Re: [lwip-users] How can I slow down the speed to receive with RAW API?

Hi, Everyone.

I think I haved found the reason.

If where somebody meet this problem, maybe it can be referenced.

It is a mistake in my porting.

I used RAW API, it forgot to enable "SYS_LIGHTWEIGHT_PROT".

functions like memp_free(), pbuf_free() should be mask interrupt when it using "memp_tab[]". It is globle.

So if you use these function in ISR, it will be destroy the link of pbuf.

Make sure to lock the globle dada.

I hope this can be usedful.

 

Best regards!

Locke_h

 

---

2008-10-27

---

发件人： address@hidden

发送时间： 2008-10-24 17:56:14

收件人： address@hidden

抄送：

主题： [lwip-users] How can I slow down the speed to receive with RAW API?

 

Hi, Every one!

My LwIP works on MCU with a 32KB SRAM.(Freescale MCF52235)

I port the "echo" function to my project. 

http://download.savannah.gnu.org/releases/lwip/contrib-1.3.0.zip

lwip/contrib/ports/unix/proj/minimal/echo.c

I connected a client to LwIP, with is made by VC. This tool software has no problem.

It can works well with long time.

But sometimes if I make rush data to LwIP. It will report that can not allocate pbuf in low_level_input(). And the packets most are droped.

The remained pbuf can not be free by LWIP.

I don't know why.

Could you please offer some advize?

Thank you!

Best Regards!

 

---

2008-10-24