 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, /* Make sure it was already recognized as a reduction computation. */ if (STMT_VINFO_DEF_TYPE (stmt_info) != vect_reduction_def + && STMT_VINFO_DEF_TYPE (stmt_info) != vect_double_reduction_def && STMT_VINFO_DEF_TYPE (stmt_info) != vect_nested_cycle) return false; @@ -6067,25 +5992,24 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, /* Leave the scalar phi in place. */ return true; - gassign *reduc_stmt = as_a <gassign *> (reduc_stmt_info->stmt); - code = gimple_assign_rhs_code (reduc_stmt); - for (unsigned k = 1; k < gimple_num_ops (reduc_stmt); ++k) - { - tree op = gimple_op (reduc_stmt, k); - if (op == phi_result) - continue; - if (k == 1 && code == COND_EXPR) - continue; - bool is_simple_use = vect_is_simple_use (op, loop_vinfo, &dt); - gcc_assert (is_simple_use); - if (dt == vect_constant_def || dt == vect_external_def) - continue; - if (!vectype_in - || (GET_MODE_SIZE (SCALAR_TYPE_MODE (TREE_TYPE (vectype_in))) - < GET_MODE_SIZE (SCALAR_TYPE_MODE (TREE_TYPE (op))))) - vectype_in = get_vectype_for_scalar_type (TREE_TYPE (op)); - break; - } + if (gassign *reduc_stmt = dyn_cast <gassign *> (reduc_stmt_info->stmt)) + for (unsigned k = 1; k < gimple_num_ops (reduc_stmt); ++k) + { + tree op = gimple_op (reduc_stmt, k); + if (op == phi_result) + continue; + if (k == 1 && gimple_assign_rhs_code (reduc_stmt) == COND_EXPR) + continue; + bool is_simple_use = vect_is_simple_use (op, loop_vinfo, &dt); + gcc_assert (is_simple_use); + if (dt == vect_constant_def || dt == vect_external_def) + continue; + if (!vectype_in + || (GET_MODE_SIZE (SCALAR_TYPE_MODE (TREE_TYPE (vectype_in))) + < GET_MODE_SIZE (SCALAR_TYPE_MODE (TREE_TYPE (op))))) + vectype_in = get_vectype_for_scalar_type (TREE_TYPE (op)); + break; + } /* For a nested cycle we might end up with an operation like phi_result * phi_result. */ if (!vectype_in) @@ -6093,10 +6017,21 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, gcc_assert (vectype_in); if (slp_node) - ncopies = 1; + { + /* The size vect_schedule_slp_instance computes is off for us. */ + vec_num = vect_get_num_vectors + (LOOP_VINFO_VECT_FACTOR (loop_vinfo) + * SLP_TREE_SCALAR_STMTS (slp_node).length (), vectype_in); + ncopies = 1; + } else - ncopies = vect_get_num_copies (loop_vinfo, vectype_in); + { + vec_num = 1; + ncopies = vect_get_num_copies (loop_vinfo, vectype_in); + } + /* Check whether we can use a single PHI node and accumulate + vectors to one before the backedge. */ stmt_vec_info use_stmt_info; if (ncopies > 1 && STMT_VINFO_RELEVANT (reduc_stmt_info) <= vect_used_only_live @@ -6104,45 +6039,129 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, && (!STMT_VINFO_IN_PATTERN_P (use_stmt_info) || !STMT_VINFO_PATTERN_DEF_SEQ (use_stmt_info)) && vect_stmt_to_vectorize (use_stmt_info) == reduc_stmt_info) - single_defuse_cycle = true; + { + single_defuse_cycle = true; + ncopies = 1; + } /* Create the destination vector */ - scalar_dest = gimple_assign_lhs (reduc_stmt); - vec_dest = vect_create_destination_var (scalar_dest, vectype_out); + vec_dest = vect_create_destination_var (phi_result, vectype_out); + /* Get the loop-entry arguments. */ + tree vec_initial_def; + auto_vec<tree> vec_initial_defs; if (slp_node) - /* The size vect_schedule_slp_instance computes is off for us. */ - vec_num = vect_get_num_vectors - (LOOP_VINFO_VECT_FACTOR (loop_vinfo) - * SLP_TREE_SCALAR_STMTS (slp_node).length (), - vectype_in); + { + vec_initial_defs.reserve (vec_num); + gcc_assert (slp_node == slp_node_instance->reduc_phis); + stmt_vec_info first = REDUC_GROUP_FIRST_ELEMENT (reduc_stmt_info); + tree neutral_op + = neutral_op_for_slp_reduction (slp_node, + STMT_VINFO_REDUC_CODE + (first ? first : reduc_stmt_info), + first != NULL); + get_initial_defs_for_reduction (slp_node_instance->reduc_phis, + &vec_initial_defs, vec_num, + first != NULL, neutral_op); + } else - vec_num = 1; + { + /* Get at the scalar def before the loop, that defines the initial + value of the reduction variable. */ + tree initial_def = PHI_ARG_DEF_FROM_EDGE (phi, + loop_preheader_edge (loop)); + /* Optimize: if initial_def is for REDUC_MAX smaller than the base + and we can't use zero for induc_val, use initial_def. Similarly + for REDUC_MIN and initial_def larger than the base. */ + if (STMT_VINFO_VEC_REDUCTION_TYPE (reduc_stmt_info) + == INTEGER_INDUC_COND_REDUCTION) + { + tree induc_val + = STMT_VINFO_VEC_INDUC_COND_INITIAL_VAL (reduc_stmt_info); + if (TREE_CODE (initial_def) == INTEGER_CST + && (STMT_VINFO_VEC_REDUCTION_TYPE (reduc_stmt_info) + == INTEGER_INDUC_COND_REDUCTION) + && !integer_zerop (induc_val) + && (((STMT_VINFO_VEC_COND_REDUC_CODE (reduc_stmt_info) + == MAX_EXPR) + && tree_int_cst_lt (initial_def, induc_val)) + || ((STMT_VINFO_VEC_COND_REDUC_CODE (reduc_stmt_info) + == MIN_EXPR) + && tree_int_cst_lt (induc_val, initial_def)))) + { + induc_val = initial_def; + /* Communicate we used the initial_def to epilouge + generation. */ + STMT_VINFO_VEC_INDUC_COND_INITIAL_VAL (reduc_stmt_info) + = NULL_TREE; + } + vec_initial_def = build_vector_from_val (vectype_out, induc_val); + } + else if (nested_cycle) + { + /* Do not use an adjustment def as that case is not supported + correctly if ncopies is not one. */ + vec_initial_def = vect_get_vec_def_for_operand (initial_def, + reduc_stmt_info); + } + else + { + tree adjustment_def = NULL_TREE; + tree *adjustment_defp = &adjustment_def; + enum tree_code code = STMT_VINFO_REDUC_CODE (reduc_stmt_info); + /* ??? For the outer loop PHI we have to do a bit of searching + to find the stmt with the code. reduc_stmt_info here is the + loop-closed PHI of the inner reduction which means we can look + at its single-arg def. */ + if (STMT_VINFO_DEF_TYPE (stmt_info) == vect_double_reduction_def) + { + tree def = gimple_phi_arg_def + (as_a <gphi *> (reduc_stmt_info->stmt), 0); + code = STMT_VINFO_REDUC_CODE + (vect_stmt_to_vectorize (loop_vinfo->lookup_def (def))); + adjustment_defp = NULL; + } + vec_initial_def + = get_initial_def_for_reduction (reduc_stmt_info, code, + initial_def, adjustment_defp); + STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT (reduc_stmt_info) + = adjustment_def; + } + vec_initial_defs.create (1); + vec_initial_defs.quick_push (vec_initial_def); + } /* Generate the reduction PHIs upfront. */ prev_phi_info = NULL; - for (j = 0; j < ncopies; j++) + for (i = 0; i < vec_num; i++) { - if (j == 0 || !single_defuse_cycle) + tree vec_init_def = vec_initial_defs[i]; + for (j = 0; j < ncopies; j++) { - for (i = 0; i < vec_num; i++) + /* Create the reduction-phi that defines the reduction + operand. */ + gphi *new_phi = create_phi_node (vec_dest, loop->header); + stmt_vec_info new_phi_info = loop_vinfo->add_stmt (new_phi); + + /* Set the loop-entry arg of the reduction-phi. */ + if (j != 0 && nested_cycle) + vec_init_def = vect_get_vec_def_for_stmt_copy (loop_vinfo, + vec_init_def); + add_phi_arg (new_phi, vec_init_def, loop_preheader_edge (loop), + UNKNOWN_LOCATION); + + /* The loop-latch arg is set in epilogue processing. */ + + if (slp_node) + SLP_TREE_VEC_STMTS (slp_node).quick_push (new_phi_info); + else { - /* Create the reduction-phi that defines the reduction - operand. */ - gimple *new_phi = create_phi_node (vec_dest, loop->header); - stmt_vec_info new_phi_info = loop_vinfo->add_stmt (new_phi); - - if (slp_node) - SLP_TREE_VEC_STMTS (slp_node).quick_push (new_phi_info); + if (j == 0) + STMT_VINFO_VEC_STMT (stmt_info) + = *vec_stmt = new_phi_info; else - { - if (j == 0) - STMT_VINFO_VEC_STMT (stmt_info) - = *vec_stmt = new_phi_info; - else - STMT_VINFO_RELATED_STMT (prev_phi_info) = new_phi_info; - prev_phi_info = new_phi_info; - } + STMT_VINFO_RELATED_STMT (prev_phi_info) = new_phi_info; + prev_phi_info = new_phi_info; } } } @@ -6633,6 +6652,7 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, gcc_assert (orig_code == MAX_EXPR || orig_code == MIN_EXPR); } } + STMT_VINFO_REDUC_CODE (stmt_info) = orig_code; reduc_fn = IFN_LAST; @@ -8378,6 +8398,7 @@ vect_transform_loop (loop_vec_info loop_vinfo) if ((STMT_VINFO_DEF_TYPE (stmt_info) == vect_induction_def || STMT_VINFO_DEF_TYPE (stmt_info) == vect_reduction_def + || STMT_VINFO_DEF_TYPE (stmt_info) == vect_double_reduction_def || STMT_VINFO_DEF_TYPE (stmt_info) == vect_nested_cycle) && ! PURE_SLP_STMT (stmt_info)) { diff --git a/gcc/tree-vectorizer.c b/gcc/tree-vectorizer.c index 8fb70b751da..50d508db7cd 100644 --- a/gcc/tree-vectorizer.c +++ b/gcc/tree-vectorizer.c @@ -639,6 +639,7 @@ vec_info::new_stmt_vec_info (gimple *stmt) STMT_VINFO_VECTORIZABLE (res) = true; STMT_VINFO_VEC_REDUCTION_TYPE (res) = TREE_CODE_REDUCTION; STMT_VINFO_VEC_COND_REDUC_CODE (res) = ERROR_MARK; + STMT_VINFO_REDUC_CODE (res) = ERROR_MARK; STMT_VINFO_REDUC_IDX (res) = -1; STMT_VINFO_SLP_VECT_ONLY (res) = false; diff --git a/gcc/tree-vectorizer.h b/gcc/tree-vectorizer.h index 370ce132f74..f140405bbd6 100644 --- a/gcc/tree-vectorizer.h +++ b/gcc/tree-vectorizer.h @@ -941,10 +941,16 @@ public: /* For INTEGER_INDUC_COND_REDUCTION, the initial value to be used. */ tree induc_cond_initial_val; + /* If not NULL the value to be added to compute final reduction value. */ + tree reduc_epilogue_adjustment; + /* On a reduction PHI the reduction type as detected by vect_force_simple_reduction. */ enum vect_reduction_type reduc_type; + /* The original reduction code, to be used in the epilogue. */ + enum tree_code reduc_code; + /* On a stmt participating in the reduction the index of the operand on the reduction SSA cycle. */ int reduc_idx; @@ -1039,6 +1045,7 @@ STMT_VINFO_BB_VINFO (stmt_vec_info stmt_vinfo) #define STMT_VINFO_VEC_REDUCTION_TYPE(S) (S)->v_reduc_type #define STMT_VINFO_VEC_COND_REDUC_CODE(S) (S)->cond_reduc_code #define STMT_VINFO_VEC_INDUC_COND_INITIAL_VAL(S) (S)->induc_cond_initial_val +#define STMT_VINFO_REDUC_EPILOGUE_ADJUSTMENT(S) (S)->reduc_epilogue_adjustment #define STMT_VINFO_REDUC_IDX(S) (S)->reduc_idx #define STMT_VINFO_DR_WRT_VEC_LOOP(S) (S)->dr_wrt_vec_loop @@ -1070,6 +1077,7 @@ STMT_VINFO_BB_VINFO (stmt_vec_info stmt_vinfo) #define STMT_VINFO_MIN_NEG_DIST(S) (S)->min_neg_dist #define STMT_VINFO_NUM_SLP_USES(S) (S)->num_slp_uses #define STMT_VINFO_REDUC_TYPE(S) (S)->reduc_type +#define STMT_VINFO_REDUC_CODE(S) (S)->reduc_code #define STMT_VINFO_REDUC_DEF(S) (S)->reduc_def #define STMT_VINFO_SLP_VECT_ONLY(S) (S)->slp_vect_only_p </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tk1/gnu-release-arm-spec2k6-O3 - Build # 25 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *binutils* in CI configuration tcwg_bmk_gnu_tk1/gnu-release-arm-spec2k6-O3. So far, this commit has regressed CI configurations: - tcwg_bmk_gnu_tk1/gnu-release-arm-spec2k6-O3 Culprit: <cut> commit 3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24 Author: GDB Administrator <gdbadmin(a)sourceware.org> Date: Wed Apr 21 00:00:45 2021 +0000 Automatic date update in version.in </cut> Results regressed to (for first_bad == 3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_marm -- artifacts/build-3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24/results_id: 1 # 456.hmmer,hmmer_base.default regressed by 105 from (for last_good == 54254f3896521d79edd4b6b9d5848a365cc483f2) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--with-mode=arm --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_marm -- artifacts/build-54254f3896521d79edd4b6b9d5848a365cc483f2/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… Results ID of last_good: tk1_32/tcwg_bmk_gnu_tk1/bisect-gnu-release-arm-spec2k6-O3/1141 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… Results ID of first_bad: tk1_32/tcwg_bmk_gnu_tk1/bisect-gnu-release-arm-spec2k6-O3/1128 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… Configuration details: Reproduce builds: <cut> mkdir investigate-binutils-3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24 cd investigate-binutils-3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24 git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd binutils # Reproduce first_bad build git checkout --detach 3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24 ../artifacts/test.sh # Reproduce last_good build git checkout --detach 54254f3896521d79edd4b6b9d5848a365cc483f2 ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tk1-gnu-release-a… Full commit (up to 1000 lines): <cut> commit 3f07ddd6899e382b28b14bc3c2d4cb0b52a61a24 Author: GDB Administrator <gdbadmin(a)sourceware.org> Date: Wed Apr 21 00:00:45 2021 +0000 Automatic date update in version.in --- bfd/version.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/bfd/version.h b/bfd/version.h index cc83f5bbda2..1f17c744c22 100644 --- a/bfd/version.h +++ b/bfd/version.h @@ -16,7 +16,7 @@ In releases, the date is not included in either version strings or sonames. */ -#define BFD_VERSION_DATE 20210420 +#define BFD_VERSION_DATE 20210421 #define BFD_VERSION @bfd_version@ #define BFD_VERSION_STRING @bfd_version_package@ @bfd_version_string@ #define REPORT_BUGS_TO @report_bugs_to@ </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tx1/gnu-master-aarch64-spec2k6-O3_LTO - Build # 16 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *gcc* in CI configuration tcwg_bmk_gnu_tx1/gnu-master-aarch64-spec2k6-O3_LTO. So far, this commit has regressed CI configurations: - tcwg_bmk_gnu_tx1/gnu-master-aarch64-spec2k6-O3_LTO Culprit: <cut> commit 39ed6a88c74db5412652f0167d03f71eca4f02fc Author: Ian Lance Taylor <iant(a)golang.org> Date: Tue May 18 18:28:17 2021 -0700 libgo: use Windows line endings in testwinsignal </cut> Results regressed to (for first_bad == 39ed6a88c74db5412652f0167d03f71eca4f02fc) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_LTO -- artifacts/build-39ed6a88c74db5412652f0167d03f71eca4f02fc/results_id: 1 # 453.povray,povray_base.default regressed by 103 from (for last_good == 01b2864757540d24c4e717a77b40b29369c064b2) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # true: 0 # benchmark -O3_LTO -- artifacts/build-01b2864757540d24c4e717a77b40b29369c064b2/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… Results ID of last_good: tx1_64/tcwg_bmk_gnu_tx1/bisect-gnu-master-aarch64-spec2k6-O3_LTO/1130 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… Results ID of first_bad: tx1_64/tcwg_bmk_gnu_tx1/bisect-gnu-master-aarch64-spec2k6-O3_LTO/1096 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… Configuration details: Reproduce builds: <cut> mkdir investigate-gcc-39ed6a88c74db5412652f0167d03f71eca4f02fc cd investigate-gcc-39ed6a88c74db5412652f0167d03f71eca4f02fc git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd gcc # Reproduce first_bad build git checkout --detach 39ed6a88c74db5412652f0167d03f71eca4f02fc ../artifacts/test.sh # Reproduce last_good build git checkout --detach 01b2864757540d24c4e717a77b40b29369c064b2 ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_gnu-bisect-tcwg_bmk_tx1-gnu-master-aa… Full commit (up to 1000 lines): <cut> commit 39ed6a88c74db5412652f0167d03f71eca4f02fc Author: Ian Lance Taylor <iant(a)golang.org> Date: Tue May 18 18:28:17 2021 -0700 libgo: use Windows line endings in testwinsignal --- libgo/go/runtime/testdata/testwinsignal/main.go | 38 ++++++++++++------------- 1 file changed, 19 insertions(+), 19 deletions(-) diff --git a/libgo/go/runtime/testdata/testwinsignal/main.go b/libgo/go/runtime/testdata/testwinsignal/main.go index 1e7c9475fd6..d8cd884ffac 100644 --- a/libgo/go/runtime/testdata/testwinsignal/main.go +++ b/libgo/go/runtime/testdata/testwinsignal/main.go @@ -1,19 +1,19 @@ -package main - -import ( - "fmt" - "os" - "os/signal" - "time" -) - -func main() { - c := make(chan os.Signal, 1) - signal.Notify(c) - - fmt.Println("ready") - sig := <-c - - time.Sleep(time.Second) - fmt.Println(sig) -} +package main + +import ( + "fmt" + "os" + "os/signal" + "time" +) + +func main() { + c := make(chan os.Signal, 1) + signal.Notify(c) + + fmt.Println("ready") + sig := <-c + + time.Sleep(time.Second) + fmt.Println(sig) +} </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tx1/llvm-release-aarch64-spec2k6-O2 - Build # 9 - Fixed!

by ci_notify＠linaro.org

Successfully identified regression in *llvm* in CI configuration tcwg_bmk_llvm_tx1/llvm-release-aarch64-spec2k6-O2. So far, this commit has regressed CI configurations: - tcwg_bmk_llvm_tx1/llvm-release-aarch64-spec2k6-O2 Culprit: <cut> commit e00f189d392dd9bf95f6a98f05f2d341d06cd65c Author: Roman Lebedev <lebedev.ri(a)gmail.com> Date: Mon Oct 5 22:35:59 2020 +0300 [InstCombine] Revert rL226781 "Teach InstCombine to canonicalize loads which are only ever stored to always use a legal integer type if one is available." (PR47592) (it was introduced in https://lists.llvm.org/pipermail/llvm-dev/2015-January/080956.html) This canonicalization seems dubious. Most importantly, while it does not create `inttoptr` casts by itself, it may cause them to appear later, see e.g. D88788. I think it's pretty obvious that it is an undesirable outcome, by now we've established that seemingly no-op `inttoptr`/`ptrtoint` casts are not no-op, and are no longer eager to look past them. Which e.g. means that given ``` %a = load i32 %b = inttoptr %a %c = inttoptr %a ``` we likely won't be able to tell that `%b` and `%c` is the same thing. As we can see in D88789 / D88788 / D88806 / D75505, we can't really teach SCEV about this (not without the https://bugs.llvm.org/show_bug.cgi?id=47592 at least) And we can't recover the situation post-inlining in instcombine. So it really does look like this fold is actively breaking otherwise-good IR, in a way that is not recoverable. And that means, this fold isn't helpful in exposing the passes that are otherwise unaware of these patterns it produces. Thusly, i propose to simply not perform such a canonicalization. The original motivational RFC does not state what larger problem that canonicalization was trying to solve, so i'm not sure how this plays out in the larger picture. On vanilla llvm test-suite + RawSpeed, this results in increase of asm instructions and final object size by ~+0.05% decreases final count of bitcasts by -4.79% (-28990), ptrtoint casts by -15.41% (-3423), and of inttoptr casts by -25.59% (-6919, *sic*). Overall, there's -0.04% less IR blocks, -0.39% instructions. See https://bugs.llvm.org/show_bug.cgi?id=47592 Differential Revision: https://reviews.llvm.org/D88789 </cut> Results regressed to (for first_bad == e00f189d392dd9bf95f6a98f05f2d341d06cd65c) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # build_llvm true: -3 # true: 0 # benchmark -O2 -- artifacts/build-e00f189d392dd9bf95f6a98f05f2d341d06cd65c/results_id: 1 # 470.lbm,lbm_base.default regressed by 111 # 470.lbm,[.] LBM_performStreamCollide regressed by 111 from (for last_good == 567462b48eba1c2d286ce97117994463f4535d2e) # reset_artifacts: -10 # build_abe binutils: -9 # build_abe stage1 -- --set gcc_override_configure=--disable-libsanitizer: -8 # build_abe linux: -7 # build_abe glibc: -6 # build_abe stage2 -- --set gcc_override_configure=--disable-libsanitizer: -5 # build_llvm true: -3 # true: 0 # benchmark -O2 -- artifacts/build-567462b48eba1c2d286ce97117994463f4535d2e/results_id: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… Results ID of last_good: tx1_64/tcwg_bmk_llvm_tx1/bisect-llvm-release-aarch64-spec2k6-O2/1124 Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… Results ID of first_bad: tx1_64/tcwg_bmk_llvm_tx1/bisect-llvm-release-aarch64-spec2k6-O2/1119 Build top page/logs: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… Configuration details: Reproduce builds: <cut> mkdir investigate-llvm-e00f189d392dd9bf95f6a98f05f2d341d06cd65c cd investigate-llvm-e00f189d392dd9bf95f6a98f05f2d341d06cd65c git clone https://git.linaro.org/toolchain/jenkins-scripts mkdir -p artifacts/manifests curl -o artifacts/manifests/build-baseline.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… --fail chmod +x artifacts/test.sh # Reproduce the baseline build (build all pre-requisites) ./jenkins-scripts/tcwg_bmk-build.sh @@ artifacts/manifests/build-baseline.sh cd llvm # Reproduce first_bad build git checkout --detach e00f189d392dd9bf95f6a98f05f2d341d06cd65c ../artifacts/test.sh # Reproduce last_good build git checkout --detach 567462b48eba1c2d286ce97117994463f4535d2e ../artifacts/test.sh cd .. </cut> History of pending regressions and results: https://git.linaro.org/toolchain/ci/base-artifacts.git/log/?h=linaro-local/… Artifacts: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… Build log: https://ci.linaro.org/job/tcwg_bmk_ci_llvm-bisect-tcwg_bmk_tx1-llvm-release… Full commit (up to 1000 lines): <cut> commit e00f189d392dd9bf95f6a98f05f2d341d06cd65c Author: Roman Lebedev <lebedev.ri(a)gmail.com> Date: Mon Oct 5 22:35:59 2020 +0300 [InstCombine] Revert rL226781 "Teach InstCombine to canonicalize loads which are only ever stored to always use a legal integer type if one is available." (PR47592) (it was introduced in https://lists.llvm.org/pipermail/llvm-dev/2015-January/080956.html) This canonicalization seems dubious. Most importantly, while it does not create `inttoptr` casts by itself, it may cause them to appear later, see e.g. D88788. I think it's pretty obvious that it is an undesirable outcome, by now we've established that seemingly no-op `inttoptr`/`ptrtoint` casts are not no-op, and are no longer eager to look past them. Which e.g. means that given ``` %a = load i32 %b = inttoptr %a %c = inttoptr %a ``` we likely won't be able to tell that `%b` and `%c` is the same thing. As we can see in D88789 / D88788 / D88806 / D75505, we can't really teach SCEV about this (not without the https://bugs.llvm.org/show_bug.cgi?id=47592 at least) And we can't recover the situation post-inlining in instcombine. So it really does look like this fold is actively breaking otherwise-good IR, in a way that is not recoverable. And that means, this fold isn't helpful in exposing the passes that are otherwise unaware of these patterns it produces. Thusly, i propose to simply not perform such a canonicalization. The original motivational RFC does not state what larger problem that canonicalization was trying to solve, so i'm not sure how this plays out in the larger picture. On vanilla llvm test-suite + RawSpeed, this results in increase of asm instructions and final object size by ~+0.05% decreases final count of bitcasts by -4.79% (-28990), ptrtoint casts by -15.41% (-3423), and of inttoptr casts by -25.59% (-6919, *sic*). Overall, there's -0.04% less IR blocks, -0.39% instructions. See https://bugs.llvm.org/show_bug.cgi?id=47592 Differential Revision: https://reviews.llvm.org/D88789 --- .../CodeGen/attr-arm-sve-vector-bits-bitcast.c | 18 +++-- clang/test/CodeGen/attr-arm-sve-vector-bits-call.c | 78 ++++++++++------------ clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c | 24 +++---- .../CodeGen/attr-arm-sve-vector-bits-globals.c | 6 +- .../InstCombine/InstCombineLoadStoreAlloca.cpp | 34 ---------- llvm/test/Transforms/InstCombine/atomic.ll | 18 ++--- llvm/test/Transforms/InstCombine/load.ll | 44 ++++++------ .../Transforms/InstCombine/loadstore-metadata.ll | 19 +----- .../InstCombine/non-integral-pointers.ll | 16 ++--- .../PhaseOrdering/instcombine-sroa-inttoptr.ll | 32 ++++----- 10 files changed, 107 insertions(+), 182 deletions(-) diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c index 3a5628d7f57e..5df1e83b13bf 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c @@ -255,22 +255,20 @@ svbool_t read_bool(struct struct_bool *s) { // CHECK-256-NEXT: entry: // CHECK-256-NEXT: [[X_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-256-NEXT: store <vscale x 16 x i1> [[X:%.*]], <vscale x 16 x i1>* [[X_ADDR]], align 16, [[TBAA15:!tbaa !.*]] -// CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[X_ADDR]] to i32* -// CHECK-256-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 16, [[TBAA6]] -// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], %struct.struct_bool* [[S:%.*]], i64 0, i32 1 -// CHECK-256-NEXT: [[TMP2:%.*]] = bitcast [3 x <4 x i8>]* [[Y]] to i32* -// CHECK-256-NEXT: store i32 [[TMP1]], i32* [[TMP2]], align 2, [[TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[X_ADDR]] to <4 x i8>* +// CHECK-256-NEXT: [[TMP1:%.*]] = load <4 x i8>, <4 x i8>* [[TMP0]], align 16, [[TBAA6]] +// CHECK-256-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], %struct.struct_bool* [[S:%.*]], i64 0, i32 1, i64 0 +// CHECK-256-NEXT: store <4 x i8> [[TMP1]], <4 x i8>* [[ARRAYIDX]], align 2, [[TBAA6]] // CHECK-256-NEXT: ret void // // CHECK-512-LABEL: @write_bool( // CHECK-512-NEXT: entry: // CHECK-512-NEXT: [[X_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-512-NEXT: store <vscale x 16 x i1> [[X:%.*]], <vscale x 16 x i1>* [[X_ADDR]], align 16, [[TBAA15:!tbaa !.*]] -// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[X_ADDR]] to i64* -// CHECK-512-NEXT: [[TMP1:%.*]] = load i64, i64* [[TMP0]], align 16, [[TBAA6]] -// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], %struct.struct_bool* [[S:%.*]], i64 0, i32 1 -// CHECK-512-NEXT: [[TMP2:%.*]] = bitcast [3 x <8 x i8>]* [[Y]] to i64* -// CHECK-512-NEXT: store i64 [[TMP1]], i64* [[TMP2]], align 2, [[TBAA6]] +// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[X_ADDR]] to <8 x i8>* +// CHECK-512-NEXT: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[TMP0]], align 16, [[TBAA6]] +// CHECK-512-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], %struct.struct_bool* [[S:%.*]], i64 0, i32 1, i64 0 +// CHECK-512-NEXT: store <8 x i8> [[TMP1]], <8 x i8>* [[ARRAYIDX]], align 2, [[TBAA6]] // CHECK-512-NEXT: ret void // void write_bool(struct struct_bool *s, svbool_t x) { diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-call.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-call.c index 5442d58e96be..13979197999e 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-call.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-call.c @@ -169,28 +169,24 @@ fixed_float64_t call_float64_ff(svbool_t pg, fixed_float64_t op1, fixed_float64_ // CHECK-NEXT: [[RETVAL_COERCE:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-NEXT: [[TMP0:%.*]] = bitcast <8 x i8>* [[OP1]] to <vscale x 16 x i1>* // CHECK-NEXT: store <vscale x 16 x i1> [[OP1_COERCE:%.*]], <vscale x 16 x i1>* [[TMP0]], align 16 -// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[OP1]] to i64* -// CHECK-NEXT: [[OP113:%.*]] = load i64, i64* [[TMP1]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP2:%.*]] = bitcast <8 x i8>* [[OP2]] to <vscale x 16 x i1>* -// CHECK-NEXT: store <vscale x 16 x i1> [[OP2_COERCE:%.*]], <vscale x 16 x i1>* [[TMP2]], align 16 -// CHECK-NEXT: [[TMP3:%.*]] = bitcast <8 x i8>* [[OP2]] to i64* -// CHECK-NEXT: [[OP224:%.*]] = load i64, i64* [[TMP3]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP4:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to i64* -// CHECK-NEXT: store i64 [[OP113]], i64* [[TMP4]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP5:%.*]] = bitcast <8 x i8>* [[OP2_ADDR]] to i64* -// CHECK-NEXT: store i64 [[OP224]], i64* [[TMP5]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP6:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to <vscale x 16 x i1>* -// CHECK-NEXT: [[TMP7:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP6]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP8:%.*]] = bitcast <8 x i8>* [[OP2_ADDR]] to <vscale x 16 x i1>* -// CHECK-NEXT: [[TMP9:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP8]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.sel.nxv16i1(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i1> [[TMP7]], <vscale x 16 x i1> [[TMP9]]) -// CHECK-NEXT: store <vscale x 16 x i1> [[TMP10]], <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]], align 16, [[TBAA13:!tbaa !.*]] -// CHECK-NEXT: [[TMP11:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to i64* -// CHECK-NEXT: [[TMP12:%.*]] = load i64, i64* [[TMP11]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP13:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to i64* -// CHECK-NEXT: store i64 [[TMP12]], i64* [[TMP13]], align 16 -// CHECK-NEXT: [[TMP14:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 -// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP14]] +// CHECK-NEXT: [[OP11:%.*]] = load <8 x i8>, <8 x i8>* [[OP1]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[OP2]] to <vscale x 16 x i1>* +// CHECK-NEXT: store <vscale x 16 x i1> [[OP2_COERCE:%.*]], <vscale x 16 x i1>* [[TMP1]], align 16 +// CHECK-NEXT: [[OP22:%.*]] = load <8 x i8>, <8 x i8>* [[OP2]], align 16, [[TBAA6]] +// CHECK-NEXT: store <8 x i8> [[OP11]], <8 x i8>* [[OP1_ADDR]], align 16, [[TBAA6]] +// CHECK-NEXT: store <8 x i8> [[OP22]], <8 x i8>* [[OP2_ADDR]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to <vscale x 16 x i1>* +// CHECK-NEXT: [[TMP3:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP2]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP4:%.*]] = bitcast <8 x i8>* [[OP2_ADDR]] to <vscale x 16 x i1>* +// CHECK-NEXT: [[TMP5:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP4]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.sel.nxv16i1(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i1> [[TMP3]], <vscale x 16 x i1> [[TMP5]]) +// CHECK-NEXT: store <vscale x 16 x i1> [[TMP6]], <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]], align 16, [[TBAA13:!tbaa !.*]] +// CHECK-NEXT: [[CASTFIXEDSVE:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to <8 x i8>* +// CHECK-NEXT: [[TMP7:%.*]] = load <8 x i8>, <8 x i8>* [[CASTFIXEDSVE]], align 16, [[TBAA6]] +// CHECK-NEXT: [[RETVAL_0__SROA_CAST:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to <8 x i8>* +// CHECK-NEXT: store <8 x i8> [[TMP7]], <8 x i8>* [[RETVAL_0__SROA_CAST]], align 16 +// CHECK-NEXT: [[TMP8:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 +// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP8]] // fixed_bool_t call_bool_ff(svbool_t pg, fixed_bool_t op1, fixed_bool_t op2) { return svsel(pg, op1, op2); @@ -260,20 +256,18 @@ fixed_float64_t call_float64_fs(svbool_t pg, fixed_float64_t op1, svfloat64_t op // CHECK-NEXT: [[RETVAL_COERCE:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-NEXT: [[TMP0:%.*]] = bitcast <8 x i8>* [[OP1]] to <vscale x 16 x i1>* // CHECK-NEXT: store <vscale x 16 x i1> [[OP1_COERCE:%.*]], <vscale x 16 x i1>* [[TMP0]], align 16 -// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[OP1]] to i64* -// CHECK-NEXT: [[OP112:%.*]] = load i64, i64* [[TMP1]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP2:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to i64* -// CHECK-NEXT: store i64 [[OP112]], i64* [[TMP2]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP3:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to <vscale x 16 x i1>* -// CHECK-NEXT: [[TMP4:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP3]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.sel.nxv16i1(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i1> [[TMP4]], <vscale x 16 x i1> [[OP2:%.*]]) -// CHECK-NEXT: store <vscale x 16 x i1> [[TMP5]], <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]], align 16, [[TBAA13]] -// CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to i64* -// CHECK-NEXT: [[TMP7:%.*]] = load i64, i64* [[TMP6]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP8:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to i64* -// CHECK-NEXT: store i64 [[TMP7]], i64* [[TMP8]], align 16 -// CHECK-NEXT: [[TMP9:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 -// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP9]] +// CHECK-NEXT: [[OP11:%.*]] = load <8 x i8>, <8 x i8>* [[OP1]], align 16, [[TBAA6]] +// CHECK-NEXT: store <8 x i8> [[OP11]], <8 x i8>* [[OP1_ADDR]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[OP1_ADDR]] to <vscale x 16 x i1>* +// CHECK-NEXT: [[TMP2:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP1]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.sel.nxv16i1(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i1> [[TMP2]], <vscale x 16 x i1> [[OP2:%.*]]) +// CHECK-NEXT: store <vscale x 16 x i1> [[TMP3]], <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]], align 16, [[TBAA13]] +// CHECK-NEXT: [[CASTFIXEDSVE:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to <8 x i8>* +// CHECK-NEXT: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[CASTFIXEDSVE]], align 16, [[TBAA6]] +// CHECK-NEXT: [[RETVAL_0__SROA_CAST:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to <8 x i8>* +// CHECK-NEXT: store <8 x i8> [[TMP4]], <8 x i8>* [[RETVAL_0__SROA_CAST]], align 16 +// CHECK-NEXT: [[TMP5:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 +// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP5]] // fixed_bool_t call_bool_fs(svbool_t pg, fixed_bool_t op1, svbool_t op2) { return svsel(pg, op1, op2); @@ -325,12 +319,12 @@ fixed_float64_t call_float64_ss(svbool_t pg, svfloat64_t op1, svfloat64_t op2) { // CHECK-NEXT: [[RETVAL_COERCE:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.sel.nxv16i1(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i1> [[OP1:%.*]], <vscale x 16 x i1> [[OP2:%.*]]) // CHECK-NEXT: store <vscale x 16 x i1> [[TMP0]], <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]], align 16, [[TBAA13]] -// CHECK-NEXT: [[TMP1:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to i64* -// CHECK-NEXT: [[TMP2:%.*]] = load i64, i64* [[TMP1]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP3:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to i64* -// CHECK-NEXT: store i64 [[TMP2]], i64* [[TMP3]], align 16 -// CHECK-NEXT: [[TMP4:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 -// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP4]] +// CHECK-NEXT: [[CASTFIXEDSVE:%.*]] = bitcast <vscale x 16 x i1>* [[SAVED_CALL_RVALUE]] to <8 x i8>* +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[CASTFIXEDSVE]], align 16, [[TBAA6]] +// CHECK-NEXT: [[RETVAL_0__SROA_CAST:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to <8 x i8>* +// CHECK-NEXT: store <8 x i8> [[TMP1]], <8 x i8>* [[RETVAL_0__SROA_CAST]], align 16 +// CHECK-NEXT: [[TMP2:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 +// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]] // fixed_bool_t call_bool_ss(svbool_t pg, svbool_t op1, svbool_t op2) { return svsel(pg, op1, op2); diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c index 17267d6038e4..8568c820ae6f 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c @@ -81,13 +81,11 @@ fixed_float64_t from_svfloat64_t(svfloat64_t type) { // CHECK-NEXT: [[TYPE_ADDR:%.*]] = alloca <8 x i8>, align 16 // CHECK-NEXT: [[TMP0:%.*]] = bitcast <8 x i8>* [[TYPE]] to <vscale x 16 x i1>* // CHECK-NEXT: store <vscale x 16 x i1> [[TYPE_COERCE:%.*]], <vscale x 16 x i1>* [[TMP0]], align 16 -// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[TYPE]] to i64* -// CHECK-NEXT: [[TYPE12:%.*]] = load i64, i64* [[TMP1]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP2:%.*]] = bitcast <8 x i8>* [[TYPE_ADDR]] to i64* -// CHECK-NEXT: store i64 [[TYPE12]], i64* [[TMP2]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP3:%.*]] = bitcast <8 x i8>* [[TYPE_ADDR]] to <vscale x 16 x i1>* -// CHECK-NEXT: [[TMP4:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP3]], align 16, [[TBAA6]] -// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP4]] +// CHECK-NEXT: [[TYPE1:%.*]] = load <8 x i8>, <8 x i8>* [[TYPE]], align 16, [[TBAA6]] +// CHECK-NEXT: store <8 x i8> [[TYPE1]], <8 x i8>* [[TYPE_ADDR]], align 16, [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i8>* [[TYPE_ADDR]] to <vscale x 16 x i1>* +// CHECK-NEXT: [[TMP2:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[TMP1]], align 16, [[TBAA6]] +// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]] // svbool_t to_svbool_t(fixed_bool_t type) { return type; @@ -98,12 +96,12 @@ svbool_t to_svbool_t(fixed_bool_t type) { // CHECK-NEXT: [[TYPE_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-NEXT: [[RETVAL_COERCE:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-NEXT: store <vscale x 16 x i1> [[TYPE:%.*]], <vscale x 16 x i1>* [[TYPE_ADDR]], align 16, [[TBAA13:!tbaa !.*]] -// CHECK-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[TYPE_ADDR]] to i64* -// CHECK-NEXT: [[TMP1:%.*]] = load i64, i64* [[TMP0]], align 16, [[TBAA6]] -// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to i64* -// CHECK-NEXT: store i64 [[TMP1]], i64* [[TMP2]], align 16 -// CHECK-NEXT: [[TMP3:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 -// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]] +// CHECK-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[TYPE_ADDR]] to <8 x i8>* +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[TMP0]], align 16, [[TBAA6]] +// CHECK-NEXT: [[RETVAL_0__SROA_CAST:%.*]] = bitcast <vscale x 16 x i1>* [[RETVAL_COERCE]] to <8 x i8>* +// CHECK-NEXT: store <8 x i8> [[TMP1]], <8 x i8>* [[RETVAL_0__SROA_CAST]], align 16 +// CHECK-NEXT: [[TMP2:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[RETVAL_COERCE]], align 16 +// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]] // fixed_bool_t from_svbool_t(svbool_t type) { return type; diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c index 5babb9c7c410..8f0d3c9f97e7 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c @@ -72,9 +72,9 @@ void write_global_bf16(svbfloat16_t v) { global_bf16 = v; } // CHECK-512-NEXT: entry: // CHECK-512-NEXT: [[V_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 16 // CHECK-512-NEXT: store <vscale x 16 x i1> [[V:%.*]], <vscale x 16 x i1>* [[V_ADDR]], align 16, [[TBAA13:!tbaa !.*]] -// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[V_ADDR]] to i64* -// CHECK-512-NEXT: [[TMP1:%.*]] = load i64, i64* [[TMP0]], align 16, [[TBAA10]] -// CHECK-512-NEXT: store i64 [[TMP1]], i64* bitcast (<8 x i8>* @global_bool to i64*), align 2, [[TBAA10]] +// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1>* [[V_ADDR]] to <8 x i8>* +// CHECK-512-NEXT: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[TMP0]], align 16, [[TBAA10]] +// CHECK-512-NEXT: store <8 x i8> [[TMP1]], <8 x i8>* @global_bool, align 2, [[TBAA10]] // CHECK-512-NEXT: ret void // void write_global_bool(svbool_t v) { global_bool = v; } diff --git a/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp b/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp index 328cbde4da67..b7bb34022ecc 100644 --- a/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp +++ b/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp @@ -554,42 +554,8 @@ static Instruction *combineLoadToOperationType(InstCombinerImpl &IC, if (LI.getPointerOperand()->isSwiftError()) return nullptr; - Type *Ty = LI.getType(); const DataLayout &DL = IC.getDataLayout(); - // Try to canonicalize loads which are only ever stored to operate over - // integers instead of any other type. We only do this when the loaded type - // is sized and has a size exactly the same as its store size and the store - // size is a legal integer type. - // Do not perform canonicalization if minmax pattern is found (to avoid - // infinite loop). - Type *Dummy; - if (!Ty->isIntegerTy() && Ty->isSized() && !isa<ScalableVectorType>(Ty) && - DL.isLegalInteger(DL.getTypeStoreSizeInBits(Ty)) && - DL.typeSizeEqualsStoreSize(Ty) && !DL.isNonIntegralPointerType(Ty) && - !isMinMaxWithLoads(InstCombiner::peekThroughBitcast( - LI.getPointerOperand(), /*OneUseOnly=*/true), - Dummy)) { - if (all_of(LI.users(), [&LI](User *U) { - auto *SI = dyn_cast<StoreInst>(U); - return SI && SI->getPointerOperand() != &LI && - !SI->getPointerOperand()->isSwiftError(); - })) { - LoadInst *NewLoad = IC.combineLoadToNewType( - LI, Type::getIntNTy(LI.getContext(), DL.getTypeStoreSizeInBits(Ty))); - // Replace all the stores with stores of the newly loaded value. - for (auto UI = LI.user_begin(), UE = LI.user_end(); UI != UE;) { - auto *SI = cast<StoreInst>(*UI++); - IC.Builder.SetInsertPoint(SI); - combineStoreToNewValue(IC, *SI, NewLoad); - IC.eraseInstFromFunction(*SI); - } - assert(LI.use_empty() && "Failed to remove all users of the load!"); - // Return the old load so the combiner can delete it safely. - return &LI; - } - } - // Fold away bit casts of the loaded value by loading the desired type. // We can do this for BitCastInsts as well as casts from and to pointer types, // as long as those are noops (i.e., the source or dest type have the same diff --git a/llvm/test/Transforms/InstCombine/atomic.ll b/llvm/test/Transforms/InstCombine/atomic.ll index 382d23d15309..fc134bcd7e4b 100644 --- a/llvm/test/Transforms/InstCombine/atomic.ll +++ b/llvm/test/Transforms/InstCombine/atomic.ll @@ -325,11 +325,9 @@ declare void @clobber() define i32 @test18(float* %p) { ; CHECK-LABEL: @test18( -; CHECK-NEXT: [[TMP1:%.*]] = bitcast float* [[P:%.*]] to i32* -; CHECK-NEXT: [[X1:%.*]] = load atomic i32, i32* [[TMP1]] unordered, align 4 +; CHECK-NEXT: [[X:%.*]] = load atomic float, float* [[P:%.*]] unordered, align 4 ; CHECK-NEXT: call void @clobber() -; CHECK-NEXT: [[TMP2:%.*]] = bitcast float* [[P]] to i32* -; CHECK-NEXT: store atomic i32 [[X1]], i32* [[TMP2]] unordered, align 4 +; CHECK-NEXT: store atomic float [[X]], float* [[P]] unordered, align 4 ; CHECK-NEXT: ret i32 0 ; %x = load atomic float, float* %p unordered, align 4 @@ -376,10 +374,8 @@ define i32 @test21(i32** %p, i8* %v) { define void @pr27490a(i8** %p1, i8** %p2) { ; CHECK-LABEL: @pr27490a( -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8** [[P1:%.*]] to i64* -; CHECK-NEXT: [[L1:%.*]] = load i64, i64* [[TMP1]], align 8 -; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8** [[P2:%.*]] to i64* -; CHECK-NEXT: store volatile i64 [[L1]], i64* [[TMP2]], align 8 +; CHECK-NEXT: [[L:%.*]] = load i8*, i8** [[P1:%.*]], align 8 +; CHECK-NEXT: store volatile i8* [[L]], i8** [[P2:%.*]], align 8 ; CHECK-NEXT: ret void ; %l = load i8*, i8** %p1 @@ -389,10 +385,8 @@ define void @pr27490a(i8** %p1, i8** %p2) { define void @pr27490b(i8** %p1, i8** %p2) { ; CHECK-LABEL: @pr27490b( -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8** [[P1:%.*]] to i64* -; CHECK-NEXT: [[L1:%.*]] = load i64, i64* [[TMP1]], align 8 -; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8** [[P2:%.*]] to i64* -; CHECK-NEXT: store atomic i64 [[L1]], i64* [[TMP2]] seq_cst, align 8 +; CHECK-NEXT: [[L:%.*]] = load i8*, i8** [[P1:%.*]], align 8 +; CHECK-NEXT: store atomic i8* [[L]], i8** [[P2:%.*]] seq_cst, align 8 ; CHECK-NEXT: ret void ; %l = load i8*, i8** %p1 diff --git a/llvm/test/Transforms/InstCombine/load.ll b/llvm/test/Transforms/InstCombine/load.ll index 4fc87219fa83..032da41de6a5 100644 --- a/llvm/test/Transforms/InstCombine/load.ll +++ b/llvm/test/Transforms/InstCombine/load.ll @@ -205,18 +205,16 @@ define i8 @test15(i8 %x, i32 %y) { define void @test16(i8* %x, i8* %a, i8* %b, i8* %c) { ; CHECK-LABEL: @test16( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8* [[X:%.*]] to i32* -; CHECK-NEXT: [[X11:%.*]] = load i32, i32* [[TMP0]], align 4 -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[A:%.*]] to i32* -; CHECK-NEXT: store i32 [[X11]], i32* [[TMP1]], align 4 -; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[B:%.*]] to i32* -; CHECK-NEXT: store i32 [[X11]], i32* [[TMP2]], align 4 -; CHECK-NEXT: [[TMP3:%.*]] = bitcast i8* [[X]] to i32* -; CHECK-NEXT: [[X22:%.*]] = load i32, i32* [[TMP3]], align 4 -; CHECK-NEXT: [[TMP4:%.*]] = bitcast i8* [[B]] to i32* -; CHECK-NEXT: store i32 [[X22]], i32* [[TMP4]], align 4 -; CHECK-NEXT: [[TMP5:%.*]] = bitcast i8* [[C:%.*]] to i32* -; CHECK-NEXT: store i32 [[X22]], i32* [[TMP5]], align 4 +; CHECK-NEXT: [[X_CAST:%.*]] = bitcast i8* [[X:%.*]] to float* +; CHECK-NEXT: [[A_CAST:%.*]] = bitcast i8* [[A:%.*]] to float* +; CHECK-NEXT: [[B_CAST:%.*]] = bitcast i8* [[B:%.*]] to float* +; CHECK-NEXT: [[X1:%.*]] = load float, float* [[X_CAST]], align 4 +; CHECK-NEXT: store float [[X1]], float* [[A_CAST]], align 4 +; CHECK-NEXT: store float [[X1]], float* [[B_CAST]], align 4 +; CHECK-NEXT: [[X2:%.*]] = load float, float* [[X_CAST]], align 4 +; CHECK-NEXT: store float [[X2]], float* [[B_CAST]], align 4 +; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8* [[C:%.*]] to float* +; CHECK-NEXT: store float [[X2]], float* [[TMP0]], align 4 ; CHECK-NEXT: ret void ; entry: @@ -240,18 +238,16 @@ entry: define void @test16-vect(i8* %x, i8* %a, i8* %b, i8* %c) { ; CHECK-LABEL: @test16-vect( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8* [[X:%.*]] to i32* -; CHECK-NEXT: [[X11:%.*]] = load i32, i32* [[TMP0]], align 4 -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[A:%.*]] to i32* -; CHECK-NEXT: store i32 [[X11]], i32* [[TMP1]], align 4 -; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[B:%.*]] to i32* -; CHECK-NEXT: store i32 [[X11]], i32* [[TMP2]], align 4 -; CHECK-NEXT: [[TMP3:%.*]] = bitcast i8* [[X]] to i32* -; CHECK-NEXT: [[X22:%.*]] = load i32, i32* [[TMP3]], align 4 -; CHECK-NEXT: [[TMP4:%.*]] = bitcast i8* [[B]] to i32* -; CHECK-NEXT: store i32 [[X22]], i32* [[TMP4]], align 4 -; CHECK-NEXT: [[TMP5:%.*]] = bitcast i8* [[C:%.*]] to i32* -; CHECK-NEXT: store i32 [[X22]], i32* [[TMP5]], align 4 +; CHECK-NEXT: [[X_CAST:%.*]] = bitcast i8* [[X:%.*]] to <4 x i8>* +; CHECK-NEXT: [[A_CAST:%.*]] = bitcast i8* [[A:%.*]] to <4 x i8>* +; CHECK-NEXT: [[B_CAST:%.*]] = bitcast i8* [[B:%.*]] to <4 x i8>* +; CHECK-NEXT: [[X1:%.*]] = load <4 x i8>, <4 x i8>* [[X_CAST]], align 4 +; CHECK-NEXT: store <4 x i8> [[X1]], <4 x i8>* [[A_CAST]], align 4 +; CHECK-NEXT: store <4 x i8> [[X1]], <4 x i8>* [[B_CAST]], align 4 +; CHECK-NEXT: [[X2:%.*]] = load <4 x i8>, <4 x i8>* [[X_CAST]], align 4 +; CHECK-NEXT: store <4 x i8> [[X2]], <4 x i8>* [[B_CAST]], align 4 +; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8* [[C:%.*]] to <4 x i8>* +; CHECK-NEXT: store <4 x i8> [[X2]], <4 x i8>* [[TMP0]], align 4 ; CHECK-NEXT: ret void ; entry: diff --git a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll index e443c6ed0075..42f6900c2ab5 100644 --- a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll +++ b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll @@ -161,24 +161,11 @@ exit: } define void @test_load_cast_combine_nonnull(float** %ptr) { -; We can't preserve nonnull metadata when converting a load of a pointer to -; a load of an integer. Instead, we translate it to range metadata. -; FIXME: We should also transform range metadata back into nonnull metadata. -; FIXME: This test is very fragile. If any LABEL lines are added after -; this point, the test will fail, because this test depends on a metadata tuple, -; which is always emitted at the end of the file. At some point, we should -; consider an option to the IR printer to emit MD tuples after the function -; that first uses them--this will allow us to refer to them like this and not -; have the tests break. For now, this function must always come last in this -; file, and no LABEL lines are to be added after this point. -; ; CHECK-LABEL: @test_load_cast_combine_nonnull( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP0:%.*]] = bitcast float** [[PTR:%.*]] to i64* -; CHECK-NEXT: [[P1:%.*]] = load i64, i64* [[TMP0]], align 8, !range ![[MD:[0-9]+]] +; CHECK-NEXT: [[P:%.*]] = load float*, float** [[PTR:%.*]], align 8, !nonnull !7 ; CHECK-NEXT: [[GEP:%.*]] = getelementptr float*, float** [[PTR]], i64 42 -; CHECK-NEXT: [[TMP1:%.*]] = bitcast float** [[GEP]] to i64* -; CHECK-NEXT: store i64 [[P1]], i64* [[TMP1]], align 8 +; CHECK-NEXT: store float* [[P]], float** [[GEP]], align 8 ; CHECK-NEXT: ret void ; entry: @@ -188,8 +175,6 @@ entry: ret void } -; This is the metadata tuple that we reference above: -; CHECK: ![[MD]] = !{i64 1, i64 0} !0 = !{!1, !1, i64 0} !1 = !{!"scalar type", !2} !2 = !{!"root"} diff --git a/llvm/test/Transforms/InstCombine/non-integral-pointers.ll b/llvm/test/Transforms/InstCombine/non-integral-pointers.ll index e8f0013604a9..a1166bf49193 100644 --- a/llvm/test/Transforms/InstCombine/non-integral-pointers.ll +++ b/llvm/test/Transforms/InstCombine/non-integral-pointers.ll @@ -41,10 +41,8 @@ define void @f_3(i8 addrspace(3)** %ptr0, i8 addrspace(3)** %ptr1) { ; integers, since pointers in address space 3 are integral. ; CHECK-LABEL: @f_3( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8 addrspace(3)** [[PTR0:%.*]] to i64* -; CHECK-NEXT: [[VAL1:%.*]] = load i64, i64* [[TMP0]], align 8 -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8 addrspace(3)** [[PTR1:%.*]] to i64* -; CHECK-NEXT: store i64 [[VAL1]], i64* [[TMP1]], align 8 +; CHECK-NEXT: [[VAL:%.*]] = load i8 addrspace(3)*, i8 addrspace(3)** [[PTR0:%.*]], align 8 +; CHECK-NEXT: store i8 addrspace(3)* [[VAL]], i8 addrspace(3)** [[PTR1:%.*]], align 8 ; CHECK-NEXT: ret void ; entry: @@ -79,13 +77,13 @@ define i64 @g(i8 addrspace(4)** %gp) { define i64 @g2(i8* addrspace(4)* %gp) { ; CHECK-LABEL: @g2( -; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* addrspace(4)* [[GP:%.*]] to i64 addrspace(4)* -; CHECK-NEXT: [[DOTPRE1:%.*]] = load i64, i64 addrspace(4)* [[TMP1]], align 8 +; CHECK-NEXT: [[DOTPRE:%.*]] = load i8*, i8* addrspace(4)* [[GP:%.*]], align 8 ; CHECK-NEXT: [[V74:%.*]] = call i8 addrspace(4)* @alloc() ; CHECK-NEXT: [[V77:%.*]] = getelementptr i8, i8 addrspace(4)* [[V74]], i64 -8 -; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8 addrspace(4)* [[V77]] to i64 addrspace(4)* -; CHECK-NEXT: store i64 [[DOTPRE1]], i64 addrspace(4)* [[TMP2]], align 8 -; CHECK-NEXT: ret i64 [[DOTPRE1]] +; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8 addrspace(4)* [[V77]] to i8* addrspace(4)* +; CHECK-NEXT: store i8* [[DOTPRE]], i8* addrspace(4)* [[TMP1]], align 8 +; CHECK-NEXT: [[V81_CAST:%.*]] = ptrtoint i8* [[DOTPRE]] to i64 +; CHECK-NEXT: ret i64 [[V81_CAST]] ; %.pre = load i8*, i8* addrspace(4)* %gp, align 8 %v74 = call i8 addrspace(4)* @alloc() diff --git a/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll b/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll index 6de0282e9448..3308a0ecc7fa 100644 --- a/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll +++ b/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll @@ -50,10 +50,10 @@ target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16 define dso_local void @_Z3gen1S(%0* noalias sret align 8 %arg, %0* byval(%0) align 8 %arg1) { ; CHECK-LABEL: @_Z3gen1S( ; CHECK-NEXT: bb: -; CHECK-NEXT: [[TMP0:%.*]] = bitcast %0* [[ARG1:%.*]] to i64* -; CHECK-NEXT: [[I21:%.*]] = load i64, i64* [[TMP0]], align 8 -; CHECK-NEXT: [[TMP1:%.*]] = bitcast %0* [[ARG:%.*]] to i64* -; CHECK-NEXT: store i64 [[I21]], i64* [[TMP1]], align 8 +; CHECK-NEXT: [[I:%.*]] = getelementptr inbounds [[TMP0:%.*]], %0* [[ARG1:%.*]], i64 0, i32 0 +; CHECK-NEXT: [[I2:%.*]] = load i32*, i32** [[I]], align 8 +; CHECK-NEXT: [[I3:%.*]] = getelementptr inbounds [[TMP0]], %0* [[ARG:%.*]], i64 0, i32 0 +; CHECK-NEXT: store i32* [[I2]], i32** [[I3]], align 8 ; CHECK-NEXT: ret void ; bb: @@ -68,13 +68,12 @@ define dso_local i32* @_Z3foo1S(%0* byval(%0) align 8 %arg) { ; CHECK-LABEL: @_Z3foo1S( ; CHECK-NEXT: bb: ; CHECK-NEXT: [[I2:%.*]] = alloca [[TMP0:%.*]], align 8 -; CHECK-NEXT: [[I1_SROA_0_0_I5_SROA_CAST:%.*]] = bitcast %0* [[ARG:%.*]] to i64* -; CHECK-NEXT: [[I1_SROA_0_0_COPYLOAD:%.*]] = load i64, i64* [[I1_SROA_0_0_I5_SROA_CAST]], align 8 -; CHECK-NEXT: [[I_SROA_0_0_I6_SROA_CAST:%.*]] = bitcast %0* [[I2]] to i64* -; CHECK-NEXT: store i64 [[I1_SROA_0_0_COPYLOAD]], i64* [[I_SROA_0_0_I6_SROA_CAST]], align 8 +; CHECK-NEXT: [[I1_SROA_0_0_I5_SROA_IDX:%.*]] = getelementptr inbounds [[TMP0]], %0* [[ARG:%.*]], i64 0, i32 0 +; CHECK-NEXT: [[I1_SROA_0_0_COPYLOAD:%.*]] = load i32*, i32** [[I1_SROA_0_0_I5_SROA_IDX]], align 8 +; CHECK-NEXT: [[I_SROA_0_0_I6_SROA_IDX:%.*]] = getelementptr inbounds [[TMP0]], %0* [[I2]], i64 0, i32 0 +; CHECK-NEXT: store i32* [[I1_SROA_0_0_COPYLOAD]], i32** [[I_SROA_0_0_I6_SROA_IDX]], align 8 ; CHECK-NEXT: tail call void @_Z7escape01S(%0* nonnull byval(%0) align 8 [[I2]]) -; CHECK-NEXT: [[TMP0]] = inttoptr i64 [[I1_SROA_0_0_COPYLOAD]] to i32* -; CHECK-NEXT: ret i32* [[TMP0]] +; CHECK-NEXT: ret i32* [[I1_SROA_0_0_COPYLOAD]] ; bb: %i = alloca %0, align 8 @@ -108,24 +107,21 @@ declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture) define dso_local i32* @_Z3bar1S(%0* byval(%0) align 8 %arg) { ; CHECK-LABEL: @_Z3bar1S( ; CHECK-NEXT: bb: -; CHECK-NEXT: [[I1_SROA_0_0_I4_SROA_CAST:%.*]] = bitcast %0* [[ARG:%.*]] to i64* -; CHECK-NEXT: [[I1_SROA_0_0_COPYLOAD:%.*]] = load i64, i64* [[I1_SROA_0_0_I4_SROA_CAST]], align 8 +; CHECK-NEXT: [[I1_SROA_0_0_I4_SROA_IDX:%.*]] = getelementptr inbounds [[TMP0:%.*]], %0* [[ARG:%.*]], i64 0, i32 0 +; CHECK-NEXT: [[I1_SROA_0_0_COPYLOAD:%.*]] = load i32*, i32** [[I1_SROA_0_0_I4_SROA_IDX]], align 8 ; CHECK-NEXT: [[I5:%.*]] = tail call i32 @_Z4condv() ; CHECK-NEXT: [[I6_NOT:%.*]] = icmp eq i32 [[I5]], 0 ; CHECK-NEXT: br i1 [[I6_NOT]], label [[BB10:%.*]], label [[BB7:%.*]] ; CHECK: bb7: ; CHECK-NEXT: tail call void @_Z5sync0v() -; CHECK-NEXT: [[TMP0:%.*]] = inttoptr i64 [[I1_SROA_0_0_COPYLOAD]] to i32* -; CHECK-NEXT: tail call void @_Z7escape0Pi(i32* [[TMP0]]) +; CHECK-NEXT: tail call void @_Z7escape0Pi(i32* [[I1_SROA_0_0_COPYLOAD]]) ; CHECK-NEXT: br label [[BB13:%.*]] ; CHECK: bb10: ; CHECK-NEXT: tail call void @_Z5sync1v() -; CHECK-NEXT: [[TMP1:%.*]] = inttoptr i64 [[I1_SROA_0_0_COPYLOAD]] to i32* -; CHECK-NEXT: tail call void @_Z7escape1Pi(i32* [[TMP1]]) +; CHECK-NEXT: tail call void @_Z7escape1Pi(i32* [[I1_SROA_0_0_COPYLOAD]]) ; CHECK-NEXT: br label [[BB13]] ; CHECK: bb13: -; CHECK-NEXT: [[DOTPRE_PHI:%.*]] = phi i32* [ [[TMP1]], [[BB10]] ], [ [[TMP0]], [[BB7]] ] -; CHECK-NEXT: ret i32* [[DOTPRE_PHI]] +; CHECK-NEXT: ret i32* [[I1_SROA_0_0_COPYLOAD]] ; bb: %i = alloca %0, align 8 </cut>

3 years, 4 months

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

linaro-toolchain