Opened 15 years ago
Closed 15 years ago
#5296 closed bug (fixed)
KDL in team_get_controlling_tty()
Reported by: | axeld | Owned by: | bonefish |
---|---|---|---|
Priority: | blocker | Milestone: | R1 |
Component: | System/Kernel | Version: | R1/Development |
Keywords: | Cc: | ||
Blocked By: | Blocking: | ||
Platform: | All |
Description
I just got a NULL pointer access in team_get_controlling_tty(). Unfortunately, I didn't have any time to look into it, ie. I don't know if the team's group or the group's session pointer was NULL.
In any case, the bug appeared at the end of a gdb session.
Ingo, maybe you remember a bit how your session/group implementation is supposed to work? I've assigned you the bug for now.
Attachments (1)
Change History (6)
by , 15 years ago
Attachment: | team_group.cpp added |
---|
comment:1 by , 15 years ago
Just ran into it again. This time I had the pleasure being able to look into it ;-)
Anyway, the team's group member is already NULL. What happens is that the team is already destructed while its threads aren't yet.
I've attached a short test program that one can use to reproduce the issue 100%: when run it launches another thread and waits for user input. Then you have to attach GDB to that second thread, and quit GDB. Say welcome to KDL.
comment:2 by , 15 years ago
This isn't possibly related to the recent thread_do_exit_work() changes is it?
comment:3 by , 15 years ago
Even if it is, that wouldn't matter, as this is definitely a kernel bug that deserves to be fixed.
test program