linaro-toolchain July 2021

linaro-toolchain@lists.linaro.org

19 participants
99 discussions

by Pascal Bourguignon

Hello, I've got a little patch for qdl, adding error messages. https://git.linaro.org/landing-teams/working/qualcomm/qdl.git/ What would be the right place to send it to? -- __Pascal Bourguignon__ http://www.sbde.fr/

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_gnu/gnu-master-arm-bootstrap_debug - Build # 13 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *gcc* in CI configuration tcwg_gnu/gnu-master-arm-bootstrap_debug. So far, this commit has regressed CI configurations: - tcwg_gnu/gnu-master-arm-bootstrap_debug Culprit: <cut> commit a110855667782dac7b674d3e328b253b3b3c919b Author: Martin Sebor <msebor(a)redhat.com> Date: Wed Jul 7 14:05:25 2021 -0600 Correct handling of variable offset minus constant in -Warray-bounds [PR100137] Resolves: PR tree-optimization/100137 - -Warray-bounds false positive on varying offset plus negative PR tree-optimization/99121 - ICE in -Warray-bounds on a multidimensional PR tree-optimization/97027 - missing warning on buffer overflow storing a larger scalar into a smaller array gcc/ChangeLog: PR tree-optimization/100137 PR tree-optimization/99121 PR tree-optimization/97027 * builtins.c (access_ref::access_ref): Also set offmax. (access_ref::offset_in_range): Define new function. (access_ref::add_offset): Set offmax. (access_ref::inform_access): Handle access_none. (handle_mem_ref): Clear ostype. (compute_objsize_r): Handle ASSERT_EXPR. * builtins.h (struct access_ref): Add offmax member. * gimple-array-bounds.cc (array_bounds_checker::check_mem_ref): Use compute_objsize() and simplify. gcc/testsuite/ChangeLog: PR tree-optimization/100137 PR tree-optimization/99121 PR tree-optimization/97027 * c-c++-common/Warray-bounds-3.c: Remove xfail * c-c++-common/Warray-bounds-4.c: Add an expected warning. * c-c++-common/Warray-bounds-9.c: New test. * c-c++-common/Warray-bounds-10.c: New test. * g++.dg/asan/asan_test.C: Suppress expected warnings. * g++.dg/pr95768.C: Same. * g++.dg/warn/Warray-bounds-10.C: Adjust text of expected messages. * g++.dg/warn/Warray-bounds-11.C: Same. * g++.dg/warn/Warray-bounds-12.C: Same. * g++.dg/warn/Warray-bounds-13.C: Same. * g++.dg/warn/Warray-bounds-17.C: Same. * g++.dg/warn/Warray-bounds-20.C: Same. * gcc.dg/Warray-bounds-29.c: Same. * gcc.dg/Warray-bounds-30.c: Add xfail. * gcc.dg/Warray-bounds-31.c: Adjust text of expected messages. * gcc.dg/Warray-bounds-32.c: Same. * gcc.dg/Warray-bounds-52.c: Same. * gcc.dg/Warray-bounds-53.c: Same. * gcc.dg/Warray-bounds-58.c: Remove xfail. * gcc.dg/Warray-bounds-63.c: Adjust text of expected messages. * gcc.dg/Warray-bounds-66.c: Same. * gcc.dg/Warray-bounds-69.c: Same. * gcc.dg/Wstringop-overflow-34.c: Same. * gcc.dg/Wstringop-overflow-47.c: Same. * gcc.dg/Wstringop-overflow-61.c: Same. * gcc.dg/Warray-bounds-77.c: New test. * gcc.dg/Warray-bounds-78.c: New test. * gcc.dg/Warray-bounds-79.c: New test. </cut> Results regressed to (for first_bad == a110855667782dac7b674d3e328b253b3b3c919b) # reset_artifacts: -10 # true: 0 # First few build errors in logs: # 00:05:59 make[3]: [armv8l-unknown-linux-gnueabihf/bits/largefile-config.h] Error 1 (ignored) # 00:11:36 /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/snapshots/gcc.git~master/gcc/tree.h:1089:58: error: array subscript -1 is outside array bounds of ‘cpp_hashnode [89478485]’ [-Werror=array-bounds] # 00:11:37 /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/snapshots/gcc.git~master/gcc/tree.h:1089:58: error: array subscript -1 is outside array bounds of ‘cpp_hashnode [89478485]’ [-Werror=array-bounds] # 00:11:37 /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/snapshots/gcc.git~master/gcc/tree.h:1089:58: error: array subscript -1 is outside array bounds of ‘cpp_hashnode [89478485]’ [-Werror=array-bounds] # 00:12:06 make[3]: *** [cp/module.o] Error 1 # 00:17:43 make[2]: *** [all-stage2-gcc] Error 2 # 00:17:43 make[1]: *** [stage2-bubble] Error 2 # 00:17:43 make: *** [all] Error 2 from (for last_good == 6278065af07634278ba30029d92a82b089969baa) # reset_artifacts: -10 # true: 0 # build_abe bootstrap_debug: 1 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13… Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13… Build top page/logs: https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13/ Configuration details: Reproduce builds: <cut> mkdir investigate-gcc-a110855667782dac7b674d3e328b253b3b3c919b cd investigate-gcc-a110855667782dac7b674d3e328b253b3b3c919b 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_gcc-bisect-gnu-master-arm-bootstrap_debug/13… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13… --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 gcc # Reproduce first_bad build git checkout --detach a110855667782dac7b674d3e328b253b3b3c919b ../artifacts/test.sh # Reproduce last_good build git checkout --detach 6278065af07634278ba30029d92a82b089969baa ../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_gcc-bisect-gnu-master-arm-bootstrap_debug/13… Build log: https://ci.linaro.org/job/tcwg_gcc-bisect-gnu-master-arm-bootstrap_debug/13… Full commit (up to 1000 lines): <cut> commit a110855667782dac7b674d3e328b253b3b3c919b Author: Martin Sebor <msebor(a)redhat.com> Date: Wed Jul 7 14:05:25 2021 -0600 Correct handling of variable offset minus constant in -Warray-bounds [PR100137] Resolves: PR tree-optimization/100137 - -Warray-bounds false positive on varying offset plus negative PR tree-optimization/99121 - ICE in -Warray-bounds on a multidimensional PR tree-optimization/97027 - missing warning on buffer overflow storing a larger scalar into a smaller array gcc/ChangeLog: PR tree-optimization/100137 PR tree-optimization/99121 PR tree-optimization/97027 * builtins.c (access_ref::access_ref): Also set offmax. (access_ref::offset_in_range): Define new function. (access_ref::add_offset): Set offmax. (access_ref::inform_access): Handle access_none. (handle_mem_ref): Clear ostype. (compute_objsize_r): Handle ASSERT_EXPR. * builtins.h (struct access_ref): Add offmax member. * gimple-array-bounds.cc (array_bounds_checker::check_mem_ref): Use compute_objsize() and simplify. gcc/testsuite/ChangeLog: PR tree-optimization/100137 PR tree-optimization/99121 PR tree-optimization/97027 * c-c++-common/Warray-bounds-3.c: Remove xfail * c-c++-common/Warray-bounds-4.c: Add an expected warning. * c-c++-common/Warray-bounds-9.c: New test. * c-c++-common/Warray-bounds-10.c: New test. * g++.dg/asan/asan_test.C: Suppress expected warnings. * g++.dg/pr95768.C: Same. * g++.dg/warn/Warray-bounds-10.C: Adjust text of expected messages. * g++.dg/warn/Warray-bounds-11.C: Same. * g++.dg/warn/Warray-bounds-12.C: Same. * g++.dg/warn/Warray-bounds-13.C: Same. * g++.dg/warn/Warray-bounds-17.C: Same. * g++.dg/warn/Warray-bounds-20.C: Same. * gcc.dg/Warray-bounds-29.c: Same. * gcc.dg/Warray-bounds-30.c: Add xfail. * gcc.dg/Warray-bounds-31.c: Adjust text of expected messages. * gcc.dg/Warray-bounds-32.c: Same. * gcc.dg/Warray-bounds-52.c: Same. * gcc.dg/Warray-bounds-53.c: Same. * gcc.dg/Warray-bounds-58.c: Remove xfail. * gcc.dg/Warray-bounds-63.c: Adjust text of expected messages. * gcc.dg/Warray-bounds-66.c: Same. * gcc.dg/Warray-bounds-69.c: Same. * gcc.dg/Wstringop-overflow-34.c: Same. * gcc.dg/Wstringop-overflow-47.c: Same. * gcc.dg/Wstringop-overflow-61.c: Same. * gcc.dg/Warray-bounds-77.c: New test. * gcc.dg/Warray-bounds-78.c: New test. * gcc.dg/Warray-bounds-79.c: New test. --- gcc/builtins.c | 69 +++++- gcc/builtins.h | 5 + gcc/gimple-array-bounds.cc | 336 ++++---------------------- gcc/testsuite/c-c++-common/Warray-bounds-10.c | 114 +++++++++ gcc/testsuite/c-c++-common/Warray-bounds-3.c | 4 +- gcc/testsuite/c-c++-common/Warray-bounds-4.c | 12 +- gcc/testsuite/c-c++-common/Warray-bounds-9.c | 144 +++++++++++ gcc/testsuite/g++.dg/asan/asan_test.C | 2 +- gcc/testsuite/g++.dg/pr95768.C | 2 +- gcc/testsuite/g++.dg/warn/Warray-bounds-10.C | 4 +- gcc/testsuite/g++.dg/warn/Warray-bounds-11.C | 4 +- gcc/testsuite/g++.dg/warn/Warray-bounds-12.C | 4 +- gcc/testsuite/g++.dg/warn/Warray-bounds-13.C | 4 +- gcc/testsuite/g++.dg/warn/Warray-bounds-17.C | 2 +- gcc/testsuite/g++.dg/warn/Warray-bounds-20.C | 8 +- gcc/testsuite/gcc.dg/Warray-bounds-29.c | 22 +- gcc/testsuite/gcc.dg/Warray-bounds-30.c | 2 +- gcc/testsuite/gcc.dg/Warray-bounds-31.c | 8 +- gcc/testsuite/gcc.dg/Warray-bounds-32.c | 26 +- gcc/testsuite/gcc.dg/Warray-bounds-52.c | 6 +- gcc/testsuite/gcc.dg/Warray-bounds-53.c | 6 +- gcc/testsuite/gcc.dg/Warray-bounds-58.c | 2 +- gcc/testsuite/gcc.dg/Warray-bounds-63.c | 6 +- gcc/testsuite/gcc.dg/Warray-bounds-66.c | 12 +- gcc/testsuite/gcc.dg/Warray-bounds-69.c | 2 +- gcc/testsuite/gcc.dg/Warray-bounds-77.c | 135 +++++++++++ gcc/testsuite/gcc.dg/Warray-bounds-78.c | 109 +++++++++ gcc/testsuite/gcc.dg/Warray-bounds-79.c | 112 +++++++++ gcc/testsuite/gcc.dg/Wstringop-overflow-34.c | 8 +- gcc/testsuite/gcc.dg/Wstringop-overflow-47.c | 14 +- gcc/testsuite/gcc.dg/Wstringop-overflow-61.c | 4 +- 31 files changed, 817 insertions(+), 371 deletions(-) diff --git a/gcc/builtins.c b/gcc/builtins.c index e52fe4291c8..39ab139b7e1 100644 --- a/gcc/builtins.c +++ b/gcc/builtins.c @@ -206,6 +206,7 @@ access_ref::access_ref (tree bound /* = NULL_TREE */, { /* Set to valid. */ offrng[0] = offrng[1] = 0; + offmax[0] = offmax[1] = 0; /* Invalidate. */ sizrng[0] = sizrng[1] = -1; @@ -457,6 +458,21 @@ access_ref::size_remaining (offset_int *pmin /* = NULL */) const return sizrng[1] - or0; } +/* Return true if the offset and object size are in range for SIZE. */ + +bool +access_ref::offset_in_range (const offset_int &size) const +{ + if (size_remaining () < size) + return false; + + if (base0) + return offmax[0] >= 0 && offmax[1] <= sizrng[1]; + + offset_int maxoff = wi::to_offset (TYPE_MAX_VALUE (ptrdiff_type_node)); + return offmax[0] > -maxoff && offmax[1] < maxoff; +} + /* Add the range [MIN, MAX] to the offset range. For known objects (with zero-based offsets) at least one of whose offset's bounds is in range, constrain the other (or both) to the bounds of the object (i.e., zero @@ -493,6 +509,8 @@ void access_ref::add_offset (const offset_int &min, const offset_int &max) if (max >= 0) { offrng[0] = 0; + if (offmax[0] > 0) + offmax[0] = 0; return; } @@ -509,6 +527,12 @@ void access_ref::add_offset (const offset_int &min, const offset_int &max) offrng[0] = 0; } + /* Set the minimum and maximmum computed so far. */ + if (offrng[1] < 0 && offrng[1] < offmax[0]) + offmax[0] = offrng[1]; + if (offrng[0] > 0 && offrng[0] > offmax[1]) + offmax[1] = offrng[0]; + if (!base0) return; @@ -4571,23 +4595,46 @@ access_ref::inform_access (access_mode mode) const return; } + if (mode == access_read_only) + { + if (allocfn == NULL_TREE) + { + if (*offstr) + inform (loc, "at offset %s into source object %qE of size %s", + offstr, ref, sizestr); + else + inform (loc, "source object %qE of size %s", ref, sizestr); + + return; + } + + if (*offstr) + inform (loc, + "at offset %s into source object of size %s allocated by %qE", + offstr, sizestr, allocfn); + else + inform (loc, "source object of size %s allocated by %qE", + sizestr, allocfn); + return; + } + if (allocfn == NULL_TREE) { if (*offstr) - inform (loc, "at offset %s into source object %qE of size %s", + inform (loc, "at offset %s into object %qE of size %s", offstr, ref, sizestr); else - inform (loc, "source object %qE of size %s", ref, sizestr); + inform (loc, "object %qE of size %s", ref, sizestr); return; } if (*offstr) inform (loc, - "at offset %s into source object of size %s allocated by %qE", + "at offset %s into object of size %s allocated by %qE", offstr, sizestr, allocfn); else - inform (loc, "source object of size %s allocated by %qE", + inform (loc, "object of size %s allocated by %qE", sizestr, allocfn); } @@ -5433,16 +5480,16 @@ handle_mem_ref (tree mref, int ostype, access_ref *pref, if (VECTOR_TYPE_P (TREE_TYPE (mref))) { - /* Hack: Give up for MEM_REFs of vector types; those may be - synthesized from multiple assignments to consecutive data - members (see PR 93200 and 96963). + /* Hack: Handle MEM_REFs of vector types as those to complete + objects; those may be synthesized from multiple assignments + to consecutive data members (see PR 93200 and 96963). FIXME: Vectorized assignments should only be present after vectorization so this hack is only necessary after it has run and could be avoided in calls from prior passes (e.g., tree-ssa-strlen.c). FIXME: Deal with this more generally, e.g., by marking up such MEM_REFs at the time they're created. */ - return false; + ostype = 0; } tree mrefop = TREE_OPERAND (mref, 0); @@ -5796,6 +5843,12 @@ compute_objsize_r (tree ptr, int ostype, access_ref *pref, tree rhs = gimple_assign_rhs1 (stmt); + if (code == ASSERT_EXPR) + { + rhs = TREE_OPERAND (rhs, 0); + return compute_objsize_r (rhs, ostype, pref, snlim, qry); + } + if (code == POINTER_PLUS_EXPR && TREE_CODE (TREE_TYPE (rhs)) == POINTER_TYPE) { diff --git a/gcc/builtins.h b/gcc/builtins.h index e71f40c300a..a64ece3f1cd 100644 --- a/gcc/builtins.h +++ b/gcc/builtins.h @@ -222,6 +222,9 @@ struct access_ref argument to the minimum. */ offset_int size_remaining (offset_int * = NULL) const; +/* Return true if the offset and object size are in range for SIZE. */ + bool offset_in_range (const offset_int &) const; + /* Return true if *THIS is an access to a declared object. */ bool ref_declared () const { @@ -261,6 +264,8 @@ struct access_ref /* Range of byte offsets into and sizes of the object(s). */ offset_int offrng[2]; offset_int sizrng[2]; + /* The minimum and maximum offset computed. */ + offset_int offmax[2]; /* Range of the bound of the access: denotes that the access is at least BNDRNG[0] bytes but no more than BNDRNG[1]. For string functions the size of the actual access is diff --git a/gcc/gimple-array-bounds.cc b/gcc/gimple-array-bounds.cc index 13f08685f8f..83b8db9755e 100644 --- a/gcc/gimple-array-bounds.cc +++ b/gcc/gimple-array-bounds.cc @@ -414,269 +414,72 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, if (warning_suppressed_p (ref, OPT_Warray_bounds)) return false; - tree arg = TREE_OPERAND (ref, 0); - /* The constant and variable offset of the reference. */ - tree cstoff = TREE_OPERAND (ref, 1); - tree varoff = NULL_TREE; - - const offset_int maxobjsize = tree_to_shwi (max_object_size ()); - - /* The zero-based array or string constant bounds in bytes. Initially - set to [-MAXOBJSIZE - 1, MAXOBJSIZE] until a tighter bound is - determined. */ - offset_int arrbounds[2] = { -maxobjsize - 1, maxobjsize }; - - /* The minimum and maximum intermediate offset. For a reference - to be valid, not only does the final offset/subscript must be - in bounds but all intermediate offsets should be as well. - GCC may be able to deal gracefully with such out-of-bounds - offsets so the checking is only enabled at -Warray-bounds=2 - where it may help detect bugs in uses of the intermediate - offsets that could otherwise not be detectable. */ - offset_int ioff = wi::to_offset (fold_convert (ptrdiff_type_node, cstoff)); - offset_int extrema[2] = { 0, wi::abs (ioff) }; - - /* The range of the byte offset into the reference. */ - offset_int offrange[2] = { 0, 0 }; - /* The statement used to allocate the array or null. */ gimple *alloc_stmt = NULL; /* For an allocation statement, the low bound of the size range. */ offset_int minbound = 0; + /* The type and size of the access. */ + tree axstype = TREE_TYPE (ref); + offset_int axssize = 0; + if (TREE_CODE (axstype) != UNION_TYPE) + if (tree access_size = TYPE_SIZE_UNIT (axstype)) + if (TREE_CODE (access_size) == INTEGER_CST) + axssize = wi::to_offset (access_size); - /* Determine the offsets and increment OFFRANGE for the bounds of each. - The loop computes the range of the final offset for expressions such - as (A + i0 + ... + iN)[CSTOFF] where i0 through iN are SSA_NAMEs in - some range. */ - const unsigned limit = param_ssa_name_def_chain_limit; - for (unsigned n = 0; TREE_CODE (arg) == SSA_NAME && n < limit; ++n) - { - gimple *def = SSA_NAME_DEF_STMT (arg); - if (is_gimple_call (def)) - { - /* Determine the byte size of the array from an allocation call. */ - wide_int sizrng[2]; - if (gimple_call_alloc_size (def, sizrng)) - { - arrbounds[0] = 0; - arrbounds[1] = offset_int::from (sizrng[1], UNSIGNED); - minbound = offset_int::from (sizrng[0], UNSIGNED); - alloc_stmt = def; - } - break; - } - - if (gimple_nop_p (def)) - { - /* For a function argument try to determine the byte size - of the array from the current function declaratation - (e.g., attribute access or related). */ - wide_int wr[2]; - tree ref = gimple_parm_array_size (arg, wr); - if (!ref) - break; - arrbounds[0] = offset_int::from (wr[0], UNSIGNED); - arrbounds[1] = offset_int::from (wr[1], UNSIGNED); - arg = ref; - break; - } - - if (!is_gimple_assign (def)) - break; - - tree_code code = gimple_assign_rhs_code (def); - if (code == POINTER_PLUS_EXPR) - { - arg = gimple_assign_rhs1 (def); - varoff = gimple_assign_rhs2 (def); - } - else if (code == ASSERT_EXPR) - { - arg = TREE_OPERAND (gimple_assign_rhs1 (def), 0); - continue; - } - else - return false; - - /* VAROFF should always be a SSA_NAME here (and not even - INTEGER_CST) but there's no point in taking chances. */ - if (TREE_CODE (varoff) != SSA_NAME) - break; - - const value_range* const vr = get_value_range (varoff); - if (!vr || vr->undefined_p () || vr->varying_p ()) - break; - - if (!vr->constant_p ()) - break; - - if (vr->kind () == VR_RANGE) - { - offset_int min - = wi::to_offset (fold_convert (ptrdiff_type_node, vr->min ())); - offset_int max - = wi::to_offset (fold_convert (ptrdiff_type_node, vr->max ())); - if (min < max) - { - offrange[0] += min; - offrange[1] += max; - } - else - { - /* When MIN >= MAX, the offset is effectively in a union - of two ranges: [-MAXOBJSIZE -1, MAX] and [MIN, MAXOBJSIZE]. - Since there is no way to represent such a range across - additions, conservatively add [-MAXOBJSIZE -1, MAXOBJSIZE] - to OFFRANGE. */ - offrange[0] += arrbounds[0]; - offrange[1] += arrbounds[1]; - } - } - else - { - /* For an anti-range, analogously to the above, conservatively - add [-MAXOBJSIZE -1, MAXOBJSIZE] to OFFRANGE. */ - offrange[0] += arrbounds[0]; - offrange[1] += arrbounds[1]; - } - - /* Keep track of the minimum and maximum offset. */ - if (offrange[1] < 0 && offrange[1] < extrema[0]) - extrema[0] = offrange[1]; - if (offrange[0] > 0 && offrange[0] > extrema[1]) - extrema[1] = offrange[0]; - - if (offrange[0] < arrbounds[0]) - offrange[0] = arrbounds[0]; - - if (offrange[1] > arrbounds[1]) - offrange[1] = arrbounds[1]; - } + access_ref aref; + if (!compute_objsize (ref, 1, &aref, ranges)) + return false; - tree reftype = NULL_TREE; - offset_int eltsize = -1; - if (arrbounds[0] >= 0) - { - /* The byte size of the array has already been determined above - based on a pointer ARG. Set ELTSIZE to the size of the type - it points to and REFTYPE to the array with the size, rounded - down as necessary. */ - reftype = TREE_TYPE (TREE_TYPE (arg)); - if (TREE_CODE (reftype) == ARRAY_TYPE) - reftype = TREE_TYPE (reftype); - if (tree refsize = TYPE_SIZE_UNIT (reftype)) - if (TREE_CODE (refsize) == INTEGER_CST) - eltsize = wi::to_offset (refsize); - - if (eltsize < 0) - return false; + if (aref.offset_in_range (axssize)) + return false; - offset_int nelts = arrbounds[1] / eltsize; - reftype = build_printable_array_type (reftype, nelts.to_uhwi ()); - } - else if (TREE_CODE (arg) == ADDR_EXPR) + if (TREE_CODE (aref.ref) == SSA_NAME) { - arg = TREE_OPERAND (arg, 0); - if (TREE_CODE (arg) != STRING_CST - && TREE_CODE (arg) != PARM_DECL - && TREE_CODE (arg) != VAR_DECL) - return false; - - /* The type of the object being referred to. It can be an array, - string literal, or a non-array type when the MEM_REF represents - a reference/subscript via a pointer to an object that is not - an element of an array. Incomplete types are excluded as well - because their size is not known. */ - reftype = TREE_TYPE (arg); - if (POINTER_TYPE_P (reftype) - || !COMPLETE_TYPE_P (reftype) - || TREE_CODE (TYPE_SIZE_UNIT (reftype)) != INTEGER_CST) - return false; - - /* Except in declared objects, references to trailing array members - of structs and union objects are excluded because MEM_REF doesn't - make it possible to identify the member where the reference - originated. */ - if (RECORD_OR_UNION_TYPE_P (reftype) - && (!VAR_P (arg) - || (DECL_EXTERNAL (arg) && array_at_struct_end_p (ref)))) - return false; - - /* FIXME: Should this be 1 for Fortran? */ - arrbounds[0] = 0; - - if (TREE_CODE (reftype) == ARRAY_TYPE) - { - /* Set to the size of the array element (and adjust below). */ - eltsize = wi::to_offset (TYPE_SIZE_UNIT (TREE_TYPE (reftype))); - /* Use log2 of size to convert the array byte size in to its - upper bound in elements. */ - const offset_int eltsizelog2 = wi::floor_log2 (eltsize); - if (tree dom = TYPE_DOMAIN (reftype)) - { - tree bnds[] = { TYPE_MIN_VALUE (dom), TYPE_MAX_VALUE (dom) }; - if (TREE_CODE (arg) == COMPONENT_REF) - { - offset_int size = maxobjsize; - if (tree fldsize = component_ref_size (arg)) - size = wi::to_offset (fldsize); - arrbounds[1] = wi::lrshift (size, eltsizelog2); - } - else if (array_at_struct_end_p (arg) || !bnds[0] || !bnds[1]) - arrbounds[1] = wi::lrshift (maxobjsize, eltsizelog2); - else - arrbounds[1] = (wi::to_offset (bnds[1]) - wi::to_offset (bnds[0]) - + 1) * eltsize; - } - else - arrbounds[1] = wi::lrshift (maxobjsize, eltsizelog2); - - /* Determine a tighter bound of the non-array element type. */ - tree eltype = TREE_TYPE (reftype); - while (TREE_CODE (eltype) == ARRAY_TYPE) - eltype = TREE_TYPE (eltype); - eltsize = wi::to_offset (TYPE_SIZE_UNIT (eltype)); - } - else + gimple *def = SSA_NAME_DEF_STMT (aref.ref); + if (is_gimple_call (def)) { - eltsize = 1; - tree size = TYPE_SIZE_UNIT (reftype); - if (VAR_P (arg)) - if (tree initsize = DECL_SIZE_UNIT (arg)) - if (tree_int_cst_lt (size, initsize)) - size = initsize; - - arrbounds[1] = wi::to_offset (size); + /* Save the allocation call and the low bound on the size. */ + alloc_stmt = def; + minbound = aref.sizrng[0]; } } - else - return false; - - offrange[0] += ioff; - offrange[1] += ioff; + + /* The range of the byte offset into the reference. Adjusted below. */ + offset_int offrange[2] = { aref.offrng[0], aref.offrng[1] }; + + /* The type of the referenced object. */ + tree reftype = TREE_TYPE (aref.ref); + /* The size of the referenced array element. */ + offset_int eltsize = 1; + /* The byte size of the array has already been determined above + based on a pointer ARG. Set ELTSIZE to the size of the type + it points to and REFTYPE to the array with the size, rounded + down as necessary. */ + if (POINTER_TYPE_P (reftype)) + reftype = TREE_TYPE (reftype); + if (TREE_CODE (reftype) == ARRAY_TYPE) + reftype = TREE_TYPE (reftype); + if (tree refsize = TYPE_SIZE_UNIT (reftype)) + if (TREE_CODE (refsize) == INTEGER_CST) + eltsize = wi::to_offset (refsize); + + const offset_int nelts = aref.sizrng[1] / eltsize; + reftype = build_printable_array_type (reftype, nelts.to_uhwi ()); /* Compute the more permissive upper bound when IGNORE_OFF_BY_ONE is set (when taking the address of the one-past-last element of an array) but always use the stricter bound in diagnostics. */ - offset_int ubound = arrbounds[1]; + offset_int ubound = aref.sizrng[1]; if (ignore_off_by_one) ubound += eltsize; - bool warned = false; /* Set if the lower bound of the subscript is out of bounds. */ - const bool lboob = (arrbounds[0] == arrbounds[1] + const bool lboob = (aref.sizrng[1] == 0 || offrange[0] >= ubound - || offrange[1] < arrbounds[0]); + || offrange[1] < 0); /* Set if only the upper bound of the subscript is out of bounds. This can happen when using a bigger type to index into an array of a smaller type, as is common with unsigned char. */ - tree axstype = TREE_TYPE (ref); - offset_int axssize = 0; - if (TREE_CODE (axstype) != UNION_TYPE) - if (tree access_size = TYPE_SIZE_UNIT (axstype)) - if (TREE_CODE (access_size) == INTEGER_CST) - axssize = wi::to_offset (access_size); - const bool uboob = !lboob && offrange[0] + axssize > ubound; if (lboob || uboob) { @@ -689,9 +492,7 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, to compute the index to print in the diagnostic; arrays in MEM_REF don't mean anything. A type with no size like void is as good as having a size of 1. */ - tree type = TREE_TYPE (ref); - while (TREE_CODE (type) == ARRAY_TYPE) - type = TREE_TYPE (type); + tree type = strip_array_types (TREE_TYPE (ref)); if (tree size = TYPE_SIZE_UNIT (type)) { offrange[0] = offrange[0] / wi::to_offset (size); @@ -699,6 +500,7 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, } } + bool warned = false; if (lboob) { if (offrange[0] == offrange[1]) @@ -720,7 +522,7 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, /* If the memory was dynamically allocated refer to it as if it were an untyped array of bytes. */ backtype = build_array_type_nelts (unsigned_char_type_node, - arrbounds[1].to_uhwi ()); + aref.sizrng[1].to_uhwi ()); warned = warning_at (location, OPT_Warray_bounds, "array subscript %<%T[%wi]%> is partly " @@ -730,46 +532,8 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, if (warned) { - if (DECL_P (arg)) - inform (DECL_SOURCE_LOCATION (arg), "while referencing %qD", arg); - else if (alloc_stmt) - { - location_t loc = gimple_location (alloc_stmt); - if (gimple_call_builtin_p (alloc_stmt, BUILT_IN_ALLOCA_WITH_ALIGN)) - { - if (minbound == arrbounds[1]) - inform (loc, "referencing a variable length array " - "of size %wu", minbound.to_uhwi ()); - else - inform (loc, "referencing a variable length array " - "of size between %wu and %wu", - minbound.to_uhwi (), arrbounds[1].to_uhwi ()); - } - else if (tree fndecl = gimple_call_fndecl (alloc_stmt)) - { - if (minbound == arrbounds[1]) - inform (loc, "referencing an object of size %wu " - "allocated by %qD", - minbound.to_uhwi (), fndecl); - else - inform (loc, "referencing an object of size between " - "%wu and %wu allocated by %qD", - minbound.to_uhwi (), arrbounds[1].to_uhwi (), fndecl); - } - else - { - tree fntype = gimple_call_fntype (alloc_stmt); - if (minbound == arrbounds[1]) - inform (loc, "referencing an object of size %wu " - "allocated by %qT", - minbound.to_uhwi (), fntype); - else - inform (loc, "referencing an object of size between " - "%wu and %wu allocated by %qT", - minbound.to_uhwi (), arrbounds[1].to_uhwi (), fntype); - } - } - + /* TODO: Determine the access from the statement and use it. */ + aref.inform_access (access_none); suppress_warning (ref, OPT_Warray_bounds); return true; } @@ -779,9 +543,9 @@ array_bounds_checker::check_mem_ref (location_t location, tree ref, /* At level 2 check also intermediate offsets. */ int i = 0; - if (extrema[i] < -arrbounds[1] || extrema[i = 1] > ubound) + if (aref.offmax[i] < -aref.sizrng[1] || aref.offmax[i = 1] > ubound) { - HOST_WIDE_INT tmpidx = extrema[i].to_shwi () / eltsize.to_shwi (); + HOST_WIDE_INT tmpidx = aref.offmax[i].to_shwi () / eltsize.to_shwi (); if (warning_at (location, OPT_Warray_bounds, "intermediate array offset %wi is outside array bounds " diff --git a/gcc/testsuite/c-c++-common/Warray-bounds-10.c b/gcc/testsuite/c-c++-common/Warray-bounds-10.c new file mode 100644 index 00000000000..cfe9a383410 --- /dev/null +++ b/gcc/testsuite/c-c++-common/Warray-bounds-10.c @@ -0,0 +1,114 @@ +/* PR tree-optimization/99475 - bogus -Warray-bounds accessing an array + element of empty structs + { dg-do compile } + { dg-options "-O2 -Wall" } */ + +struct S +{ +#if SOME_CONFIG_MACRO + /* Suppose the contents are empty in the development configuration + but non-empty in others. Out of bounds accesses to elements of + the arrays below should be diagnosed in all configurations, + including when S is empty, even if they are folded away. */ + int member; +#endif +}; + +extern struct S sa3[3]; +extern struct S sa2_3[2][3]; +extern struct S sa3_4_5[3][4][5]; + +void sink (void*); + + +void access_sa3 (struct S s) +{ + sa3[0] = s; + sa3[1] = s; + sa3[2] = s; + sa3[3] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } +} + +void access_sa3_ptr (struct S s) +{ + struct S *p = &sa3[0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[3] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } +} + +void access_sa2_3_ptr (struct S s) +{ + struct S *p = &sa2_3[0][0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[6] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } +} + +void access_sa3_4_5_ptr (struct S s, int i) +{ + struct S *p = &sa3_4_5[0][0][0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[60] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } +} + + +void access_vla3 (struct S s, unsigned n) +{ + struct S vla3[3 < n ? 3 : n]; + + vla3[0] = s; + vla3[1] = s; + vla3[2] = s; + vla3[3] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } + + sink (vla3); +} + +void access_vla3_ptr (struct S s, unsigned n) +{ + struct S vla3[3 < n ? 3 : n]; + struct S *p = &vla3[0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[3] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } + + sink (vla3); +} + +void access_vla2_3_ptr (struct S s, unsigned n) +{ + struct S vla2_3[2 < n ? 2 : n][3 < n ? 3 : n]; + struct S *p = &vla2_3[0][0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[6] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } + + sink (vla2_3); +} + +void access_vla3_4_5_ptr (struct S s, unsigned n) +{ + struct S vla3_4_5[3 < n ? 3 : n][4 < n ? 4 : n][5 < n ? 5 : n]; + struct S *p = &vla3_4_5[0][0][0]; + + p[0] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[1] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[2] = s; // { dg-bogus "\\\[-Warray-bounds" } + p[60] = s; // { dg-warning "\\\[-Warray-bounds" "pr?????" { xfail *-*-* } } + + sink (vla3_4_5); +} + +// { dg-prune-output "empty struct has size 0 in C" } diff --git a/gcc/testsuite/c-c++-common/Warray-bounds-3.c b/gcc/testsuite/c-c++-common/Warray-bounds-3.c index aae4999ea13..3d7c7687374 100644 --- a/gcc/testsuite/c-c++-common/Warray-bounds-3.c +++ b/gcc/testsuite/c-c++-common/Warray-bounds-3.c @@ -158,7 +158,7 @@ void test_memcpy_overflow (char *d, const char *s, size_t n) but known access size is detected. This works except with small sizes that are powers of 2 due to bug . */ T (char, 1, arr + SR (DIFF_MAX - 1, DIFF_MAX), s, 1); - T (char, 1, arr + SR (DIFF_MAX - 1, DIFF_MAX), s, 2); /* { dg-warning "pointer overflow between offset \\\[\[0-9\]+, \[0-9\]+] and size 2 accessing array " "bug " { xfail non_strict_align } } */ + T (char, 1, arr + SR (DIFF_MAX - 1, DIFF_MAX), s, 2); /* { dg-warning "\\\[-Warray-bounds" } */ T (char, 1, arr + SR (DIFF_MAX - 2, DIFF_MAX), s, 3); /* { dg-warning "pointer overflow between offset \\\[\[0-9\]+, \[0-9\]+] and size 3 accessing array " "memcpy" } */ T (char, 1, arr + SR (DIFF_MAX - 4, DIFF_MAX), s, 5); /* { dg-warning "pointer overflow between offset \\\[\[0-9\]+, \[0-9\]+] and size 5 accessing array " "memcpy" } */ } @@ -178,7 +178,7 @@ void test_memcpy_bounds_memarray_range (void) TM (ma.a5, ma.a5 + i, ma.a5, 1); TM (ma.a5, ma.a5 + i, ma.a5, 3); - TM (ma.a5, ma.a5 + i, ma.a5, 5); + TM (ma.a5, ma.a5 + i, ma.a5, 5); /* { dg-warning "\\\[-Warray-bounds" } */ TM (ma.a5, ma.a5 + i, ma.a5, 7); /* diagnosed with -Warray-bounds=2 */ } diff --git a/gcc/testsuite/c-c++-common/Warray-bounds-4.c b/gcc/testsuite/c-c++-common/Warray-bounds-4.c index 22a23a12651..1f73f11943f 100644 --- a/gcc/testsuite/c-c++-common/Warray-bounds-4.c +++ b/gcc/testsuite/c-c++-common/Warray-bounds-4.c @@ -43,7 +43,17 @@ void test_memcpy_bounds_memarray_range (void) TM (ma.a5, ma.a5 + j, ma.a5, 1); TM (ma.a5, ma.a5 + j, ma.a5, 3); - TM (ma.a5, ma.a5 + j, ma.a5, 5); + + /* The copy below is invalid for two reasons: 1) it overlaps and 2) it + writes past the end of ma.a5. The warning is a little cryptic here + because the GIMPLE is: + _4 = &ma.a5 + prephitmp_14; + MEM <unsigned char[5]> [(char * {ref-all})_4] + = MEM <unsigned char[5]> [(char * {ref-all})&ma]; + and could be improved. Just verify that one is issued but not its + full text. */ + TM (ma.a5, ma.a5 + j, ma.a5, 5); /* { dg-warning "\\\[-Warray-bounds" } */ + TM (ma.a5, ma.a5 + j, ma.a5, 7); /* { dg-warning "offset \\\[5, 7] from the object at .ma. is out of the bounds of referenced subobject .$MA::$?a5. with type .char ?\\\[5]. at offset 0" } */ TM (ma.a5, ma.a5 + j, ma.a5, 9); /* { dg-warning "offset \\\[5, 9] from the object at .ma. is out of the bounds of referenced subobject .$MA::$?a5. with type .char ?\\\[5]. at offset 0" } */ } diff --git a/gcc/testsuite/c-c++-common/Warray-bounds-9.c b/gcc/testsuite/c-c++-common/Warray-bounds-9.c new file mode 100644 index 00000000000..be05775fa91 --- /dev/null +++ b/gcc/testsuite/c-c++-common/Warray-bounds-9.c @@ -0,0 +1,144 @@ +/* PR tree-optimization/99121 - ICE in -Warray-bounds on a multidimensional + VLA + { dg-do compile } + { dg-options "-O2 -Wall -ftrack-macro-expansion=0" } */ + +#define NOIPA __attribute__ ((noipa)) + +void sink (void*, ...); +#define T(a, x) sink (a, x) + + +NOIPA void a_0_n (int n) +{ + int a[0][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_n_0 (int n) +{ + int a[n][0]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + + +NOIPA void a_1_n_0 (int n) +{ + int a[1][n][0]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_1_0_n (int n) +{ + int a[1][0][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_0_1_n (int n) +{ + int a[0][1][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_0_n_1 (int n) +{ + int a[0][n][1]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_n_0_n (int n) +{ + int a[n][0][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_n_n_0 (int n) +{ + int a[n][n][0]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_0_n_n (int n) +{ + int a[0][n][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_0_0_n (int n) +{ + int a[0][0][n]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_n_0_0 (int n) +{ + int a[n][0][0]; + + sink (a); + + T (a, ((int *) a)[0]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((char *) a)[1]); // { dg-warning "\\\[-Warray-bounds" } + T (a, ((float *) a)[n]); // { dg-warning "\\\[-Warray-bounds" } +} + +NOIPA void a_n_n_n (int n) +{ + int a[n][n][n]; + + sink (a); + + T (a, ((int *) a)[-1]); // { dg-warning "\\\[-Warray-bounds" "pr99140" } + T (a, ((int *) a)[0]); + T (a, ((char *) a)[1]); + T (a, ((float *) a)[n]); +} diff --git a/gcc/testsuite/g++.dg/asan/asan_test.C b/gcc/testsuite/g++.dg/asan/asan_test.C index c62568f48ec..49933ab8e4d 100644 --- a/gcc/testsuite/g++.dg/asan/asan_test.C +++ b/gcc/testsuite/g++.dg/asan/asan_test.C @@ -2,7 +2,7 @@ // { dg-skip-if "" { *-*-* } { "*" } { "-O2" } } // { dg-skip-if "" { *-*-* } { "-flto" } { "" } } // { dg-additional-sources "asan_globals_test-wrapper.cc" } -// { dg-options "-std=c++11 -fsanitize=address -fno-builtin -Wall -Werror -Wno-alloc-size-larger-than -Wno-stringop-overflow -g -DASAN_UAR=0 -DASAN_HAS_EXCEPTIONS=1 -DASAN_HAS_BLACKLIST=0 -DSANITIZER_USE_DEJAGNU_GTEST=1 -lasan -lpthread" } +// { dg-options "-std=c++11 -fsanitize=address -fno-builtin -Wall -Werror -Wno-alloc-size-larger-than -Wno-array-bounds -Wno-stringop-overflow -g -DASAN_UAR=0 -DASAN_HAS_EXCEPTIONS=1 -DASAN_HAS_BLACKLIST=0 -DSANITIZER_USE_DEJAGNU_GTEST=1 -lasan -lpthread" } // { dg-additional-options "-ldl" { target { ! *-*-freebsd* } } } // { dg-additional-options "-DASAN_NEEDS_SEGV=1" { target { ! arm*-*-* } } } // { dg-additional-options "-DASAN_LOW_MEMORY=1 -DASAN_NEEDS_SEGV=0" { target arm*-*-* } } diff --git a/gcc/testsuite/g++.dg/pr95768.C b/gcc/testsuite/g++.dg/pr95768.C index 5e2c8c44ad0..d34d5133134 100644 --- a/gcc/testsuite/g++.dg/pr95768.C +++ b/gcc/testsuite/g++.dg/pr95768.C @@ -1,6 +1,6 @@ /* PR c++/95768 - pretty-printer ICE on -Wuninitialized with allocated storage { dg-do compile } - { dg-options "-O2 -Wall" } */ + { dg-options "-O2 -Wall -Wno-array-bounds" } */ </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 33 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *glibc* in CI configuration tcwg_cross/gnu-master-aarch64-build_cross. So far, this commit has regressed CI configurations: - tcwg_cross/gnu-master-aarch64-build_cross Culprit: <cut> commit fab3a6daf137c734e684471e197148aa058102e3 Author: Siddhesh Poyarekar <siddhesh(a)sourceware.org> Date: Wed Jul 7 13:25:49 2021 +0530 tests-exclude-mcheck: Fix typo It's tst-realloc, not tst-posix-realloc. Verified this time to ensure that the total number of tests reduced by 1. Reported-by: Stefan Liebler <stli(a)linux.ibm.com> </cut> Results regressed to (for first_bad == fab3a6daf137c734e684471e197148aa058102e3) # reset_artifacts: -10 # true: 0 # build_abe binutils: 1 # build_abe stage1: 2 # build_abe linux: 3 # First few build errors in logs: # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:499: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/start.os] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/crtn.o] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/start.o] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/crti.o] Error 1 # 00:00:14 make[2]: *** [../o-iterator.mk:9: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/check_fds.os] Error 1 # 00:00:14 make[2]: *** [../o-iterator.mk:9: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/check_fds.o] Error 1 # 00:00:14 ../sysdeps/aarch64/nptl/tls.h:91:19: error: ‘__builtin_thread_pointer’ is not supported on this target # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:485: /home/tcwg-buildslave/workspace/tcwg_gnu_1/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/libc-tls.o] Error 1 # 00:00:14 ../sysdeps/unix/sysv/linux/aarch64/sysdep.h:190:17: error: invalid register name for ‘_x0’ # 00:00:14 ../sysdeps/unix/sysv/linux/aarch64/sysdep.h:174:22: error: invalid register name for ‘_x8’ from (for last_good == 7a5db2e82fbb6c3a6e3fdae02b7166c5d0e8c7a8) # reset_artifacts: -10 # true: 0 # build_abe binutils: 1 # build_abe stage1: 2 # build_abe linux: 3 # build_abe glibc: 4 # build_abe stage2: 5 # build_abe qemu: 6 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Build top page/logs: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Configuration details: Reproduce builds: <cut> mkdir investigate-glibc-fab3a6daf137c734e684471e197148aa058102e3 cd investigate-glibc-fab3a6daf137c734e684471e197148aa058102e3 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-master-aarch64-build_cross/… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_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 glibc # Reproduce first_bad build git checkout --detach fab3a6daf137c734e684471e197148aa058102e3 ../artifacts/test.sh # Reproduce last_good build git checkout --detach 7a5db2e82fbb6c3a6e3fdae02b7166c5d0e8c7a8 ../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-master-aarch64-build_cross/… Build log: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Full commit (up to 1000 lines): <cut> commit fab3a6daf137c734e684471e197148aa058102e3 Author: Siddhesh Poyarekar <siddhesh(a)sourceware.org> Date: Wed Jul 7 13:25:49 2021 +0530 tests-exclude-mcheck: Fix typo It's tst-realloc, not tst-posix-realloc. Verified this time to ensure that the total number of tests reduced by 1. Reported-by: Stefan Liebler <stli(a)linux.ibm.com> --- malloc/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/malloc/Makefile b/malloc/Makefile index 2af1203a0a..37a9a4efab 100644 --- a/malloc/Makefile +++ b/malloc/Makefile @@ -94,7 +94,7 @@ tests-exclude-mcheck = tst-mallocstate \ tst-malloc_info \ tst-memalign \ tst-posix_memalign \ - tst-posix-realloc + tst-realloc tests-mcheck = $(filter-out $(tests-exclude-mcheck), $(tests)) </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 37 - Successful!

by ci_notify＠linaro.org

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tx1/gnu-release-aarch64-spec2k6-O2 - Build # 15 - Successful!

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 a4dfaad2e5594d871fe00a1116005e28f95d644e Author: Richard Sandiford <richard.sandiford(a)arm.com> Date: Mon Sep 30 16:20:44 2019 +0000 Remove global call sets: gcse.c This is another case in which we can conservatively treat partial kills as full kills. Again this is in principle a bug fix for TARGET_HARD_REGNO_CALL_PART_CLOBBERED targets, but in practice it probably doesn't make a difference. 2019-09-30 Richard Sandiford <richard.sandiford(a)arm.com> gcc/ * gcse.c: Include function-abi.h. (compute_hash_table_work): Use insn_callee_abi to get the ABI of the call insn target. Invalidate partially call-clobbered registers as well as fully call-clobbered ones. From-SVN: r276323 </cut> Results regressed to (for first_bad == a4dfaad2e5594d871fe00a1116005e28f95d644e) # 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-a4dfaad2e5594d871fe00a1116005e28f95d644e/results_id: 1 # 447.dealII,[.] _ZNK12SparseMatrixIdE5vmultI6VectorIdES3_EEvRT regressed by 112 from (for last_good == c1b582720a39c3cb944aaff0298f721cdd034e3f) # 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-c1b582720a39c3cb944aaff0298f721cdd034e3f/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-O2/1425 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-O2/1421 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-a4dfaad2e5594d871fe00a1116005e28f95d644e cd investigate-gcc-a4dfaad2e5594d871fe00a1116005e28f95d644e 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 a4dfaad2e5594d871fe00a1116005e28f95d644e ../artifacts/test.sh # Reproduce last_good build git checkout --detach c1b582720a39c3cb944aaff0298f721cdd034e3f ../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 a4dfaad2e5594d871fe00a1116005e28f95d644e Author: Richard Sandiford <richard.sandiford(a)arm.com> Date: Mon Sep 30 16:20:44 2019 +0000 Remove global call sets: gcse.c This is another case in which we can conservatively treat partial kills as full kills. Again this is in principle a bug fix for TARGET_HARD_REGNO_CALL_PART_CLOBBERED targets, but in practice it probably doesn't make a difference. 2019-09-30 Richard Sandiford <richard.sandiford(a)arm.com> gcc/ * gcse.c: Include function-abi.h. (compute_hash_table_work): Use insn_callee_abi to get the ABI of the call insn target. Invalidate partially call-clobbered registers as well as fully call-clobbered ones. From-SVN: r276323 --- gcc/ChangeLog | 7 +++++++ gcc/gcse.c | 10 ++++++++-- 2 files changed, 15 insertions(+), 2 deletions(-) diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 9d8b083d539..1b1c31cf009 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,10 @@ +2019-09-30 Richard Sandiford <richard.sandiford(a)arm.com> + + * gcse.c: Include function-abi.h. + (compute_hash_table_work): Use insn_callee_abi to get the ABI of + the call insn target. Invalidate partially call-clobbered + registers as well as fully call-clobbered ones. + 2019-09-30 Richard Sandiford <richard.sandiford(a)arm.com> * function.c (aggregate_value_p): Work out which ABI the diff --git a/gcc/gcse.c b/gcc/gcse.c index 9bde8619b7b..aeb59c645e1 100644 --- a/gcc/gcse.c +++ b/gcc/gcse.c @@ -160,6 +160,7 @@ along with GCC; see the file COPYING3. If not see #include "dbgcnt.h" #include "gcse.h" #include "gcse-common.h" +#include "function-abi.h" /* We support GCSE via Partial Redundancy Elimination. PRE optimizations are a superset of those done by classic GCSE. @@ -1528,8 +1529,13 @@ compute_hash_table_work (struct gcse_hash_table_d *table) if (CALL_P (insn)) { hard_reg_set_iterator hrsi; - EXECUTE_IF_SET_IN_HARD_REG_SET (regs_invalidated_by_call, - 0, regno, hrsi) + + /* We don't track modes of hard registers, so we need + to be conservative and assume that partial kills + are full kills. */ + HARD_REG_SET callee_clobbers + = insn_callee_abi (insn).full_and_partial_reg_clobbers (); + EXECUTE_IF_SET_IN_HARD_REG_SET (callee_clobbers, 0, regno, hrsi) record_last_reg_set_info (insn, regno); if (! RTL_CONST_OR_PURE_CALL_P (insn) </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 35 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *glibc* in CI configuration tcwg_cross/gnu-master-aarch64-build_cross. So far, this commit has regressed CI configurations: - tcwg_cross/gnu-master-aarch64-build_cross Culprit: <cut> commit 6ec6c77867af4ddfec7323e0ac6ede89effca852 Author: liuhongt <hongtao.liu(a)intel.com> Date: Wed Jul 7 10:07:59 2021 +0800 soft-fp: Add __extendhfsf2/__extendhfdf2, __truncsfhf2/__truncdfhf2, __eqhf2/__nehf2 1. Add __extendhfdf2/__extendhfsf2 to return an IEEE half converted to IEEE double/single. 2. Add __truncdfhf2/__extendsfhf2 to truncate IEEE double/single into IEEE half. 3. Add __eqhf2/__nehf2 to return 0 if a == b and a,b are not NAN, otherwise return 1. These are needed by x86 _Float16: https://software.intel.com/content/www/us/en/develop/download/intel-avx512-… support in GCC. </cut> Results regressed to (for first_bad == 6ec6c77867af4ddfec7323e0ac6ede89effca852) # reset_artifacts: -10 # true: 0 # build_abe binutils: 1 # build_abe stage1: 2 # build_abe linux: 3 # First few build errors in logs: # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/start.o] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:499: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/start.os] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/crtn.o] Error 1 # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:483: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/crti.o] Error 1 # 00:00:14 make[2]: *** [../o-iterator.mk:9: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/check_fds.o] Error 1 # 00:00:14 make[2]: *** [../o-iterator.mk:9: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/check_fds.os] Error 1 # 00:00:14 ../sysdeps/aarch64/nptl/tls.h:91:19: error: ‘__builtin_thread_pointer’ is not supported on this target # 00:00:14 make[2]: *** [/home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/sysd-rules:485: /home/tcwg-buildslave/workspace/tcwg_gnu_0/abe/builds/x86_64-pc-linux-gnu/aarch64-linux-gnu/glibc-glibc.git~master/csu/libc-tls.o] Error 1 # 00:00:14 ../sysdeps/unix/sysv/linux/aarch64/sysdep.h:190:17: error: invalid register name for ‘_x0’ # 00:00:14 ../sysdeps/unix/sysv/linux/aarch64/sysdep.h:174:22: error: invalid register name for ‘_x8’ from (for last_good == b46cfcef3f7dab8e90729c6616ac1420f2642245) # reset_artifacts: -10 # true: 0 # build_abe binutils: 1 # build_abe stage1: 2 # build_abe linux: 3 # build_abe glibc: 4 # build_abe stage2: 5 # build_abe qemu: 6 Artifacts of last_good build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Artifacts of first_bad build: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Build top page/logs: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Configuration details: Reproduce builds: <cut> mkdir investigate-glibc-6ec6c77867af4ddfec7323e0ac6ede89effca852 cd investigate-glibc-6ec6c77867af4ddfec7323e0ac6ede89effca852 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-master-aarch64-build_cross/… --fail curl -o artifacts/manifests/build-parameters.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… --fail curl -o artifacts/test.sh https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_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 glibc # Reproduce first_bad build git checkout --detach 6ec6c77867af4ddfec7323e0ac6ede89effca852 ../artifacts/test.sh # Reproduce last_good build git checkout --detach b46cfcef3f7dab8e90729c6616ac1420f2642245 ../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-master-aarch64-build_cross/… Build log: https://ci.linaro.org/job/tcwg_cross-bisect-gnu-master-aarch64-build_cross/… Full commit (up to 1000 lines): <cut> commit 6ec6c77867af4ddfec7323e0ac6ede89effca852 Author: liuhongt <hongtao.liu(a)intel.com> Date: Wed Jul 7 10:07:59 2021 +0800 soft-fp: Add __extendhfsf2/__extendhfdf2, __truncsfhf2/__truncdfhf2, __eqhf2/__nehf2 1. Add __extendhfdf2/__extendhfsf2 to return an IEEE half converted to IEEE double/single. 2. Add __truncdfhf2/__extendsfhf2 to truncate IEEE double/single into IEEE half. 3. Add __eqhf2/__nehf2 to return 0 if a == b and a,b are not NAN, otherwise return 1. These are needed by x86 _Float16: https://software.intel.com/content/www/us/en/develop/download/intel-avx512-… support in GCC. --- soft-fp/eqhf2.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++ soft-fp/extendhfdf2.c | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++ soft-fp/extendhfsf2.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++ soft-fp/half.h | 2 ++ soft-fp/truncdfhf2.c | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++ soft-fp/truncsfhf2.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++ 6 files changed, 253 insertions(+) diff --git a/soft-fp/eqhf2.c b/soft-fp/eqhf2.c new file mode 100644 index 0000000000..6d6634e5c5 --- /dev/null +++ b/soft-fp/eqhf2.c @@ -0,0 +1,49 @@ +/* Software floating-point emulation. + Return 0 iff a == b, 1 otherwise + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#include "soft-fp.h" +#include "half.h" + +CMPtype +__eqhf2 (HFtype a, HFtype b) +{ + FP_DECL_EX; + FP_DECL_H (A); + FP_DECL_H (B); + CMPtype r; + + FP_INIT_EXCEPTIONS; + FP_UNPACK_RAW_H (A, a); + FP_UNPACK_RAW_H (B, b); + FP_CMP_EQ_H (r, A, B, 1); + FP_HANDLE_EXCEPTIONS; + + return r; +} + +strong_alias (__eqhf2, __nehf2); diff --git a/soft-fp/extendhfdf2.c b/soft-fp/extendhfdf2.c new file mode 100644 index 0000000000..337ba791d4 --- /dev/null +++ b/soft-fp/extendhfdf2.c @@ -0,0 +1,53 @@ +/* Software floating-point emulation. + Return an IEEE half converted to IEEE double + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#define FP_NO_EXACT_UNDERFLOW +#include "soft-fp.h" +#include "half.h" +#include "double.h" + +DFtype +__extendhfdf2 (HFtype a) +{ + FP_DECL_EX; + FP_DECL_H (A); + FP_DECL_D (R); + DFtype r; + + FP_INIT_EXCEPTIONS; + FP_UNPACK_RAW_H (A, a); +#if _FP_W_TYPE_SIZE < _FP_FRACBITS_D + FP_EXTEND (D, H, 2, 1, R, A); +#else + FP_EXTEND (D, H, 1, 1, R, A); +#endif + FP_PACK_RAW_D (r, R); + FP_HANDLE_EXCEPTIONS; + + return r; +} diff --git a/soft-fp/extendhfsf2.c b/soft-fp/extendhfsf2.c new file mode 100644 index 0000000000..a02f46d9a9 --- /dev/null +++ b/soft-fp/extendhfsf2.c @@ -0,0 +1,49 @@ +/* Software floating-point emulation. + Return an IEEE half converted to IEEE single + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#define FP_NO_EXACT_UNDERFLOW +#include "soft-fp.h" +#include "half.h" +#include "single.h" + +SFtype +__extendhfsf2 (HFtype a) +{ + FP_DECL_EX; + FP_DECL_H (A); + FP_DECL_S (R); + SFtype r; + + FP_INIT_EXCEPTIONS; + FP_UNPACK_RAW_H (A, a); + FP_EXTEND (S, H, 1, 1, R, A); + FP_PACK_RAW_S (r, R); + FP_HANDLE_EXCEPTIONS; + + return r; +} diff --git a/soft-fp/half.h b/soft-fp/half.h index ca2c1fad03..70af2712fc 100644 --- a/soft-fp/half.h +++ b/soft-fp/half.h @@ -167,4 +167,6 @@ union _FP_UNION_H #define _FP_FRAC_HIGH_RAW_H(X) _FP_FRAC_HIGH_1 (X) #define _FP_FRAC_HIGH_DW_H(X) _FP_FRAC_HIGH_1 (X) +#define FP_CMP_EQ_H(r, X, Y, ex) _FP_CMP_EQ (H, 1, (r), X, Y, (ex)) + #endif /* !SOFT_FP_HALF_H */ diff --git a/soft-fp/truncdfhf2.c b/soft-fp/truncdfhf2.c new file mode 100644 index 0000000000..8bcb278769 --- /dev/null +++ b/soft-fp/truncdfhf2.c @@ -0,0 +1,52 @@ +/* Software floating-point emulation. + Truncate IEEE double into IEEE half. + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#include "soft-fp.h" +#include "half.h" +#include "double.h" + +HFtype +__truncdfhf2 (DFtype a) +{ + FP_DECL_EX; + FP_DECL_D (A); + FP_DECL_H (R); + HFtype r; + + FP_INIT_ROUNDMODE; + FP_UNPACK_SEMIRAW_D (A, a); +#if _FP_W_TYPE_SIZE < _FP_FRACBITS_D + FP_TRUNC (H, D, 1, 2, R, A); +#else + FP_TRUNC (H, D, 1, 1, R, A); +#endif + FP_PACK_SEMIRAW_H (r, R); + FP_HANDLE_EXCEPTIONS; + + return r; +} diff --git a/soft-fp/truncsfhf2.c b/soft-fp/truncsfhf2.c new file mode 100644 index 0000000000..25bee29f7f --- /dev/null +++ b/soft-fp/truncsfhf2.c @@ -0,0 +1,48 @@ +/* Software floating-point emulation. + Truncate IEEE single into IEEE half. + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#include "soft-fp.h" +#include "half.h" +#include "single.h" + +HFtype +__truncsfhf2 (SFtype a) +{ + FP_DECL_EX; + FP_DECL_S (A); + FP_DECL_H (R); + HFtype r; + + FP_INIT_ROUNDMODE; + FP_UNPACK_SEMIRAW_S (A, a); + FP_TRUNC (H, S, 1, 1, R, A); + FP_PACK_SEMIRAW_H (r, R); + FP_HANDLE_EXCEPTIONS; + + return r; +} </cut>

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 21 - Successful!

by ci_notify＠linaro.org

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 38 - Successful!

by ci_notify＠linaro.org

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_cross/gnu-master-aarch64-build_cross - Build # 36 - Successful!

by ci_notify＠linaro.org

3 years, 4 months

[CI-NOTIFY]: TCWG Bisect tcwg_bmk_tx1/llvm-release-aarch64-spec2k6-Os - Build # 6 - Successful!

by ci_notify＠linaro.org

Successfully identified regression in *llvm* in CI configuration tcwg_bmk_llvm_tx1/llvm-release-aarch64-spec2k6-Os. So far, this commit has regressed CI configurations: - tcwg_bmk_llvm_tx1/llvm-release-aarch64-spec2k6-Os Culprit: <cut> commit 87d7757bbe14fed420092071ded3430072053316 Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin(a)amd.com> Date: Tue Nov 24 10:42:43 2020 -0800 [SLP] Control maximum vectorization factor from TTI D82227 has added a proper check to limit PHI vectorization to the maximum vector register size. That unfortunately resulted in at least a couple of regressions on SystemZ and x86. This change reverts PHI handling from D82227 and replaces it with a more general check in SLPVectorizerPass::tryToVectorizeList(). Moved to tryToVectorizeList() it allows to restart vectorization if initial chunk fails. However, this function is more general and handles not only PHI but everything which SLP handles. If vectorization factor would be limited to maximum vector register size it would limit much more vectorization than before leading to further regressions. Therefore a new TTI callback getMaximumVF() is added with the default 0 to preserve current behavior and limit nothing. Then targets can decide what is better for them. The callback gets ElementSize just like a similar getMinimumVF() function and the main opcode of the chain. The latter is to avoid regressions at least on the AMDGPU. We can have loads and stores up to 128 bit wide, and <2 x 16> bit vector math on some subtargets, where the rest shall not be vectorized. I.e. we need to differentiate based on the element size and operation itself. Differential Revision: https://reviews.llvm.org/D92059 </cut> Results regressed to (for first_bad == 87d7757bbe14fed420092071ded3430072053316) # 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 -Os -- artifacts/build-87d7757bbe14fed420092071ded3430072053316/results_id: 1 # 444.namd,namd_base.default regressed by 102 from (for last_good == c21df2a79c268d1e0f467ec25a1ec7cb4aff5dfb) # 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 -Os -- artifacts/build-c21df2a79c268d1e0f467ec25a1ec7cb4aff5dfb/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-Os/1415 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-Os/1413 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-87d7757bbe14fed420092071ded3430072053316 cd investigate-llvm-87d7757bbe14fed420092071ded3430072053316 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 87d7757bbe14fed420092071ded3430072053316 ../artifacts/test.sh # Reproduce last_good build git checkout --detach c21df2a79c268d1e0f467ec25a1ec7cb4aff5dfb ../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 87d7757bbe14fed420092071ded3430072053316 Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin(a)amd.com> Date: Tue Nov 24 10:42:43 2020 -0800 [SLP] Control maximum vectorization factor from TTI D82227 has added a proper check to limit PHI vectorization to the maximum vector register size. That unfortunately resulted in at least a couple of regressions on SystemZ and x86. This change reverts PHI handling from D82227 and replaces it with a more general check in SLPVectorizerPass::tryToVectorizeList(). Moved to tryToVectorizeList() it allows to restart vectorization if initial chunk fails. However, this function is more general and handles not only PHI but everything which SLP handles. If vectorization factor would be limited to maximum vector register size it would limit much more vectorization than before leading to further regressions. Therefore a new TTI callback getMaximumVF() is added with the default 0 to preserve current behavior and limit nothing. Then targets can decide what is better for them. The callback gets ElementSize just like a similar getMinimumVF() function and the main opcode of the chain. The latter is to avoid regressions at least on the AMDGPU. We can have loads and stores up to 128 bit wide, and <2 x 16> bit vector math on some subtargets, where the rest shall not be vectorized. I.e. we need to differentiate based on the element size and operation itself. Differential Revision: https://reviews.llvm.org/D92059 --- llvm/include/llvm/Analysis/TargetTransformInfo.h | 9 ++++ .../llvm/Analysis/TargetTransformInfoImpl.h | 2 + llvm/lib/Analysis/TargetTransformInfo.cpp | 5 ++ .../Target/AMDGPU/AMDGPUTargetTransformInfo.cpp | 6 +++ llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h | 1 + llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp | 33 +++++-------- .../Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll | 55 +++++++++++++++++----- llvm/test/Transforms/SLPVectorizer/AMDGPU/round.ll | 4 +- .../Transforms/SLPVectorizer/slp-max-phi-size.ll | 17 +++++-- 9 files changed, 95 insertions(+), 37 deletions(-) diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h index 3ba77c9a8dc9..b9b9df35cdb0 100644 --- a/llvm/include/llvm/Analysis/TargetTransformInfo.h +++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h @@ -941,6 +941,11 @@ public: /// applies when shouldMaximizeVectorBandwidth returns true. unsigned getMinimumVF(unsigned ElemWidth) const; + /// \return The maximum vectorization factor for types of given element + /// bit width and opcode, or 0 if there is no maximum VF. + /// Currently only used by the SLP vectorizer. + unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const; + /// \return True if it should be considered for address type promotion. /// \p AllowPromotionWithoutCommonHeader Set true if promoting \p I is /// profitable without finding other extensions fed by the same input. @@ -1498,6 +1503,7 @@ public: virtual unsigned getMinVectorRegisterBitWidth() = 0; virtual bool shouldMaximizeVectorBandwidth(bool OptSize) const = 0; virtual unsigned getMinimumVF(unsigned ElemWidth) const = 0; + virtual unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const = 0; virtual bool shouldConsiderAddressTypePromotion( const Instruction &I, bool &AllowPromotionWithoutCommonHeader) = 0; virtual unsigned getCacheLineSize() const = 0; @@ -1917,6 +1923,9 @@ public: unsigned getMinimumVF(unsigned ElemWidth) const override { return Impl.getMinimumVF(ElemWidth); } + unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const override { + return Impl.getMaximumVF(ElemWidth, Opcode); + } bool shouldConsiderAddressTypePromotion( const Instruction &I, bool &AllowPromotionWithoutCommonHeader) override { return Impl.shouldConsiderAddressTypePromotion( diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h index b4847844cd0e..2c206094ac4a 100644 --- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h +++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h @@ -356,6 +356,8 @@ public: unsigned getMinimumVF(unsigned ElemWidth) const { return 0; } + unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const { return 0; } + bool shouldConsiderAddressTypePromotion(const Instruction &I, bool &AllowPromotionWithoutCommonHeader) { diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp index f327d0cad426..086a212ee65b 100644 --- a/llvm/lib/Analysis/TargetTransformInfo.cpp +++ b/llvm/lib/Analysis/TargetTransformInfo.cpp @@ -635,6 +635,11 @@ unsigned TargetTransformInfo::getMinimumVF(unsigned ElemWidth) const { return TTIImpl->getMinimumVF(ElemWidth); } +unsigned TargetTransformInfo::getMaximumVF(unsigned ElemWidth, + unsigned Opcode) const { + return TTIImpl->getMaximumVF(ElemWidth, Opcode); +} + bool TargetTransformInfo::shouldConsiderAddressTypePromotion( const Instruction &I, bool &AllowPromotionWithoutCommonHeader) const { return TTIImpl->shouldConsiderAddressTypePromotion( diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp index ab83419183ee..2cfb0299eea9 100644 --- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp +++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp @@ -288,6 +288,12 @@ unsigned GCNTTIImpl::getMinVectorRegisterBitWidth() const { return 32; } +unsigned GCNTTIImpl::getMaximumVF(unsigned ElemWidth, unsigned Opcode) const { + if (Opcode == Instruction::Load || Opcode == Instruction::Store) + return 32 * 4 / ElemWidth; + return (ElemWidth == 16 && ST->has16BitInsts()) ? 2 : 1; +} + unsigned GCNTTIImpl::getLoadVectorFactor(unsigned VF, unsigned LoadSize, unsigned ChainSizeInBytes, VectorType *VecTy) const { diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h index 05a054ac30ff..1785a2df2eba 100644 --- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h +++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h @@ -170,6 +170,7 @@ public: unsigned getNumberOfRegisters(unsigned RCID) const; unsigned getRegisterBitWidth(bool Vector) const; unsigned getMinVectorRegisterBitWidth() const; + unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const; unsigned getLoadVectorFactor(unsigned VF, unsigned LoadSize, unsigned ChainSizeInBytes, VectorType *VecTy) const; diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp index e1c1c6edf08c..cda43521e9ba 100644 --- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp +++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp @@ -126,6 +126,10 @@ static cl::opt<int> MaxVectorRegSizeOption("slp-max-reg-size", cl::init(128), cl::Hidden, cl::desc("Attempt to vectorize for this register size in bits")); +static cl::opt<unsigned> +MaxVFOption("slp-max-vf", cl::init(0), cl::Hidden, + cl::desc("Maximum SLP vectorization factor (0=unlimited)")); + static cl::opt<int> MaxStoreLookup("slp-max-store-lookup", cl::init(32), cl::Hidden, cl::desc("Maximum depth of the lookup for consecutive stores.")); @@ -741,6 +745,12 @@ public: return MinVecRegSize; } + unsigned getMaximumVF(unsigned ElemWidth, unsigned Opcode) const { + unsigned MaxVF = MaxVFOption.getNumOccurrences() ? + MaxVFOption : TTI->getMaximumVF(ElemWidth, Opcode); + return MaxVF ? MaxVF : UINT_MAX; + } + /// Check if homogeneous aggregate is isomorphic to some VectorType. /// Accepts homogeneous multidimensional aggregate of scalars/vectors like /// {[4 x i16], [4 x i16]}, { <2 x float>, <2 x float> }, @@ -6191,6 +6201,7 @@ bool SLPVectorizerPass::tryToVectorizeList(ArrayRef<Value *> VL, BoUpSLP &R, unsigned Sz = R.getVectorElementSize(I0); unsigned MinVF = std::max(2U, R.getMinVecRegSize() / Sz); unsigned MaxVF = std::max<unsigned>(PowerOf2Floor(VL.size()), MinVF); + MaxVF = std::min(R.getMaximumVF(Sz, S.getOpcode()), MaxVF); if (MaxVF < 2) { R.getORE()->emit([&]() { return OptimizationRemarkMissed(SV_NAME, "SmallVF", I0) @@ -7633,7 +7644,6 @@ bool SLPVectorizerPass::vectorizeChainsInBlock(BasicBlock *BB, BoUpSLP &R) { bool Changed = false; SmallVector<Value *, 4> Incoming; SmallPtrSet<Value *, 16> VisitedInstrs; - unsigned MaxVecRegSize = R.getMaxVecRegSize(); bool HaveVectorizedPhiNodes = true; while (HaveVectorizedPhiNodes) { @@ -7660,27 +7670,8 @@ bool SLPVectorizerPass::vectorizeChainsInBlock(BasicBlock *BB, BoUpSLP &R) { // Look for the next elements with the same type. SmallVector<Value *, 4>::iterator SameTypeIt = IncIt; - Type *EltTy = (*IncIt)->getType(); - - assert(EltTy->isSized() && - "Instructions should all be sized at this point"); - TypeSize EltTS = DL->getTypeSizeInBits(EltTy); - if (EltTS.isScalable()) { - // For now, just ignore vectorizing scalable types. - ++IncIt; - continue; - } - - unsigned EltSize = EltTS.getFixedSize(); - unsigned MaxNumElts = MaxVecRegSize / EltSize; - if (MaxNumElts < 2) { - ++IncIt; - continue; - } - while (SameTypeIt != E && - (*SameTypeIt)->getType() == EltTy && - static_cast<unsigned>(SameTypeIt - IncIt) < MaxNumElts) { + (*SameTypeIt)->getType() == (*IncIt)->getType()) { VisitedInstrs.insert(*SameTypeIt); ++SameTypeIt; } diff --git a/llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll b/llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll index 514cf0749086..c7985b90f012 100644 --- a/llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll +++ b/llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll @@ -123,12 +123,18 @@ bb: ret <2 x i16> %ins.1 } -; FIXME: Should not vectorize define <2 x i32> @uadd_sat_v2i32(<2 x i32> %arg0, <2 x i32> %arg1) { ; GCN-LABEL: @uadd_sat_v2i32( ; GCN-NEXT: bb: -; GCN-NEXT: [[TMP0:%.*]] = call <2 x i32> @llvm.uadd.sat.v2i32(<2 x i32> [[ARG0:%.*]], <2 x i32> [[ARG1:%.*]]) -; GCN-NEXT: ret <2 x i32> [[TMP0]] +; GCN-NEXT: [[ARG0_0:%.*]] = extractelement <2 x i32> [[ARG0:%.*]], i64 0 +; GCN-NEXT: [[ARG0_1:%.*]] = extractelement <2 x i32> [[ARG0]], i64 1 +; GCN-NEXT: [[ARG1_0:%.*]] = extractelement <2 x i32> [[ARG1:%.*]], i64 0 +; GCN-NEXT: [[ARG1_1:%.*]] = extractelement <2 x i32> [[ARG1]], i64 1 +; GCN-NEXT: [[ADD_0:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[ARG0_0]], i32 [[ARG1_0]]) +; GCN-NEXT: [[ADD_1:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[ARG0_1]], i32 [[ARG1_1]]) +; GCN-NEXT: [[INS_0:%.*]] = insertelement <2 x i32> undef, i32 [[ADD_0]], i64 0 +; GCN-NEXT: [[INS_1:%.*]] = insertelement <2 x i32> [[INS_0]], i32 [[ADD_1]], i64 1 +; GCN-NEXT: ret <2 x i32> [[INS_1]] ; bb: %arg0.0 = extractelement <2 x i32> %arg0, i64 0 @@ -145,8 +151,15 @@ bb: define <2 x i32> @usub_sat_v2i32(<2 x i32> %arg0, <2 x i32> %arg1) { ; GCN-LABEL: @usub_sat_v2i32( ; GCN-NEXT: bb: -; GCN-NEXT: [[TMP0:%.*]] = call <2 x i32> @llvm.usub.sat.v2i32(<2 x i32> [[ARG0:%.*]], <2 x i32> [[ARG1:%.*]]) -; GCN-NEXT: ret <2 x i32> [[TMP0]] +; GCN-NEXT: [[ARG0_0:%.*]] = extractelement <2 x i32> [[ARG0:%.*]], i64 0 +; GCN-NEXT: [[ARG0_1:%.*]] = extractelement <2 x i32> [[ARG0]], i64 1 +; GCN-NEXT: [[ARG1_0:%.*]] = extractelement <2 x i32> [[ARG1:%.*]], i64 0 +; GCN-NEXT: [[ARG1_1:%.*]] = extractelement <2 x i32> [[ARG1]], i64 1 +; GCN-NEXT: [[ADD_0:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[ARG0_0]], i32 [[ARG1_0]]) +; GCN-NEXT: [[ADD_1:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[ARG0_1]], i32 [[ARG1_1]]) +; GCN-NEXT: [[INS_0:%.*]] = insertelement <2 x i32> undef, i32 [[ADD_0]], i64 0 +; GCN-NEXT: [[INS_1:%.*]] = insertelement <2 x i32> [[INS_0]], i32 [[ADD_1]], i64 1 +; GCN-NEXT: ret <2 x i32> [[INS_1]] ; bb: %arg0.0 = extractelement <2 x i32> %arg0, i64 0 @@ -163,8 +176,15 @@ bb: define <2 x i32> @sadd_sat_v2i32(<2 x i32> %arg0, <2 x i32> %arg1) { ; GCN-LABEL: @sadd_sat_v2i32( ; GCN-NEXT: bb: -; GCN-NEXT: [[TMP0:%.*]] = call <2 x i32> @llvm.sadd.sat.v2i32(<2 x i32> [[ARG0:%.*]], <2 x i32> [[ARG1:%.*]]) -; GCN-NEXT: ret <2 x i32> [[TMP0]] +; GCN-NEXT: [[ARG0_0:%.*]] = extractelement <2 x i32> [[ARG0:%.*]], i64 0 +; GCN-NEXT: [[ARG0_1:%.*]] = extractelement <2 x i32> [[ARG0]], i64 1 +; GCN-NEXT: [[ARG1_0:%.*]] = extractelement <2 x i32> [[ARG1:%.*]], i64 0 +; GCN-NEXT: [[ARG1_1:%.*]] = extractelement <2 x i32> [[ARG1]], i64 1 +; GCN-NEXT: [[ADD_0:%.*]] = call i32 @llvm.sadd.sat.i32(i32 [[ARG0_0]], i32 [[ARG1_0]]) +; GCN-NEXT: [[ADD_1:%.*]] = call i32 @llvm.sadd.sat.i32(i32 [[ARG0_1]], i32 [[ARG1_1]]) +; GCN-NEXT: [[INS_0:%.*]] = insertelement <2 x i32> undef, i32 [[ADD_0]], i64 0 +; GCN-NEXT: [[INS_1:%.*]] = insertelement <2 x i32> [[INS_0]], i32 [[ADD_1]], i64 1 +; GCN-NEXT: ret <2 x i32> [[INS_1]] ; bb: %arg0.0 = extractelement <2 x i32> %arg0, i64 0 @@ -181,8 +201,15 @@ bb: define <2 x i32> @ssub_sat_v2i32(<2 x i32> %arg0, <2 x i32> %arg1) { ; GCN-LABEL: @ssub_sat_v2i32( ; GCN-NEXT: bb: -; GCN-NEXT: [[TMP0:%.*]] = call <2 x i32> @llvm.ssub.sat.v2i32(<2 x i32> [[ARG0:%.*]], <2 x i32> [[ARG1:%.*]]) -; GCN-NEXT: ret <2 x i32> [[TMP0]] +; GCN-NEXT: [[ARG0_0:%.*]] = extractelement <2 x i32> [[ARG0:%.*]], i64 0 +; GCN-NEXT: [[ARG0_1:%.*]] = extractelement <2 x i32> [[ARG0]], i64 1 +; GCN-NEXT: [[ARG1_0:%.*]] = extractelement <2 x i32> [[ARG1:%.*]], i64 0 +; GCN-NEXT: [[ARG1_1:%.*]] = extractelement <2 x i32> [[ARG1]], i64 1 +; GCN-NEXT: [[ADD_0:%.*]] = call i32 @llvm.ssub.sat.i32(i32 [[ARG0_0]], i32 [[ARG1_0]]) +; GCN-NEXT: [[ADD_1:%.*]] = call i32 @llvm.ssub.sat.i32(i32 [[ARG0_1]], i32 [[ARG1_1]]) +; GCN-NEXT: [[INS_0:%.*]] = insertelement <2 x i32> undef, i32 [[ADD_0]], i64 0 +; GCN-NEXT: [[INS_1:%.*]] = insertelement <2 x i32> [[INS_0]], i32 [[ADD_1]], i64 1 +; GCN-NEXT: ret <2 x i32> [[INS_1]] ; bb: %arg0.0 = extractelement <2 x i32> %arg0, i64 0 @@ -267,8 +294,14 @@ define <4 x i16> @uadd_sat_v4i16(<4 x i16> %arg0, <4 x i16> %arg1) { ; ; GFX8-LABEL: @uadd_sat_v4i16( ; GFX8-NEXT: bb: -; GFX8-NEXT: [[TMP0:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0:%.*]], <4 x i16> [[ARG1:%.*]]) -; GFX8-NEXT: ret <4 x i16> [[TMP0]] +; GFX8-NEXT: [[TMP0:%.*]] = shufflevector <4 x i16> [[ARG0:%.*]], <4 x i16> undef, <2 x i32> <i32 0, i32 1> +; GFX8-NEXT: [[TMP1:%.*]] = shufflevector <4 x i16> [[ARG1:%.*]], <4 x i16> undef, <2 x i32> <i32 0, i32 1> +; GFX8-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]]) +; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[ARG0]], <4 x i16> undef, <2 x i32> <i32 2, i32 3> +; GFX8-NEXT: [[TMP4:%.*]] = shufflevector <4 x i16> [[ARG1]], <4 x i16> undef, <2 x i32> <i32 2, i32 3> +; GFX8-NEXT: [[TMP5:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP3]], <2 x i16> [[TMP4]]) +; GFX8-NEXT: [[INS_3:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3> +; GFX8-NEXT: ret <4 x i16> [[INS_3]] ; bb: %arg0.0 = extractelement <4 x i16> %arg0, i64 0 diff --git a/llvm/test/Transforms/SLPVectorizer/AMDGPU/round.ll b/llvm/test/Transforms/SLPVectorizer/AMDGPU/round.ll index 84d25fe79a27..7a8e08de4138 100644 --- a/llvm/test/Transforms/SLPVectorizer/AMDGPU/round.ll +++ b/llvm/test/Transforms/SLPVectorizer/AMDGPU/round.ll @@ -18,9 +18,9 @@ bb: ret <2 x half> %tmp5 } -; TODO: Should probably not really be vectorizing this ; GCN-LABEL: @round_v2f32( -; GCN: call <2 x float> @llvm.round.v2f32 +; GCN: call float @llvm.round.f32( +; GCN: call float @llvm.round.f32( define <2 x float> @round_v2f32(<2 x float> %arg) { bb: %tmp = extractelement <2 x float> %arg, i64 0 diff --git a/llvm/test/Transforms/SLPVectorizer/slp-max-phi-size.ll b/llvm/test/Transforms/SLPVectorizer/slp-max-phi-size.ll index 466e83d0260a..ce33efc951fc 100644 --- a/llvm/test/Transforms/SLPVectorizer/slp-max-phi-size.ll +++ b/llvm/test/Transforms/SLPVectorizer/slp-max-phi-size.ll @@ -1,7 +1,12 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt -slp-vectorizer -S -slp-max-reg-size=32 < %s | FileCheck -check-prefix=MAX32 %s -; RUN: opt -slp-vectorizer -S -slp-max-reg-size=256 < %s | FileCheck -check-prefix=MAX256 %s -; RUN: opt -slp-vectorizer -S -slp-max-reg-size=1024 < %s | FileCheck -check-prefix=MAX1024 %s +; RUN: opt -slp-vectorizer -S -slp-max-vf=1 < %s | FileCheck -check-prefix=MAX32 %s +; RUN: opt -slp-vectorizer -S -slp-max-vf=8 < %s | FileCheck -check-prefix=MAX256 %s +; RUN: opt -slp-vectorizer -S -slp-max-vf=32 < %s | FileCheck -check-prefix=MAX1024 %s +; RUN: opt -slp-vectorizer -S < %s | FileCheck -check-prefix=MAX1024 %s + +; Make sure we do not vectorize to create PHI wider than requested. +; On AMDGPU target wider vectorization will result in a higher register pressure, +; spilling, or even inability to allocate registers. define void @phi_float32(half %hval, float %fval) { ; MAX32-LABEL: @phi_float32( @@ -120,6 +125,7 @@ define void @phi_float32(half %hval, float %fval) { ; MAX32-NEXT: [[PHI30:%.*]] = phi float [ [[I63]], [[BB3]] ], [ [[FVAL]], [[BB4]] ], [ [[FVAL]], [[BB5]] ], [ [[FVAL]], [[BB1]] ] ; MAX32-NEXT: [[PHI31:%.*]] = phi float [ [[I65]], [[BB3]] ], [ [[FVAL]], [[BB4]] ], [ [[I65]], [[BB5]] ], [ [[I65]], [[BB1]] ] ; MAX32-NEXT: [[PHI32:%.*]] = phi float [ [[I67]], [[BB3]] ], [ [[I67]], [[BB4]] ], [ [[FVAL]], [[BB5]] ], [ [[I67]], [[BB1]] ] +; MAX32-NEXT: store float [[PHI31]], float* undef, align 4 ; MAX32-NEXT: ret void ; ; MAX256-LABEL: @phi_float32( @@ -296,6 +302,8 @@ define void @phi_float32(half %hval, float %fval) { ; MAX256-NEXT: [[TMP154:%.*]] = phi <8 x float> [ [[TMP28]], [[BB3]] ], [ [[TMP93]], [[BB4]] ], [ [[TMP132]], [[BB5]] ], [ [[TMP54]], [[BB1]] ] ; MAX256-NEXT: [[TMP155:%.*]] = phi <8 x float> [ [[TMP30]], [[BB3]] ], [ [[TMP103]], [[BB4]] ], [ [[TMP142]], [[BB5]] ], [ [[TMP64]], [[BB1]] ] ; MAX256-NEXT: [[TMP156:%.*]] = phi <8 x float> [ [[TMP32]], [[BB3]] ], [ [[TMP113]], [[BB4]] ], [ [[TMP152]], [[BB5]] ], [ [[TMP74]], [[BB1]] ] +; MAX256-NEXT: [[TMP157:%.*]] = extractelement <8 x float> [[TMP156]], i32 6 +; MAX256-NEXT: store float [[TMP157]], float* undef, align 4 ; MAX256-NEXT: ret void ; ; MAX1024-LABEL: @phi_float32( @@ -481,6 +489,8 @@ define void @phi_float32(half %hval, float %fval) { ; MAX1024-NEXT: br label [[BB2]] ; MAX1024: bb2: ; MAX1024-NEXT: [[TMP165:%.*]] = phi <32 x float> [ [[TMP38]], [[BB3]] ], [ [[TMP125]], [[BB4]] ], [ [[TMP164]], [[BB5]] ], [ [[TMP86]], [[BB1]] ] +; MAX1024-NEXT: [[TMP166:%.*]] = extractelement <32 x float> [[TMP165]], i32 30 +; MAX1024-NEXT: store float [[TMP166]], float* undef, align 4 ; MAX1024-NEXT: ret void ; bb: @@ -603,5 +613,6 @@ bb2: %phi30 = phi float [ %i63, %bb3 ], [ %fval, %bb4 ], [ %fval, %bb5 ], [ %fval, %bb1 ] %phi31 = phi float [ %i65, %bb3 ], [ %fval, %bb4 ], [ %i65, %bb5 ], [ %i65, %bb1 ] %phi32 = phi float [ %i67, %bb3 ], [ %i67, %bb4 ], [ %fval, %bb5 ], [ %i67, %bb1 ] + store float %phi31, float* undef ret void } </cut>

3 years, 4 months

← Newer
1
2
3
4
5
6
7
8
9
10
Older →

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

linaro-toolchain July 2021