Re: [lwip-users] sys_msleep() implementation

[address@hidden]

Hi Marko,

What Piero said is correct: the code 'blocks' on a semaphore using 
sys_sem_wait. In contrast to sys_arch_sem_wait (note the 'arch' in the 
name) or your proposed OS_sleep, sys_sem_wait only blocks for a specific 
time.

Example: if you have the ARP timer running (5 seconds interval) and call 
sys_msleep(15000), the code in sys_sem_wait will sleep 5 seconds, 
process the ARP timer callback, sleep 5 seconds, and so on. This is 
repeated until the original time passed to sys_msleep has passed.

This is what the comment says: "Timeouts are processed while sleeping."
The actual blocking on the semaphore (with a specific timeout) is done 
with the (arch-dependent) function sys_arch_sem_wait.

So the simple answer to your original question is: No, you cannot use 
your OS'es sleep function since timeouts are not processed while 
sleeping and the stack depends on them. (More advanced answer: in 
application threads, you may use the OSes sleep function directly, since 
timeouts are only used in the tcpip_thread.)

Simon


Marko Panger wrote:
> Hi,
>
> In the original implementation it blocks anyway by pending on semaphore. 
> The effect is the same. Its just the implementation as I see it, but I 
> could be wrong.
>
> marko
>
> Piero 74 wrote:
>   
> > Hi marko
> >
> > the comment of this function said:
> >
> >  * Sleep for some ms. Timeouts are processed while sleeping.
> >  *
> >  
> > (and you can see how work in the lwip code)
> > Lwip needs to checks its timeout list during sleep operation, so, you 
> > cannot block it using RTOS delay
> >
> > Someone can confirm my answer?
> >
> > Bye, Piero.
> >
> > 2008/5/6 Marko Panger <address@hidden 
> > <mailto:address@hidden>>:
> >
> >     Hi all,
> >
> >     I was looking at the implementation of sys_msleep() function in
> >     sys.c. As per default in crates
> >     a semaphore and then pends on it for the timeout value. Could I
> >     redefine this function in sys_arch.c
> >     to better suit my RTOS or is there any particular reason for such
> >     implementation ?
> >
> >     For example I would implement it as:
> >
> >     void sys_msleep(u32_t ms) {
> >       TSK_Sleep(ms);   }
> >
> >     Any thoughts ?
> >
> >     Thanks,
> >     Marko
> >
> >
> >
> >     _______________________________________________
> >     lwip-users mailing list
> >     address@hidden <mailto:address@hidden>
> >     http://lists.nongnu.org/mailman/listinfo/lwip-users
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > lwip-users mailing list
> > address@hidden
> > http://lists.nongnu.org/mailman/listinfo/lwip-users
> >     
>
>
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users
>
>
>