cURL / Mailing Lists / curl-library / Single Mail


failing test cases

From: Daniel Stenberg <>
Date: Sat, 21 May 2005 00:39:46 +0200 (CEST)


I just wanted to mention that I'm perfectly aware of the fact that there are
FTP test cases that fail wrongly and seemingly randomly.

I'm struggling hard to find the cause and fix this nuisance. I know it is
related to the 'sockfilt' program since I've followed the tracks back to it.

However, I've not yet understood why sockfilt sometimes do this (strace log
when curl disconnects from the server and it should go back to the listening
state to await the next connect):

00:36:54.748929 close(4) = 0
00:36:54.748956 munmap(0xb7fe8000, 4096) = 0
00:36:54.748986 select(4, [0 3], [], [], {120, 0}) = ? ERESTARTNOHAND (To be
00:36:55.771488 +++ killed by SIGKILL +++

I guess it takes some more tearing hair and long hours staring at the same
code over and over again until I'll have this nailed..

  Commercial curl and libcurl Technical Support:
Received on 2005-05-21