cvs commit: src/sys/amd64/amd64 mptable_pci.c msi.c nexus.c src/sys/amd64/include intr_machdep.h src/sys/amd64/pci pci_bus.c src/sys/dev/acpica acpi_pcib_acpi.c acpi_pcib_pci.c src/sys/dev/pci pci.c pci_if.m pci_pci.c pci_private.h pcib_if.m ...

John Baldwin jhb at freebsd.org
Fri May 4 19:47:05 UTC 2007


On Friday 04 May 2007 11:49:35 am Rui Paulo wrote:
> At Wed, 2 May 2007 17:50:36 +0000 (UTC),
> John Baldwin wrote:
> > 
> > jhb         2007-05-02 17:50:36 UTC
> > 
> >   FreeBSD src repository
> > 
> >   Modified files:
> >     sys/amd64/amd64      mptable_pci.c msi.c nexus.c 
> >     sys/amd64/include    intr_machdep.h 
> >     sys/amd64/pci        pci_bus.c 
> >     sys/dev/acpica       acpi_pcib_acpi.c acpi_pcib_pci.c 
> >     sys/dev/pci          pci.c pci_if.m pci_pci.c pci_private.h 
> >                          pcib_if.m pcib_private.h pcivar.h 
> >     sys/i386/i386        mptable_pci.c msi.c nexus.c 
> >     sys/i386/include     intr_machdep.h 
> >     sys/i386/pci         pci_bus.c 
> >   Log:
> >   Revamp the MSI/MSI-X code a bit to achieve two main goals:
> >   - Simplify the amount of work that has be done for each architecture by
> >     pushing more of the truly MI code down into the PCI bus driver.
> >   - Don't bind MSI-X indicies to IRQs so that we can allow a driver to map
> >     multiple MSI-X messages into a single IRQ when handling a message
> >     shortage.
> 
> 
> With this change, I think, I get a WITNESS panic when booting the
> second core on my MacBook
> 
> After SMP: AP #1 Launched, I get this panic:
> panic: blockable sleep lock (sx) msi @ i386/i386/msi.c:374
> 
> The backtrace is:
> witness_checkorder()
> _sx_lock()
> msi_map()
> nexus_map_msi()
> acpi_pcib_map_msi()
> pcib_map_msi()
> pci_remap_msi_irq()
> msi_assign_cpu()
> 
> Any idea why?

Hmm, I know why, I'll look at it.  I think msi_assign_cpu() is called while 
holding a lock and that is leading to the problem.

-- 
John Baldwin


More information about the cvs-all mailing list