Possible GnuTLS-related hard hang

Possible GnuTLS-related hard hang

[Lars Ingebrigtsen]

My laptop went into sleep mode, and when it came back, Emacs hung hard.
It may have changed its IP address or something.

strace showed the following:

Process 12143 attached with 3 threads - interrupt to quit
[pid 12143] sendto(10, "\340\204s\374\204\22\231\26M;?\307\220=^\206T\371Q\374o\324\246i\31\256\31$\314\315\367\362"..., 136, 0, NULL, 0 <unfinished ...>
[pid 12145] restart_syscall(<... resuming interrupted call ...> <unfinished ...>
[pid 12144] restart_syscall(<... resuming interrupted call ...> <unfinished ...>
[pid 12143] <... sendto resumed> )      = -1 EAGAIN (Resource temporarily unavailable)
[pid 12143] sendto(10, "\340\204s\374\204\22\231\26M;?\307\220=^\206T\371Q\374o\324\246i\31\256\31$\314\315\367\362"..., 136, 0, NULL, 0) = -1 EAGAIN (Resource temporarily unavailable)
[pid 12143] sendto(10, "\340\204s\374\204\22\231\26M;?\307\220=^\206T\371Q\374o\324\246i\31\256\31$\314\315\367\362"..., 136, 0, NULL, 0) = -1 EAGAIN (Resource temporarily unavailable)
[pid 12143] sendto(10, "\340\204s\374\204\22\231\26M;?\307\220=^\206T\371Q\374o\324\246i\31\256\31$\314\315\367\362"..., 136, 0, NULL, 0) = -1 EAGAIN (Resource temporarily unavailable)

larsi@rusty:~/pgnus$ ls -l /proc/12143/fd/10 
lrwx------ 1 larsi larsi 64 2012-01-27 13:14 /proc/12143/fd/10 -> socket:[586673]
larsi@rusty:~/pgnus$ lsof | grep 586673
emacs     12143      larsi   10u     IPv4             586673      0t0     TCP 10.0.1.2:50099->dough.gmane.org:nntp (ESTABLISHED)

This is a STARTTLS connection, so I'm guessing that the sendto up there
is from GnuTLS.

Assuming that my guess is correct, does anybody know whether there's
some option we can set to make GnuTLS less, er, willing to infloop while
trying to send data?

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Sent from my Rome