Hi! The optimization prepared in PR103798 ("memchr with a (small) constant string should be expanded inline", c6cf555a88f8ffb8ec85c2b94fe656ab217260ea) caused a regression for avr-elf, pru-elf and rl78-elf: .../gcc/configure --prefix=... --enable-werror-always --enable-languages=all --disable-gcov --disable-shared --disable-threads --target=pru-elf --without-headers [...] make V=1 all-gcc [...] /usr/lib/gcc-snapshot/bin/g++ -fno-PIE -c -g -O2 -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE -fno-exceptions -fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings -W cast-qual -Wmissing-format-attribute -Woverloaded-virtual -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Werror -fno-common -DHAVE_CONFIG_H -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. -I../../gcc/gcc/../include -I../../gcc/gcc/../libcpp/include -I../../gcc/gcc/../libcody -I../../gcc/gcc/../libdecnumber -I../../gcc/gcc/../libdecnumber/dpd -I../libdecnumber -I../../gcc/gcc/../libbacktrace -o tree-ssa-forwprop.o -MT tree-ssa-forwprop.o -MMD -MP -MF ./.deps/tree-ssa-forwprop.TPo ../../gcc/gcc/tree-ssa-forwprop.cc ../../gcc/gcc/tree-ssa-forwprop.cc: In function 'bool simplify_builtin_call(gimple_stmt_iterator*, tree)': ../../gcc/gcc/tree-ssa-forwprop.cc:1258:42: error: array subscript 1 is outside array bounds of 'tree_node* [1]' [-Werror=array-bounds] 1258 | op[i - 1] = fold_convert_loc (loc, boolean_type_node, | ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~ 1259 | fold_build2_loc (loc, | ~~~~~~~~~~~~~~~~~~~~~ 1260 | BIT_IOR_EXPR, | ~~~~~~~~~~~~~ 1261 | boolean_type_node, | ~~~~~~~~~~~~~~~~~~ 1262 | op[i - 1], | ~~~~~~~~~~ 1263 | op[i])); | ~~~~~~~ In file included from ../../gcc/gcc/system.h:707, from ../../gcc/gcc/tree-ssa-forwprop.cc:21: ../../gcc/gcc/../include/libiberty.h:733:36: note: at offset 8 into object of size [0, 8] allocated by '__builtin_alloca' 733 | # define alloca(x) __builtin_alloca(x) | ~~~~~~~~~~~~~~~~^~~ ../../gcc/gcc/../include/libiberty.h:365:40: note: in expansion of macro 'alloca' 365 | #define XALLOCAVEC(T, N) ((T *) alloca (sizeof (T) * (N))) | ^~~~~~ ../../gcc/gcc/tree-ssa-forwprop.cc:1250:22: note: in expansion of macro 'XALLOCAVEC' 1250 | tree *op = XALLOCAVEC (tree, isize); | ^~~~~~~~~~ cc1plus: all warnings being treated as errors make[1]: *** [Makefile:1146: tree-ssa-forwprop.o] Error 1 (Last confirmed at c2c3e4f6698925c8c969d8525677fbfe98f78909.)
Looks like gcc can't figure out that isize can't be 0 even if there was a check for integer_zerop (size) before hand. There must be a way to add an assert there to allow gcc to figure that out.
Obviously this only happens because you use --enable-werror-always
Sure, just a warning, but still there (as of 31303c9b5bab200754cdb7ef8cd91ae4918f3018) and affecting three targets. Maybe we get that understood/fixed?
I am just going to test: diff --git a/gcc/tree-ssa-forwprop.cc b/gcc/tree-ssa-forwprop.cc index 06f19868ade..0326e6733e8 100644 --- a/gcc/tree-ssa-forwprop.cc +++ b/gcc/tree-ssa-forwprop.cc @@ -1231,14 +1231,14 @@ simplify_builtin_call (gimple_stmt_iterator *gsi_p, tree callee2) tree size = gimple_call_arg (stmt2, 2); /* Size must be a constant which is <= UNITS_PER_WORD and <= the string length. */ - if (TREE_CODE (size) != INTEGER_CST || integer_zerop (size)) + if (TREE_CODE (size) != INTEGER_CST) break; if (!tree_fits_uhwi_p (size)) break; unsigned HOST_WIDE_INT sz = tree_to_uhwi (size); - if (sz > UNITS_PER_WORD || sz >= slen) + if (sz == 0 || sz > UNITS_PER_WORD || sz >= slen) break; tree ch = gimple_call_arg (stmt2, 1); This does not change the behavior of the code at all, just makes it obvious sz cannot be 0 and therefore isize won't be 0 either.
Patch posted: https://gcc.gnu.org/pipermail/gcc-patches/2023-May/618762.html
The trunk branch has been updated by Andrew Pinski <pinskia@gcc.gnu.org>: https://gcc.gnu.org/g:f65af1eeef670f2c249b1896726ef57bbf65fe2f commit r14-937-gf65af1eeef670f2c249b1896726ef57bbf65fe2f Author: Andrew Pinski <apinski@marvell.com> Date: Tue May 16 14:34:05 2023 -0700 Fix PR 106900: array-bounds warning inside simplify_builtin_call The problem here is that VRP cannot figure out isize could not be 0 due to using integer_zerop. This patch removes the use of integer_zerop and instead checks for 0 directly after converting the tree to an unsigned HOST_WIDE_INT. This allows VRP to figure out isize is not 0 and `isize - 1` will always be >= 0. This patch is just to avoid the warning that GCC could produce sometimes and does not change any code generation or even VRP. OK? Bootstrapped and tested on x86_64-linux-gnu with no regressions. gcc/ChangeLog: * tree-ssa-forwprop.cc (simplify_builtin_call): Check against 0 instead of calling integer_zerop.
Fixed on the trunk; not really worth backporting since it is only an issue with --enable-werror-always which almost nobody uses.
Thanks a lot! I scheduled builds for the three affected targets (from my target list.) The box is quite loaded right now (and a few jobs a before those three), so I guess it'll take a few hours.
All three target configurations reported a successful build. Thanks!