utrace_do_stop() sets utrace-stopped but leaves the tracee in TASK_STOPPED
state. This means SIGCONT can wake up the tracee and fool the tracer.
Signed-off-by: Oleg Nesterov o...@redhat.com
---
kernel/utrace.c |3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
---
utrace_do_stop() sets utrace-stopped but leaves the tracee in TASK_STOPPED
state. This means SIGCONT can wake up the tracee and fool the tracer.
IMHO this one belongs before 2/4 and 3/4 and all the comments changed to
reflect that SIGKILL is the only case. But the incremental order really
On 07/28, Roland McGrath wrote:
utrace_do_stop() sets utrace-stopped but leaves the tracee in TASK_STOPPED
state. This means SIGCONT can wake up the tracee and fool the tracer.
IMHO this one belongs before 2/4 and 3/4 and all the comments changed to
reflect that SIGKILL is the only case.
OK, I'll send the first patch upstream, then I re-send 2-4 with updated
comments to you.
Ok, good. When I get those I'll merge those upstream-bound ones into the
tracehook git branch (now just signals-tracehook_notify_jctl-change.patch)
and merge the utrace changes into the utrace branch.