Denys Vlasenko wrote:
On Mon, 2008-12-08 at 16:05 +0100, Denys Vlasenko wrote:
I built 2.6.28-rc7 + utrace and sigstep.c indeed fails.
Here is the shortened version of sigstep.c.
Looks like when we PTRACE_SINGLESTEP after raise (SIGUSR2),
the child is left to run freely.
(Sorry about GNU indent style... utrace-tests requires that)
Can you confirm that it also fails for you on 2.6.28-rc7 + utrace
but works on vanilla 2.6.28-rc7?
Please also test simplified version of the second test,
see below.
It appears that PTRACE_SINGLESTEP is a culprit here too,
no need to play with signals as in previous test,
it happens without any signals.
It also does not require many single-steps or forks,
for me it happens on the very first iteration.
(I removed the part which makes lots of forks).
--
vda
Thanks for your wonderful simplification.
I occasionally found these two test cases. Sorry, don't remember where
they were from. Maybe they had been updated by somebody from original
texts. And I also made some modifications. The current code is mixed
version which maybe confuse people. I am a little lazy ^-^.
The two tests works fine on 2.6.27.5-41.fc9.x86_64 as you mentioned.
Test passed on 2.6.28-rc7:
[EMAIL PROTECTED] uname -a
Linux 2.6.28-rc7 #28 SMP Mon Dec 1
22:06:56 KST 2008 i686 i686 i386 GNU/Linux
[EMAIL PROTECTED] cat /proc/kallsyms |grep utrace
[EMAIL PROTECTED] ./sigstep
[EMAIL PROTECTED] echo $?
0
[EMAIL PROTECTED] ./multi-step-same-time
[EMAIL PROTECTED] echo $?
0
Only failed on 2.6.28-rc7+ latest utrace.
Regards,
Wenji
