|
From: | Bill Auerbach |
Subject: | RE: [lwip-users] Queued segments not being sent |
Date: | Fri, 7 Mar 2008 17:17:29 -0500 |
Yeah, you’re right. The good news is, I found the
problem. That tcp_poll routine sneaks in and was calling send_data (my
version of it) and this threw off my one to one relationship of calling
tcp_write and having tcp_sent be used to free the data. I’m not
sure what the purpose of tcp_poll is, but it was getting in my way. Bill From: address@hidden
[mailto:address@hidden On Behalf Of Taranowski, Thomas (SWCOE) It’s always been my recommendation
to leave early on a Friday, for precisely this reason. Working hard on a
Friday always causes problems. From:
address@hidden
[mailto:address@hidden On Behalf Of Bill Auerbach The problem appears to be the first
segment in the big list. It is being sent, and wireshark’s
“Follow TCP stream” contains the following: LINK address@hidden@V......>address@hidden:
0 ..drop: 0 ..chkerr: 0 ..lenerr: 0 ..memerr: 0 ..rterr: 0 ..proterr: 0 ..opterr: 0 ..err: 0 ..cachehit: 0 <snip> LINK is correct and was the first
segment sent by itself (not sure why that happens…) The next
segment were these lines: LWIP_PLATFORM_DIAG(("xmit:
%"STAT_COUNTER_F"\n\t", proto->xmit)); LWIP_PLATFORM_DIAG(("rexmit:
%"STAT_COUNTER_F"\n\t", proto->rexmit)); LWIP_PLATFORM_DIAG(("recv:
%"STAT_COUNTER_F"\n\t", proto->recv)); Which is bad, and it picks up with fw:
which is correct to the end. The hex of the bad area is: 49 4e 4b 0a 09
00
.LINK... 00 40 54 68 01 08 fb 10 03 67 00
00 40 56 18 01 address@hidden address@hidden 09 03 14 02 3e 00 00 00 40 57 c8
01 09 0b 18 01 ....>... @W...... 15 00 00 66 77 3a 20 30 0a 09 00 64
72 6f 70 3a ...fw: 0 ...drop: Nothing in the bad data makes any
sense. It’s not freeing early or reusing data because strings would
appear or just the one word would be trashed. The good news in all of this is that
it’s Friday. J Bill |
[Prev in Thread] | Current Thread | [Next in Thread] |