Hello Haavard,

A few remarks:

> From: Remy Bohmer <hskinnemoen@atmel.com>

My name, at your address ;-)))

> This patch splits up the interrupt handler of the serial port
> into a interrupt top-half and a tasklet.

I see you moved the handling of the sysrq-key to the tasklet. This was
actually a very nice feature in the IRQ-top half on preempt-RT. This
helps debugging running away RT-processes.

> In this version of the patch, we try to only do things that are
> absolutely necessary in the interrupt handler, storing away the
> status register along with the received character and letting the
> tasklet handle break, sysrq, error flags, etc.

Preempt-RT now absolutely requires my (4th) IRQ_NODELAY patch, because
the spinlock now is always inside the code, and not only in
theexception path, and thus without my NO_DELAY patch we have a panic
during boot.
On preempt-RT this spinlock must be a raw-spinlock. (If this type is
known in the mainline kernel, you can apply that patch it anyway)

BTW: Attached I have added a 2nd patch that I use for Preempt-RT. (For
cleaner startup, and to get rid of useless IRQ-threads.

> This patch should apply on top of the cleanup patch I sent earlier

For the cleanup patch:
Acked-by: Remy Bohmer <linux@bohmer.net>

> today. Or at least I think so...I'll send the full series once
> everyone are happy.

So, for this patch: I am almost happy ;-)


Kind Regards,

Remy