[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: recvmsg/pflocal
From: |
Roland McGrath |
Subject: |
Re: recvmsg/pflocal |
Date: |
Thu, 30 Aug 2001 19:15:15 -0400 (EDT) |
You can't always believe gdb. Do a "info frame" on a few of those
"duplicate" frames and the pointer values will show you they're really all
the same one.
> #5 0x464c457f in ?? ()
"\x46\x4c\x45\x7f" is little-endian for our old friend "ELF\^?", the first
thing in the first page mapped in from any ELF file, sometimes found at 0.
I bet frame #8 has a frame pointer (%ebp) of 0 (as it should).
That backtrace is what you always see in a normal idle signal thread.
In the first pflocal backtrace file there are two io_select calls blocking
for read on the same pipe, looks pretty normal. In the second backtrace
file syslogd looks normal, blocked in the socket_recv RPC; but in pflocal,
there is a socket_send blocking until someone reads out of the pipe, and
there is no socket_recv RPC in progress.
- recvmsg/pflocal, Marcus Brinkmann, 2001/08/30
- Re: recvmsg/pflocal,
Roland McGrath <=