https://bugs.kde.org/show_bug.cgi?id=385409
--- Comment #57 from Andreas Arnez <ar...@linux.ibm.com> --- Created attachment 114932 --> https://bugs.kde.org/attachment.cgi?id=114932&action=edit Implement VLL with aligned loads This fixes complaints from memcheck in cases where VLL is used for strlen, strcpy, etc. This is achieved by implementing a VLL that stays within a 16-byte aligned chunk with an aligned load instead. Then memcheck will apply its "partial-loads-OK" heuristic and suppress complaints about reading past the allocated string. One problem still remains with VLBB (vector load to block boundary), which assures that no page boundary is crossed and is often used for loading the initial bytes of a string. VLBB may be safely used to load from an unaligned address, but memcheck's heuristic does not apply in this case. Splitting the load into two aligned 16-byte loads wouldn't help either, because the second load would not necessarily overlap with the allocated string at all. I wonder whether memcheck's heuristic could be adjusted such that it allows for unaligned 16-byte loads as long as no page boundary is crossed? Or are there any other suggestions? -- You are receiving this mail because: You are watching all bug changes.
