Hi @simonmichal I'm only aware of a problem when the AsyncSocket is re-opened and reassigned to another poller.
I just re-pushed a change to specify relaxed memory order on the std::atomic introduced by this patch; it took a while to convince myself this was ok (not that this makes any difference in generated code, in this case and on intel). I see we generally use relaxed order with std::atomics, and a comment in the XrdSys XrdSysRAtomic utility class contains a a suggestion to use a mutex to some cases, where a stronger guarantee is needed, explicit.
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1