The patch was incomplete. Please use the version attached to this message.
REgards
Henrik
On Mon, 12 Sep 2005, Henrik Nordstrom wrote:
On Sun, 11 Sep 2005, OpenMacNews wrote:
2005/09/11 09:23:57| comm_poll: 1+0 FDs ready
looking at this, i don't thing i've verified your theory, but i'm not at
all sure what *is* happening ...
I think it is the same.. just trapped a bit earlier than I expected.. I have
attached a small patch which fixes what I saw in the source. Please give it a
try.
off to google on com_poll, unless i hear from you first :)
It's a Squid internal function, keeping track of which connections are ready
for reading/writing. Google will probably find you a lot of discussion
mentioning this function but none relevant to your problem.
Regards
Henrik
? q
Index: src/net_db.c
===================================================================
RCS file: /cvsroot/squid/squid/src/net_db.c,v
retrieving revision 1.158.2.10
diff -u -p -r1.158.2.10 net_db.c
--- src/net_db.c 30 Jun 2005 18:50:56 -0000 1.158.2.10
+++ src/net_db.c 11 Sep 2005 22:17:29 -0000
@@ -571,7 +571,13 @@ netdbExchangeHandleReply(void *data, cha
size -= hdr_sz;
p += hdr_sz;
} else {
- size = 0;
+ if (size >= ex->buf_sz) {
+ debug(38, 3) ("netdbExchangeHandleReply: Too big HTTP header, aborting\n");
+ netdbExchangeDone(ex);
+ return;
+ } else {
+ size = 0;
+ }
}
}
debug(38, 5) ("netdbExchangeHandleReply: start parsing loop, size = %ld\n",
