Dave wrote
> On 02/07/2008 09:21 AM, Jürgen Keil wrote:
> > Do you get an "excessive interrupt processing. Disabling port"
> > message from the nv_sata driver before that WARNING, perhaps?
> > See bug 6642154:
> > http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6642154
>
> I don't see anything with the "excessive interrupt processing" message
> in /var/adm/messages.
Is your root (/var) filesystem disk using the nv_sata driver, too?
If it is using nv_sata, the kernel won't be able to log the
message to /var/adm/messages any more, or write a crash
dump that would contain this message in it, after the nv_sata
driver has disabled the sata port due to "excessive interrupt
processing".
I think you would only see that message on the console
(X server not running) or on a serial console.
> I do see these for IRQ20, 21, and 22:
>
> unix: [ID 954099 kern.info] NOTICE: IRQ22 is being shared by drivers with
> different interrupt levels.
>
> I assume these are benign?
The "interrupt sharing" message should be harmless.
> >
> >
> > I'm now running opensolaris with an nv_sata driver changed like this,
> > and so far, there have been no more "excessive interrupt processing"
> > hangs:
> >
> > diff -r a183e54573d2 usr/src/uts/common/sys/sata/adapters/nv_sata/nv_sata.h
> > --- a/usr/src/uts/common/sys/sata/adapters/nv_sata/nv_sata.h Thu Feb 07
> > 06:05:33 2008 -0800
> > +++ b/usr/src/uts/common/sys/sata/adapters/nv_sata/nv_sata.h Thu Dec 13
> > 10:23:15 2007 +0100
> > @@ -625,7 +625,7 @@ typedef struct prde {
> > * processing loop, declare the interrupt wedged and
> > * disable.
> > */
> > -#define NV_MAX_INTR_LOOP 10
> > +#define NV_MAX_INTR_LOOP 32
> >
> > /*
> > * flag values for nv_copy_regs_out
>
> Would this be a waste of time to try, since I'm not seeing the
> "excessive interrupt" messages in /var/adm/messages?
Maybe no waste of time, most likely your system won't be
able to record that message to /var/adm/message after the
sata port is disabled by the driver.
> >> I wonder, can I force Solaris to use the legacy ata
> >> driver for the MCP55 chip in SXCE b81?
> >
> > Should work by removing the relevant "nv_sata"
> > driver binding entry from the /etc/driver_aliases file.
> >
> > Note that if your root filesystem is on "nv_sata"
> > this will break booting from that drive, until you
> > manually fix the bootpath in /boot/solaris/bootenv.rc,
> > and update /etc/vfstab with the new disk device
> > names.
> >
>
> Very good. I will try this as a last resort. I really
> would like to have the improved performance of native SATA.
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
