When one of the string args to strncmp has alignment of less than 8 bytes, the builtin expansion code generates a runtime check to see if the string might cross a 4k boundary. However the runtime check looks for at least the length N before the next 4k boundary and then the code generated may load a doubleword, crossing the 4k boundary if N is less than 8.
Author: acsawdey Date: Mon Feb 13 16:00:22 2017 New Revision: 245392 URL: https://gcc.gnu.org/viewcvs?rev=245392&root=gcc&view=rev Log: 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com> PR target/79449 * gcc.dg/strncmp-2.c: New. Test strncmp and memcmp builtin expansion for reading beyond a 4k boundary. 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com> PR target/79449 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime boundary crossing check and subsequent code generation agree. Added: trunk/gcc/testsuite/gcc.dg/strncmp-2.c Modified: trunk/gcc/ChangeLog trunk/gcc/config/rs6000/rs6000.c trunk/gcc/testsuite/ChangeLog
Fix checked in to 245392.
Author: acsawdey Date: Mon Mar 27 15:40:20 2017 New Revision: 246504 URL: https://gcc.gnu.org/viewcvs?rev=246504&root=gcc&view=rev Log: 2017-03-27 Aaron Sawdey <acsawdey@linux.vnet.ibm.com> Backport from trunk PR target/79449 PR target/79170 * gcc.dg/strncmp-2.c: New. Test strncmp and memcmp builtin expansion for reading beyond a 4k boundary. * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime boundary crossing check and subsequent code generation agree. * gcc.dg/memcmp-1.c: Improved to catch failures seen in PR 79170. * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed. (setb_unsigned) New pattern for setb with CCUNS. * config/rs6000/rs6000.c (expand_block_compare): Use a different subfc./subfe sequence to avoid overflow problems. Generate a shorter sequence with cmpld/setb for power9. * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern for generating subfc. instruction. (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence now uses this instruction. * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg. * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand strcmp. Fix bug where comparison didn't stop with zero byte. Fix case where N arg is SIZE_MAX. * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare. (cmpstrsi): Add pattern. * gcc.dg/memcmp-1.c: New. * gcc.dg/strncmp-1.c: New. * config/rs6000/rs6000-protos.h (expand_strn_compare): Declare. * config/rs6000/rs6000.md (UNSPEC_CMPB): New unspec. (cmpb<mode>3): pattern for generating cmpb. (cmpstrnsi): pattern to expand strncmp (). * config/rs6000/rs6000.opt (mstring-compare-inline-limit): Add a new target option for controlling how much code inline expansion of strncmp() will be allowed to generate. * config/rs6000/rs6000.c (expand_strncmp_align_check): generate code for runtime page crossing check of strncmp () args. (expand_strn_compare): Function to do builtin expansion of strncmp (). * config/i386/i386.md (cmpstrnsi): New test to bail out if neither string input is a string constant. * builtins.c (expand_builtin_strncmp): Attempt expansion of strncmp via cmpstrnsi even if neither string is constant. Modified: branches/ibm/gcc-6-branch/gcc/ChangeLog.ibm branches/ibm/gcc-6-branch/gcc/builtins.c branches/ibm/gcc-6-branch/gcc/config/i386/i386.md branches/ibm/gcc-6-branch/gcc/config/rs6000/altivec.md branches/ibm/gcc-6-branch/gcc/config/rs6000/rs6000-protos.h branches/ibm/gcc-6-branch/gcc/config/rs6000/rs6000.c branches/ibm/gcc-6-branch/gcc/config/rs6000/rs6000.md branches/ibm/gcc-6-branch/gcc/config/rs6000/rs6000.opt