https://bugs.kde.org/show_bug.cgi?id=356174
--- Comment #2 from Daniel Trebbien <dtrebb...@gmail.com> --- Created attachment 95853 --> https://bugs.kde.org/attachment.cgi?id=95853&action=edit Fix handling of an 'Hc-1' packet Looking through the packets logged just before the segfault, I noticed that the response to an 'Hc-1' packet was not correct. For example: < 8> send packet: $Hc-1#09 < 7> read packet: $E01#a6 This is due to two issues: 1. Within server_main(), `strtoul (&own_buf[2], NULL, 16);' actually expands to a call to VG_(strtoull16). However, unlike the standard C library strtoul() routine, VG_(strtoull16) did not support a minus sign. 2. gdb_id_to_thread_id() does not handle the case where the gdb thread ID is -1 (which means "all threads" in this case: https://sourceware.org/gdb/current/onlinedocs/gdb/Packets.html#Packets ). The attached patch adds minus sign and overflow handling to VG_(strtoull10) and VG_(strtoull16) so that their behavior is closer to the standard strtoull() function: http://pubs.opengroup.org/onlinepubs/000095399/functions/strtoull.html The patch also modifies server_main() to handle the case where `gdb_id' is `(unsigned long)-1'. -- You are receiving this mail because: You are watching all bug changes.
