github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
Geth MoarVM: 23daabadc9 | (Stefan Seifert)++ | src/io/signals.c
Fix memory corruption after cancelling signal

According to libuv docs, memory MUST only be released once uv_close on a handle runs the supplied callback. Otherwise, uv_run's cleanup will overwrite memory that has already been freed. Therefore we must ensure that we hold on to the async task (and it's SignalInfo structure) until uv_close calls the callback and only then remove the taks from the work queue.
Fixes github.com/rakudo/rakudo/issues/3299		 08:47
nine It's certainly worth to look at other parts of IO as they may have the same issue with cancellation as ^^^! 08:50
(that means someone else should do it, as I'm still on vacation and should get some 2 hours of sleep before going on www.darkskyproject.co.nz/tours/sum...xperience/ ) 08:51
nwc10 sleep well 08:52
lizmat nine++ 08:53
jnthn Interesting: twitter.com/damageboy/status/11947...3708057608 10:06
(The whole thread, but the tweet I linked is the reason I mention it here)
oh, wait, I maybe linked the thread...anyway, the bit about JITs. :) 10:07
lizmat yikes 12:37
Guest13443 nine: how was the Observatory? 13:42
Geth MoarVM: 1b2f72e5e9 | (Stefan Seifert)++ | src/io/signals.c
Remove dead code

Accidentally committed leftover from debugging. Spotted by patzim++		 15:02
lizmat jnthn brrt: git.kernel.org/pub/scm/linux/kerne...d29c7a96f3 16:03
seems related ?
nine Guest13443: absolutely fantastic! 19:26
brrt jnthn: I saw it.... I think 'suffer the consequences' is a bit much, but yeah 19:37
I'd prefer testing the real impact over rushing to implement the 'fix'
(which would be difficult to do given how dynasm works)

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!