helgrind: fix real (and imaginary) race conditions #7208
Conversation
|
Does this mean we can/should rip out lockdep? I'm pretty sure it's broken anyway. |
|
lockdep is orders of magnitude faster than helgrind. All librbd unit tests run with lockdep enabled -- helgrind didn't catch any librbd bugs not already previously fixed by lockdep reports. |
| @@ -67,6 +70,7 @@ SimpleMessenger::~SimpleMessenger() | |||
| assert(!did_bind); // either we didn't bind or we shut down the Accepter | |||
| assert(rank_pipe.empty()); // we don't have any running Pipes. | |||
| assert(!reaper_started); // the reaper thread is stopped | |||
| ceph_spin_destroy(&global_seq_lock); | |||
There was a problem hiding this comment.
could you help to do this in async messenger too?
There was a problem hiding this comment.
These were exposed running librbd test cases -- it would probably open a whole new set of issues to helgrind ceph-mon/ceph-osd (with or without running w/ async messenger).
dillaman
force-pushed
the
wip-14163-jewel
branch
from
afa8564
to
09dea70
Compare
| @@ -1342,6 +1344,7 @@ void Pipe::fault(bool onread) | |||
| return; | |||
| } | |||
|
|
|||
| recv_reset(); | |||
| shutdown_socket(); | |||
There was a problem hiding this comment.
is it safe to leave out recv_reset() for the other two callsites of shutdown_socket()? seems like this could use an explanation in its own commit
There was a problem hiding this comment.
The other two sites were for shutdown cases -- so no worries if you leave state behind. I can move to a new commit.
Provides necessary support for annotating helgrind false-positives. Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Fixes: ceph#14163 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Derived classes should be responsible for registration to avoid accessing a partially constructed object. Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
dillaman
force-pushed
the
wip-14163-jewel
branch
from
09dea70
to
da9a470
Compare
helgrind: fix real (and imaginary) race conditions Reviewed-by: Josh Durgin <jdurgin@redhat.com>
|
Added issues for remaining potential bugs discovered: |
No description provided.