[PATCH] Rework DSDT generation in bhyve

[John Baldwin]

On Saturday, December 28, 2013 10:14:01 pm Neel Natu wrote:
> Hi John,
> 
> On Fri, Dec 27, 2013 at 9:05 AM, John Baldwin <jhb at freebsd.org> wrote:
> > While hacking on the power button support, I also started rototilling the DSDT
> > generation code a bit.  My initial goal was to enumerate the LPC serial ports
> > (COM1 and COM2) properly via ACPI.  I ended up doing the following:
> >
> > - Moved the info for the top-level PCI bus into the PCI emulation code and
> >   added ResourceProducer entries for the memory ranges decoded by the bus
> >   for memory BARs.
> > - I added a framework to allow each PCI emulation driver to optionally write
> >   an entry into the DSDT under the \_SB_.PCI0 namespace.  The LPC driver uses
> >   this to write a node for the LPC bus (\_SB_.PCI0.ISA).
> > - I added a linker set to allow any LPC devices to write entries into the
> >   DSDT below the LPC node.  I moved the existing block for the RTC out of
> >   acpi.c and into the RTC driver.  I added DSDT nodes for the AT PIC,
> >   the 8254 ISA timer, and the LPC UART devices.
> > - I also added a "SuperIO" device under the LPC node to claim "system
> >   resources" (as is done in real hardware).  I added a linker set to allow
> >   various drivers to add IO or memory ranges that should be claimed by
> >   SuperIO and then added the extended RTC IO range and the registers used
> >   for ACPI power management as system resources.
> >
> > The end result is that for the stock VM created by vmrun.sh, the attimer0,
> > uart0, and uart1 devices move from isa0 to acpi0.  There is also a PIC
> > device that would be claimed by 'device atpic' (but the stock amd64 kernel
> > doesn't include that).  The DSDT is also a bit more fleshed out, and also
> > looks "nice" as devices are mostly laid out in the normal tree rather than
> > using separate Scope() sections for each device.
> >
> > Note that I did add some helper routines for writing out DSDT lines and
> > resource entries to try to simplify the code in the various DSDT handlers
> > and make the code that generates DSDT lines a bit more readable (e.g. the
> > implicit newlines make things more readable IMO).
> >
> > The patch is relative to the previous power button patch and is at
> > http://www.FreeBSD.org/~jhb/patches/bhyve_dsdt.patch
> >
> 
> This looks great!
> 
> This is not related to your patch (it has always been there), but I
> was trying to understand the significance of this entry:
> pci_lpc_write_dsdt(): OperationRegion (P40C, PCI_Config, 0x60, 0x04)
> 
> Any idea on why this is needed/interesting?

Oh, I assumed it was just copied and pasted from whatever template was
used for the initial DSDT.  I was thinking of removing it because it is
unused.  If you are happy with me doing so I'll gladly axe it.

> best
> Neel
> 
> > Assuming these are ok, the next thing I might work on is cleaning up
> > the PCI INTx interrupts.  They should really all be rounted to the I/O APIC
> > intpins above 15 (and we should leave ISA IRQs 5, 10, and 11 alone in APIC
> > mode and only fall back to that if we are using the 8259As and a real ELCR).
> > The MP Table output would need some minor tweaking for that, and for ACPI we
> > should generate a _PRT table under _SB_.PCI0 in pci_emul.c.
> >
> > --
> > John Baldwin
> > _______________________________________________
> > freebsd-virtualization at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-virtualization
> > To unsubscribe, send any mail to "freebsd-virtualization-unsubscribe at freebsd.org"
> 

-- 
John Baldwin