I caught one case where i wasn't returning ERR_OK in the receive callback which
lead to the pbuf being set to refused and my callback was called again with a
already freed pbuf from the 250ms timer. But I still see the assert on really
high traffic from e.g. an intense scan by nmap.. there i got
Well...
You do know that p->payload is a void*, that is, a pointer to some
bytes, do you ?
http://www.nongnu.org/lwip/2_0_x/structpbuf.html#a8d32dc3e964369d4eec638fc37fbc460
Then, why do you want to peek at data between your driver and lwip ? You
are not using the RAW API, you are looking at r
Are you sure you are not freeing an already freed pbuf ? That is what
the assert is supposed to catch.
Try to trace back from the assert.
___
lwip-users mailing list
lwip-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/lwip-users
Ha, I love the word "severe" being used here...
I don't know, I've read some guys finding bugs on the Cube drivers, and
seems to vary depending on which micro you are using. Search the list.
The fact is that your messages are getting lost and you need to find
where...
I suggest running an echo w
Hi,
note that for pinging link local addresses you must spacify an
interface (-I option in GNU/Linux)
For global addresses, I had the same problem and it was because
Ethernet multicast was not enabled in my interface. Take a look at
[1]. as it may be your problem too.
---
Yes, I noticed that. It happens a lot when I have the TCP_SNDBUF high but goes
away when I set it to 2920 (2xMSS). Then, however, transmitting is very slow.
Do you have an idea why this is so?
I use the CubeMX driver and environment from ST, hoping that it does not have
severe bugs.
-Urspr
I doing below configuration to enable IPv6 in lwIP 2.0.2.
But some how I am not able to ping on link local or global address.
I am using Ethernet.
#define LWIP_IPV6 1
#define LWIP_IPV6_MLD 1
#define LWIP_IPV6_AUTOCONFIG 1
#define LWIP_ICM
Hello,
I have got some issues in reading the payload of the pbuf.
I'm using a STM32 microcontroller and run on it the LwIP Stack in the raw
API without a RTOS.
It's based on a example from ST with a Webserver and is working fine.
But now I want to read the payload, when a new packet is recieve
I don't know in the API you are using, nor with NO_SYS != 1, but with
RAW API and NO_SYS=1, in 1.4.1, I don't set the netif up, DHCP does, and
I have to check for (netif->dhcp->state == DHCP_BOUND) before using the
netif.
In 2.0+ yes, the "up" is administrative and you set the ifc up before
sta
It is stated in the RFCs that after a connection is closed, it should
not be reused for some time. This is to account for data arriving later.
This has been already discussed in this list.
If you need to accept further connections during that time, you need to
have more pcbs available.
When yo
You have missing frames here.
Frame #8 sends 1460 bytes starting @5841
Frame #9 sends 1460 bytes starting @8761
8761 - 5841 = 2920 = 2*1460
You have a frame "#8.5" (the one that should have been sent between 8
and 9) that is missing. It is getting lost somewhere between your sender
and your rece
Hi Noam,
Thanks for your comments. I am using the RAW API.
The complete data array I want to send is 32k. I use individual packets of 1460
bytes each.
> I think that you have set a too large window size for the mem size you have.
You mean TCP_WND? It looks like only a fraction of my MEM_SIZE
Hi Axel,
there have been some buggy drivers using DMA where transmit part "looped
around" and resent some frames.
Wrong pointers can write (trash) anywhere, etc. (DMA or non-DMA driver rx ptrs,
or user application)
___
lwip-users mailing list
lwip-use
Hi Adrian,
Are you using RAW, netcon or Socket API ?
I think that you have set a too large window size for the mem size you have.
You should not send 32K blocks in one go. You need to send smaller block, say
1K or 1.5K,
check available memory of send buffer and if available send another chun
It seems like I do not have this problem if I choose TCP_SND_BUF to be exactly
double the TCP_MSS of 1460. If I choose higher values, the problem happens.
Is there an explanation for this?
Von: lwip-users
[mailto:lwip-users-bounces+adrian.figueroa=tu-dresden...@nongnu.org] Im Auftrag
von Adria
Hello all,
I am trying to send large amounts of data continuously. I use lwIP as a TCP
client on an STM32F7 and Matlab on a PC as a TCP server.
I send blocks of 32768 bytes. Here are some settings for the memory sizes in
lwIP:
#define MEM_ALIGNMENT 4
#define MEM_SIZE 17520
#define MEMP_NUM_TCP
Thanks Simon!
I've gone through my project and resolved some issues and the /pbc->next
! pcb /hasn't appeared anymore. In the process i also updated to lwip 2.0.2.
But i've got another issue now.. My device is listening to port 502 and
as soon as someone connects and it starts receiving package
17 matches
Mail list logo