Thanks for looking at this.
I apologize for the tone of that last update I posted on gerrithub as
well, last week wasn't great for me and I probably spent too long on
this patchset.


William Allen Simpson wrote on Sun, Jul 30, 2017 at 08:54:32PM -0400:
> That is, it was a superficial bandage over the real problem.  I'd added
> logging for multiple calls, but never saw it during testing.

logging unfortunately often serializes events :(

> What could be done instead is call xp_destroy (as is done now), but
> each transport would unhook the epoll, then spawn another task to do
> the actual free later.
> 
> That task could spin checking for the reference counter to be zero.
> Then free.
> 
> Essentially, we'd use the worker locking/queuing system as a
> serialization mechanism.  Because the new task would be added to the
> tail of the queue, it's very likely to come after any epoll task.
> 
> We really don't care that we're adding task switches at the close of
> the connection, in exchange for low overhead during the connection.

Sounds good to me, maybe requeue itself instead of spinning? I'm not
sure if it can get stuck waiting for some later work item in the queue.


I also realize I haven't given nearly enough information for the last
stack/crash I have on nfs 4.1, can you confirm you cannot reproduce it
either ? I don't remember what I was doing exactly but it crashed right
away, could have been on a mount or basic ls/touch commands within the
export, on a simple VFS export built with ASAN (-DSANITIZE_ADDRESS=ON)

If you can't hit it I will provide more info there as well but it would
surprise me as there isn't much specific at this point.
-- 
Dominique Martinet

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Nfs-ganesha-devel mailing list
Nfs-ganesha-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs-ganesha-devel

Reply via email to