Messages in this thread | | | From | Roland McGrath <> | Subject | Re: [PATCHSET RFC] ptrace,signal: clean transition between STOPPED and TRACED | Date | Tue, 1 Feb 2011 21:39:31 -0800 (PST) |
| |
> Also, the first test of xcheck seems to enter infinite loop.
It's a test for race-condition bugs, where the original test was to run the infinite loop until it crashed. As now written, it runs the loop until $TESTTIME seconds have passed, defaulting to 10. With various buggy kernels in various configurations and on various machines, the time it takes to get to a crash has varied widely.
> I see. Yeah, if there are users which expect /proc/pid/status to be > certain value, we can either emulate it or delay TRACED transition to > the next PTRACE call *after* ATTACH/wait(2) sequence, but I think both > are quite ugly and would like to avoid if at all possible.
I agree. We just need to be more clearly sure about userland requirements before we make changes. I hope the answer is that nothing other than these synthetic test cases actually relies on which one /proc/pid/status reports. (That is, if they do check it, they only notice the "T".)
Thanks, Roland
| |