https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #10 from Yilong Li ---
(In reply to Mark Thomas from comment #9)
> I'm trying to reproduce these results. Could you share how you configured
> the tests to run with RV-Predict.
The general
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #11 from Yilong Li ---
(In reply to Yilong Li from comment #10)
> (In reply to Mark Thomas from comment #9)
> > I'm trying to reproduce these results. Could you share how you configured
> > the
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #9 from Mark Thomas ---
I'm trying to reproduce these results. Could you share how you configured the
tests to run with RV-Predict.
--
You are receiving this mail because:
You are the assignee for the bug.
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #8 from Remy Maucherat ---
Yes, I think that's too many questions for BZ. NIO2 does not allow concurrency,
and in addition to the SSL engine, concurrency would also corrupt all the
buffers used. The socket wrapper
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #4 from Mark Thomas ---
Given bug 57799 which suggests that there might be a concurrency issue with
NIO2 and TLS it might be worth taking a closer look at the stack traces in this
report to see if they offer any
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #3 from Remy Maucherat ---
No idea, it sounds too specific to the NIO2 behavior. Sure the completion
handler is going to be called directly by the main thread pool, but only once
per IO operation, and since a read
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #5 from Remy Maucherat ---
Yes, but I don't see how a concurrent read is possible, and this is a "non
blocking" read, that other bug is supposed to be using blocking IO.
--
You are receiving this mail because:
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #7 from Yilong Li ---
(In reply to Remy Maucherat from comment #3)
> No idea, it sounds too specific to the NIO2 behavior. Sure the completion
> handler is going to be called directly by the main
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #6 from Yilong Li ---
(In reply to Remy Maucherat from comment #3)
> No idea, it sounds too specific to the NIO2 behavior. Sure the completion
> handler is going to be called directly by the main
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
Remy Maucherat changed:
What|Removed |Added
Status|NEW |RESOLVED
https://bz.apache.org/bugzilla/show_bug.cgi?id=58321
--- Comment #2 from Yilong Li ---
(In reply to Remy Maucherat from comment #1)
> This looks invalid as the IO operations are protected by pending flags
> and/or semaphores. I get it your tool seems to use
11 matches
Mail list logo