Comment 4 for bug 1527636

Revision history for this message
Ben Franksen (bfrk) wrote : Re: [Bug 1527636] Re: Client exit delays caused by reverse name lookup

On 12/18/2015 05:22 PM, mdavidsaver wrote:
>> I'll do a little test to see if this works.
>
> With glibc gethostbyaddr(), and probably all resolver functions handle
> EINTR internally by retrying. So they are completely impervious to
> signals. Thanks glibc...

Götz Pfeiffer (here at HZB) has written a patch that allows forceful
thread termination (under Posix only), using pthread_cancel and
pthread_join rather than signals. This seems to work. Judging from past
discussions I was certain that proposing something like this for
addition to base would be hopeless. In case I was mistaken, I am sure
Götz would be delighted to provide more details including his patch. It
could at least serve as a basis for further discussion.

(For reference, let me include the bug tracker URL here
https://bugs.launchpad.net/bugs/1527636 since I doubt that Götz is
subscribed to the launchpad tracker).

Cheers
Ben
--
"Make it so they have to reboot after every typo." ― Scott Adams