Allow two threads to share the same timer - provided they use distinct
sets of timer channels - without occasionally corrupting registers or
state flags.
lwIP has already used its own log level when deciding what to print.
If we reach the LWIP_PLATFORM_DIAG message no further filtering should
happen, so set the log level to match the macro used.
This fixes the missing IP addresses from ifconfig. Before:
> ifconfig
Iface ET0 HWaddr: 24:0a:c4:e6:0e:9f Link: up State: up
Link type: wired
inet addr: mask: gw:
inet6 addr: scope: link state: valid preferred
inet6 addr: scope: global state: valid preferred
Iface ET1 HWaddr: 24:0a:c4:e6:0e:9c Link: up State: up
Link type: wireless
inet addr: mask: gw:
inet6 addr: scope: link state: valid preferred
inet6 addr: scope: global state: valid preferred
>
With this change:
> ifconfig
Iface ET0 HWaddr: 24:0a:c4:e6:0e:9f Link: up State: up
Link type: wired
inet addr: 10.4.4.81 mask: 255.255.254.0 gw: 10.4.4.1
inet6 addr: fe80:0:0:0:260a:c4ff:fee6:e9f scope: link state: valid preferred
inet6 addr: 2001:db8:0:0:260a:c4ff:fee6:e9f scope: global state: valid preferred
Iface ET1 HWaddr: 24:0a:c4:e6:0e:9c Link: up State: up
Link type: wireless
inet addr: 10.4.4.86 mask: 255.255.254.0 gw: 10.4.4.1
inet6 addr: fe80:0:0:0:260a:c4ff:fee6:e9c scope: link state: valid preferred
inet6 addr: 2001:db8:0:0:260a:c4ff:fee6:e9c scope: global state: valid preferred
>
The mbox code contains a race condition in `mbox_put()`: When it
waits for a slot in the queue to become available, it is woken up with
IRQs enabled. It disables IRQs again as first thing, but by then
another thread may already have preempted the running thread and filled
the queue back up. In this case, a message in the queue would be
silently overwritten.
The MPU based stack guard is very unpleased by the stack overflow
happening during the test. The increase in stack size makes the MPU
stack guard happy again.
The `channels` member should not be set to the number of hardware
channels *n*, but to *n* - 1 instead. The last channel is implicitly
used in `timer_read()`. Hence out of *n* hardware channels, only *n* - 1
are available to the application.
This fixes a bug introduced by 4d02e15247
which incorrectly set the channel number to *n* rather than to
*n* - 1.
