Re: bug #94 alive again

Hi Clemens,

At 21.36 29/09/2002, PsychoTekk .de wrote:
> > In Your cache.log, between 21/9 and today, there is some events like:
> >
> > "comm_select: FD xxx signaled but with nulled descriptor. Closing"
> >
> > If no, the Pedro's patch doesn't work.
>
>i did a search that lists all lines that begin with "comm_select: FD",
>but only got hits like "comm_select: FD xxx ready for reading/writing".
>
>regards,
> Clemens

Try the following change in http.c:

Index: http.c
===================================================================
RCS file: /cvsroot/squid/squid/src/http.c,v
retrieving revision 1.20.6.5
diff -u -p -r1.20.6.5 http.c
--- http.c 24 Sep 2002 17:45:31 -0000 1.20.6.5
+++ http.c 29 Sep 2002 20:01:06 -0000
@@ -948,7 +948,10 @@ httpSendRequest(HttpStateData * httpStat
      else if (entry->mem_obj == NULL)
         cfd = -1;
      else
+ {
         cfd = entry->mem_obj->fd;
+ assert(FD_SOCKET == fd_table[cfd].type);
+ }
      assert(-1 == cfd || FD_SOCKET == fd_table[cfd].type);
      if (p != NULL)
         httpState->flags.proxying = 1;

So we can be sure that the crash is in the wrong fd_table entry.

Regards

Guido

-
=======================================================
Serassio Guido
Via Albenga, 11/4 10134 - Torino - ITALY
E-mail: guido.serassio@serassio.it
WWW: http://www.serassio.it
Received on Sun Sep 29 2002 - 14:03:52 MDT