Bug?
1 view
Skip to first unread message
fdsa
Nov 4, 2008, 6:09:03 PM11/4/08
to ebbebb
On Mac OS X, when a client disconnects prior to the request
finishing, libebb will start looping and spike CPU until its
timeout triggers and the connection is closed.
dtruss shows what's going on:
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
...
For some reason libebb doesn't immediately close the connection
when recv() returns zero (ebb.c:311). This looks intentional,
though. What's going on here?
-- mike
finishing, libebb will start looping and spike CPU until its
timeout triggers and the connection is closed.
dtruss shows what's going on:
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
recvfrom(0x7, 0xBFFEF1F1, 0xFFFF) = 0 0
kevent(0x3, 0x3006A0, 0x0) = 1 0
...
For some reason libebb doesn't immediately close the connection
when recv() returns zero (ebb.c:311). This looks intentional,
though. What's going on here?
-- mike
ry dahl
Nov 5, 2008, 4:27:01 PM11/5/08
to ebb...@googlegroups.com
yes - bug (left over from some debugging?)
thank you.
thank you.
Reply all
Reply to author
Forward
0 new messages