Update of /cvsroot/ltp/ltp/runtest
In directory ddv4jf1.ch3.sourceforge.com:/tmp/cvs-serv13931/ltp/runtest
Garret added a new testcase for ptrace() syscall: The following calls ptrace(PTRACE_TRACEME, 0) for all signals and validates that all signals apart from SIGKILL are properly intercepted and WIFSTOPPED remains true, and SIGKILL is properly treated as WIFSIGNALED() && WTERMSIG() returns SIGKILL.I looked at it again this weekend, and my criterion for kill (..., 0) was incorrect. The newer version passes, and tests correct criterion, as per the manpage:
1. This test app helped unroot another issue with our custom Linux platform here at Cisco,
2. This test application does function properly on a Gentoo Linux based x86_64 / 2.6.30 kernel however, so it's an issue with our OS platform,
I will add this application to testcases/kernel/syscalls/ptrace, if someone else sees the value in it. I agree that additional error checking could be added for the calls to kill(2)... I just whipped this up in 45 mins after finding this strange behavior on our platform. Signed-off-by: Garrett Cooper <yanegomi@...>.
RCS file: /cvsroot/ltp/ltp/runtest/syscalls,v
retrieving revision 1.300
retrieving revision 1.301
diff -C2 -d -r1.300 -r1.301
*** syscalls 13 Oct 2009 14:07:57 -0000 1.300
--- syscalls 29 Oct 2009 18:15:37 -0000 1.301
*** 732,735 ****
--- 732,736 ----
+ ptrace05 ptrace05
# Broken test; See: testcases/kernel/syscalls/ptrace/Makefile for more details.