lwip-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[lwip-users] Assertion "unsent_oversize mismatch (pcb vs., last_unsent)"


From: Sandra Gilge (ADATIS)
Subject: [lwip-users] Assertion "unsent_oversize mismatch (pcb vs., last_unsent)"
Date: Thu, 7 Jul 2016 18:15:07 +0200
User-agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.1.1

Hallo Simon,

I think now this error is caused by the driver. (See also my last question: "Problem with blackfin port")

I did a change in this driver and the assertion didn't occur anymore.

In the driver there was a command that caused a MEMP_POOL malloc for RX buffers in the TX-callback function. In my opinion this got there by accident.
Thus PBUFs are allocated as RX buffers when they shouldn't. It doesn't cause errors as long as I receive more packets than I send (the RX event then consumes these buffers).

But when I let my JPEG streamer run I seem to send statistically more packets than receive. After about 2 hours of streaming I seem to reach the end of the buffers. Still I would not expect a assertion in this case, but a simple memerr. And when I stop streaming the network should be fine after a short while?

Best regards,

Sandra


Sandra Gilge wrote:
Hallo,
Hello 
?
Now I?m getting at heavy load the following assertion:
Assertion "unsent_oversize mismatch (pcb vs. last_unsent)" failed at line 465 in .\src\core\tcp_out.c
[..]
Anyone an idea?
Unfortunately no. Can you be absolutely sure you (or the port) don't violate lwIP's threading/reentrance requirements?
Debugging the target when the assertion fails might help (e.g. what are the pcb's members)?
I'd be happy if you could reproduce this, best with the win32 port. 

Simon

 

reply via email to

[Prev in Thread] Current Thread [Next in Thread]