[Pan-users] Segfault in today's pan.git updated

[walt]

I've hit a few confusing problems after compiling pan.git today, but this
is the worst one:

(gdb) bt
#0  0x00007ffff5178e6f in g_io_channel_unix_get_fd (channel=0x0) at 
giounix.c:657
#1  0x0000000000575951 in pan::GIOChannelSocketGnuTLS::open (this=0x7fffd80010c0, 
address=..., port=563, setme_err="")
    at ../../../pan/tasks/socket-impl-openssl.cc:349
#2  0x00000000005733cd in (anonymous namespace)::ThreadWorker::do_work 
(this=0xb743c0) at ../../../pan/tasks/socket-impl-main.cc:99
#3  0x00000000005bbeda in pan::WorkerPool::Worker::worker_thread_func (g=0xb743c0, 
unused=<optimized out>)
    at ../../../pan/general/worker-pool.cc:75
#4  0x00007ffff5155bbd in g_thread_pool_thread_proxy (data=0x92b370) at 
gthreadpool.c:309
#5  0x00007ffff51555fb in g_thread_proxy (data=0x923540) at gthread.c:801
#6  0x00007ffff3ad901b in start_thread (arg=0x7fffebbe9700) at 
pthread_create.c:308
#7  0x00007ffff380d74d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)
#0  0x00007ffff5178e6f in g_io_channel_unix_get_fd (channel=0x0) at 
giounix.c:657
#1  0x0000000000575951 in pan::GIOChannelSocketGnuTLS::open (this=0x7fffd80010c0, 
address=..., port=563, setme_err="")
    at ../../../pan/tasks/socket-impl-openssl.cc:349
#2  0x00000000005733cd in (anonymous namespace)::ThreadWorker::do_work 
(this=0xb743c0) at ../../../pan/tasks/socket-impl-main.cc:99
#3  0x00000000005bbeda in pan::WorkerPool::Worker::worker_thread_func (g=0xb743c0, 
unused=<optimized out>)
    at ../../../pan/general/worker-pool.cc:75
#4  0x00007ffff5155bbd in g_thread_pool_thread_proxy (data=0x92b370) at 
gthreadpool.c:309
#5  0x00007ffff51555fb in g_thread_proxy (data=0x923540) at gthread.c:801
#6  0x00007ffff3ad901b in start_thread (arg=0x7fffebbe9700) at 
pthread_create.c:308
#7  0x00007ffff380d74d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:114

There are other problems when pan reads an old servers.xml file copied from a 
different
~/.pan2 directory, but I'll try to debug that tomorrow when I'm awake :)

For now, I'll just mention that xzver slows header fetching to almost a 
standstill,
with occasional bursts of speed lasting less than a second.  I have only two 
news
servers to test against, and one of the two obviously doesn't support xzver at 
all,
so that leaves me with only one server to test against.

Anyone else seeing the same with today's pan.git?  (commit 4f681cd2)

BTW, I don't know what diablo and xfeature are/do, so I don't know how to test 
them.

Cheers, Heinrich, and thanks as always for your good work.