All right people, but since my main goal is to know the exact amount of
memory the lwIP uses, could I just find that out without setting
MEMP_MEM_MALLOC to 1, since it seems that it requires some changes in
the kernel itself, as Richard and Simon explained. I am still not daring
to make such changes since I am not really experienced, and especially
if I can avoid that (at least for now).
So, I have returned MEMP_MEM_MALLOC to 0, and generated with Xilinx SDK
a .map file (it's in attachments). Can we say that lwIP for sure won't
take more memory then what we can see in that file, ram_heap plus sum of
everything that makes memp_memory (which includes entire pbuf_pool
(PBUF_POOL_BUFSIZE*PBUF_POOL_SIZE) as well)? Is there anyway I can
verify that with mentioned iperf application, for example, to add some
enormous array which will fill the rest of RAM, and try to write to it,
and see whether that will make some influence on lwIP?
Noam, I had a look at your lwipopts.h, and I must say that your
suggestions have helped when it comes to lwIP performance, especially
the MEMP_NUM_PBUF define.
The Best,
Nenad
_______________________________________________
lwip-users mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/lwip-users
