svn commit: r305722 - head/sys/x86/x86

[Konstantin Belousov]

On Mon, Sep 12, 2016 at 04:47:35PM +0800, Sepherosa Ziehau wrote:
> On Mon, Sep 12, 2016 at 4:39 PM, Sepherosa Ziehau <sephe at freebsd.org> wrote:
> > On Mon, Sep 12, 2016 at 4:32 PM, Konstantin Belousov
> > <kostikbel at gmail.com> wrote:
> >> On Mon, Sep 12, 2016 at 04:57:58AM +0000, Sepherosa Ziehau wrote:
> >>> Author: sephe
> >>> Date: Mon Sep 12 04:57:58 2016
> >>> New Revision: 305722
> >>> URL: https://svnweb.freebsd.org/changeset/base/305722
> >>>
> >>> Log:
> >>>   x86: Use sx lock for interrupt sources.
> >>>
> >>>   - Certain pic_assign_cpu, e.g. msi_assign_cpu can have quite a long
> >>>     call chain.  For msi_assign_cpu, mutex makes complex PCI bridge
> >>>     drivers more tricky, e.g. sleep can note be called, etc, it will
> >>>     be pretty tricky for upcoming Hyper-V PCI bridge driver for PCI
> >>>     pass-through.
> >>
> >>>   - It is not used on any hot code path nor non-sleepable context, so
> >>>     sx should have the same effect as mutex.
> >> Did you tested ACPI_DMAR interrupt remapping mode with your patch and
> >> witness enabled ?
> >
> > Nope, I am about to try it now.  BTW, I just checked the code for
> > ACPI_DMAR, I didn't see anything obviously w/ the lock type change
> > though.
> 
> I didn't notice anything wrong w/ options
> ACPI_DMAR+WITNESS+INVARIANTS.  If you saw anything wrong, please let
> me know.

Options are not enough, you should explicitely enable interrupt remapping
with loader tunable hw.dmar.ir=1.