From 00f34291175981393b2a9802ada879e72335dcd0 Mon Sep 17 00:00:00 2001 From: Martin Liska Date: Fri, 8 Oct 2021 09:48:58 +0200 Subject: Come up with OPTION_SET_P macro. gcc/ada/ChangeLog: * gcc-interface/misc.c (gnat_post_options): Use new macro OPTION_SET_P. (gnat_init_gcc_eh): Likewise. (gnat_init_gcc_fp): Likewise. gcc/c-family/ChangeLog: * c-opts.c (c_common_post_options): Use new macro OPTION_SET_P. gcc/ChangeLog: * config/alpha/alpha.c (alpha_option_override): Use new macro OPTION_SET_P. * config/arc/arc.c (arc_override_options): Likewise. * config/arm/arm.c (arm_option_override): Likewise. * config/bfin/bfin.c (bfin_load_pic_reg): Likewise. * config/c6x/c6x.c (c6x_option_override): Likewise. * config/csky/csky.c: Likewise. * config/darwin.c (darwin_override_options): Likewise. * config/frv/frv.c (frv_option_override): Likewise. * config/i386/djgpp.h: Likewise. * config/i386/i386.c (ix86_stack_protect_guard): Likewise. (ix86_max_noce_ifcvt_seq_cost): Likewise. * config/ia64/ia64.c (ia64_option_override): Likewise. (ia64_override_options_after_change): Likewise. * config/m32c/m32c.c (m32c_option_override): Likewise. * config/m32r/m32r.c (m32r_init): Likewise. * config/m68k/m68k.c (m68k_option_override): Likewise. * config/microblaze/microblaze.c (microblaze_option_override): Likewise. * config/mips/mips.c (mips_option_override): Likewise. * config/nios2/nios2.c (nios2_option_override): Likewise. * config/nvptx/nvptx.c (nvptx_option_override): Likewise. * config/pa/pa.c (pa_option_override): Likewise. * config/riscv/riscv.c (riscv_option_override): Likewise. * config/rs6000/aix71.h: Likewise. * config/rs6000/aix72.h: Likewise. * config/rs6000/aix73.h: Likewise. * config/rs6000/rs6000.c (darwin_rs6000_override_options): Likewise. (rs6000_override_options_after_change): Likewise. (rs6000_linux64_override_options): Likewise. (glibc_supports_ieee_128bit): Likewise. (rs6000_option_override_internal): Likewise. (rs6000_file_start): Likewise. (rs6000_darwin_file_start): Likewise. * config/rs6000/rtems.h: Likewise. * config/rs6000/sysv4.h: Likewise. * config/rs6000/vxworks.h (SUB3TARGET_OVERRIDE_OPTIONS): Likewise. * config/s390/s390.c (s390_option_override): Likewise. * config/sh/linux.h: Likewise. * config/sh/netbsd-elf.h (while): Likewise. * config/sh/sh.c (sh_option_override): Likewise. * config/sol2.c (solaris_override_options): Likewise. * config/sparc/sparc.c (sparc_option_override): Likewise. * config/tilegx/tilegx.c (tilegx_option_override): Likewise. * config/visium/visium.c (visium_option_override): Likewise. * config/vxworks.c (vxworks_override_options): Likewise. * lto-opts.c (lto_write_options): Likewise. * omp-expand.c (expand_omp_simd): Likewise. * omp-general.c (omp_max_vf): Likewise. * omp-offload.c (oacc_xform_loop): Likewise. * opts.h (OPTION_SET_P): Likewise. * targhooks.c (default_max_noce_ifcvt_seq_cost): Likewise. * toplev.c (process_options): Likewise. * tree-predcom.c: Likewise. * tree-sra.c (analyze_all_variable_accesses): Likewise. gcc/cp/ChangeLog: * constexpr.c (maybe_warn_about_constant_value): Use new macro OPTION_SET_P. * decl.c (redeclaration_error_message): Likewise. (cxx_init_decl_processing): Likewise. gcc/d/ChangeLog: * d-lang.cc (d_post_options): Use new macro OPTION_SET_P. gcc/fortran/ChangeLog: * options.c (gfc_post_options): Use new macro OPTION_SET_P. gcc/objc/ChangeLog: * objc-next-runtime-abi-01.c: Use new macro OPTION_SET_P. * objc-next-runtime-abi-02.c (objc_next_runtime_abi_02_init): Likewise. --- gcc/tree-predcom.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'gcc/tree-predcom.c') diff --git a/gcc/tree-predcom.c b/gcc/tree-predcom.c index 6b195d1..ce1f08f 100644 --- a/gcc/tree-predcom.c +++ b/gcc/tree-predcom.c @@ -233,6 +233,7 @@ along with GCC; see the file COPYING3. If not see #include "tree-scalar-evolution.h" #include "tree-affine.h" #include "builtins.h" +#include "opts.h" /* The maximum number of iterations between the considered memory references. */ @@ -3492,7 +3493,7 @@ public: only if predictive commoning isn't set explicitly, and it doesn't allow unrolling. */ if (flag_tree_loop_vectorize - && !global_options_set.x_flag_predictive_commoning) + && !OPTION_SET_P (flag_predictive_commoning)) return true; return false; -- cgit v1.1 From 0ee3dc6052361290c92bba492cc0a9e556b31055 Mon Sep 17 00:00:00 2001 From: Richard Sandiford Date: Mon, 4 Oct 2021 23:55:43 +0100 Subject: loop: Fix profile updates after unrolling [PR102385] In g:62acc72a957b5614 I'd stopped the unroller from using an epilogue loop in cases where the iteration count was known to be a multiple of the unroll factor. The epilogue and non-epilogue cases still shared this (preexisting) code to update the edge frequencies: basic_block exit_bb = single_pred (loop->latch); new_exit = find_edge (exit_bb, rest); new_exit->probability = profile_probability::always () .apply_scale (1, new_est_niter + 1); [etc] But of course (in hindsight) that only makes sense for the epilogue case, where we've already moved the main loop's exit edge to be a sibling of the latch edge. For the non-epilogue case, the exit edge stays (and needs to stay) in its original position. I don't really understand what the code is trying to do for the epilogue case. It has: /* Ensure that the frequencies in the loop match the new estimated number of iterations, and change the probability of the new exit edge. */ profile_count freq_h = loop->header->count; profile_count freq_e = (loop_preheader_edge (loop))->count (); if (freq_h.nonzero_p ()) { ... scale_loop_frequencies (loop, freq_e.probability_in (freq_h)); } Here, freq_e.probability_in (freq_h) is freq_e / freq_h, so for the header block, this has the effect of: new header count = freq_h * (freq_e / freq_h) i.e. we say that the header executes exactly as often as the preheader edge, which would only make sense if the loop never iterates. Also, after setting the probability of the nonexit edge (correctly) to new_est_niter / (new_est_niter + 1), the code does: scale_bbs_frequencies (&loop->latch, 1, prob); for this new probability. I think that only makes sense if the nonexit edge was previously unconditional (100%). But the code carefully preserved the probability of the original exit edge when creating the new one. All I'm trying to do here though is fix the mess I created and get the probabilities right for the non-epilogue case. Things are simpler there since we don't have to worry about loop versioning. Hopefully the comments explain the approach. The function's current interface implies that it can cope with multiple exit edges and that the function only needs the iteration count relative to one of those edges in order to work correctly. In practice that's not the case: it assumes there is exactly one exit edge and all current callers also ensure that the exit test dominates the latch. I think the function is easier to follow if we remove the implied generality. gcc/ PR tree-optimization/102385 * predict.h (change_edge_frequency): Declare. * predict.c (change_edge_frequency): New function. * tree-ssa-loop-manip.h (tree_transform_and_unroll_loop): Remove edge argument. (tree_unroll_loop): Likewise. * gimple-loop-jam.c (tree_loop_unroll_and_jam): Update accordingly. * tree-predcom.c (pcom_worker::tree_predictive_commoning_loop): Likewise. * tree-ssa-loop-prefetch.c (loop_prefetch_arrays): Likewise. * tree-ssa-loop-manip.c (tree_unroll_loop): Likewise. (tree_transform_and_unroll_loop): Likewise. Use single_dom_exit to retrieve the exit edges. Make all the old profile update code conditional on !single_loop_p -- the case it was written for -- and use a different approach for the single-loop case. gcc/testsuite/ * gcc.dg/pr102385.c: New test. --- gcc/tree-predcom.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'gcc/tree-predcom.c') diff --git a/gcc/tree-predcom.c b/gcc/tree-predcom.c index ce1f08f..208e755 100644 --- a/gcc/tree-predcom.c +++ b/gcc/tree-predcom.c @@ -3398,8 +3398,7 @@ pcom_worker::tree_predictive_commoning_loop (bool allow_unroll_p) the phi nodes in execute_pred_commoning_cbck. A bit hacky. */ replace_phis_by_defined_names (m_chains); - edge exit = single_dom_exit (m_loop); - tree_transform_and_unroll_loop (m_loop, unroll_factor, exit, &desc, + tree_transform_and_unroll_loop (m_loop, unroll_factor, &desc, execute_pred_commoning_cbck, &dta); eliminate_temp_copies (m_loop, tmp_vars); } -- cgit v1.1