Squid and SIGPIPE

I have been overlooking these for a while on the assumption that it was
gdb catching something when it should have left it. I have just started
wondering if this was in fact bad unexpected behaviour for the squid
signals. Should I be looking into it?

I run all my test builds under gdb with -ND to catch any big problems
visibly instead of silent child restarts.

It has been giving me
Program received signal SIGPIPE, Broken pipe.
0xffffe410 in __kernel_vsyscall ()
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7ce60b3 in __write_nocancel () from /lib/tls/i686/cmov/libc.so.6
#2 0x081128e2 in commHandleWrite (fd=14, data=0x8326884) at comm.cc:1963
#3 0x080876d5 in comm_select (msec=10) at comm_epoll.cc:309
#4 0x0811623e in CommSelectEngine::checkEvents (this=0xbfbb92fc,
timeout=10) at comm.cc:2614
#5 0x0809ba81 in EventLoop::checkEngine (this=0xbfbb92bc,
engine=0xbfbb92fc, primary=true) at EventLoop.cc:50
#6 0x0809bc1a in EventLoop::runOnce (this=0xbfbb92bc) at EventLoop.cc:124
#7 0x0809bd4a in EventLoop::run (this=0xbfbb92bc) at EventLoop.cc:100
#8 0x080d62ee in main (argc=4, argv=0xbfbb93b4) at main.cc:1310

or similar under some easily repeatable conditions.
Received on Thu Jul 05 2007 - 05:40:08 MDT