Source: strace Version: 4.10-1 Severity: serious Hi,
strace 4.10 FTBFS on a bunch of arches due to testsuite failures. I've had a look at a few of them on the arches I have access too. Hopefully some of this helps a bit. https://buildd.debian.org/status/package.php?p=strace http://buildd.debian-ports.org/status/package.php?p=strace&suite=sid FAIL: bexecve.test ----- This test fails if the executable being run does not match the native architecture. This only affects platforms which build strace64. On these platforms the strace invocation in the test outputs something like this: > [ Process PID=11033 runs in 32 bit mode. ] > Process 11033 detached The test case doesn't handle the first line and dies when it sees it. FAIL: net-yy.test, unix-yy.test ----- The awk scripts these tests use depend on gawk features so they need a 'check_prog gawk' check. It looks like this might be fixed upstream after lots of stuff was refactored. FAIL: stat32-v.test ----- tests/stat.c tries to generate a stat syscall to test the output of strace. The code contains this (NR_stat is defined on mips): > #ifdef NR_stat > if (sizeof(stb.st_size) > 4) > return 77; > assert(syscall(NR_stat, av[1], &stb) == 0); > #else > assert(stat(av[1], &stb) == 0); > #endif Unfortunately on mips, this generates an indirect 'syscall' syscall instead of directly calling the 'stat' syscall. The indirect syscall is printed completely differently by strace. One one hand it might be a good idea for strace to print indirect syscalls as if they were called directly. Otherwise you can workaround this by forcing stat from glibc to be called. Thanks, James
signature.asc
Description: This is a digitally signed message part
