Re: [lwip-users] Could sys_restart_timeouts be provided also for !NO_SYS targets ?

[Ari Suutari]

On 26 Sep 2016, at 21:34 , Ari Suutari <address@hidden> wrote:

I’ll enable sys_restart_timeouts and see if it works also, I think it should

because the effect should be same as my workaround.

I quickly tested sys_restart_timeouts for my case, it seems to work

ok. Without it, dhcp timers are not running properly after sleep and

system is not able to get dhcp lease. When using sys_restart_timeouts

just after system has waken up, dhcp client works perfectly.

   Ari S.

On 26 Sep 2016, at 21:20 , Sylvain Rochet <address@hidden> wrote:

Hi Ari,

On Mon, Sep 26, 2016 at 02:13:10PM +0300, Ari Suutari wrote:

Hi,

This related to http://savannah.nongnu.org/bugs/index.php?48568

I have a system which sleeps long times but sys_now() progresses during
sleep.
Currently this still seems to cause some surprises with timers.

While there seems to be discussion about moving to absolute time
in timers instead of current implementation, could it be possible that
timeouts.c provided
sys_restart_timeouts() function also for targets that are using RTOS in the
meantime ?

The problem here is that sys_timeouts_mbox_fetch() is still running when
you go to sleep, therefore calling sys_restart_timeouts() would do
nothing, this is why this function is currently not available in !NO_SYS
mode.

Anyway, could you try enabling sys_restart_timeouts() for !NO_SYS and
then calling it with tcpip_callback() to see what happen ? Theoretically
it should work but I'm not confident enough to certainly state there is
no corner case doing so.

(And yes, absolute timers does not have this problem ;)

Sylvain
_______________________________________________
lwip-users mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/lwip-users

_______________________________________________
lwip-users mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/lwip-users