Re: Emacs hangs with concurrent TLS connections

[Ted Zlatanov <tzz@lifelogs.com>]

On Sun, 28 Nov 2010 15:20:50 +0100 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> This can be rather tricky to reproduce, but I seem to be able to
LMI> reproduce it kinda reliably with the following.

LMI> Don't eval it -- it might hang your Emacs completely.  Start a new one
LMI> and try it there:

LMI> (progn
LMI>   (require 'gnutls)
LMI>   (url-retrieve "https://www.google.no" #'ignore)
LMI>   (url-retrieve "https://www.google.no" #'ignore))

LMI> If it doesn't hang your Emacs, try some other URLs, and more of them.
  
LMI> Emacs becomes completely unresponsive, and the only thing that can be
LMI> done is to kill Emacs.

LMI> I get an infinite loop like this:

LMI> select(11, [3 4 6 8 9], [], NULL, {0, 19533}) = 1 (in [9], left {0, 19531})
LMI> getrusage(RUSAGE_SELF, {ru_utime={6, 424401}, ru_stime={1, 744109}, ...}) = 0
LMI> times({tms_utime=642, tms_stime=174, tms_cutime=0, tms_cstime=0}) = 1846926770
LMI> getrusage(RUSAGE_SELF, {ru_utime={6, 424401}, ru_stime={1, 744109}, ...}) = 0
LMI> times({tms_utime=642, tms_stime=174, tms_cutime=0, tms_cstime=0}) = 1846926770
LMI> read(3, 0xda6174, 4096)                 = -1 EAGAIN (Resource temporarily unavailable)
LMI> poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}], 2, 0) = 0 (Timeout)
LMI> select(11, [3 4 6 8 9], [], NULL, {0, 19388}) = 1 (in [9], left {0, 19386})

LMI> where fd 3 is

LMI> lrwx------ 1 larsi larsi 64 Nov 28 15:14 3 -> socket:[5510547]

LMI> which is

LMI> emacs     20504       larsi    3u     IPv4            5510547       0t0        TCP localhost.localdomain:57459->localhost.localdomain:6010 (ESTABLISHED)

LMI> which is the ssh-tunnelled X connection.

LMI> So something is up when trying to initialise two gnutls connections at
LMI> the same time.  Or something.  A race condition?  Some state shared that
LMI> shouldn't be shared?  I don't know.

Have you looked at a verbose log with gnutls-log-level set high?

Ted