Hi,

I just got access to an ACER and will try the new driver with the ISO. 
But it looks like the IOMMU support in the BIOS is broken.
The BIOS is supposed to report a MAGAW i.e. maximum adjusted guest 
address width of 39/48/... . Instead it reports a MAGAW of 12 which is 
completely broken.

There are some things you can try.

1. Cold boot into the Solaris DVD. If you *enable* the IOMMU or 
virtualization in the BIOS and then warm boot into Solaris - you will have 
this problem. This is a known issue that BIOS vendors and Intel warn 
against. Note, this doesn't mean that you cant warm boot into Solaris, 
only that the first boot after *toggling* IOMMU/virtualization in the BIOS 
must be a cold boot.

2. The ACER that I have does not have an option to disable/enable 
Virtualization i.e. the BIOS always enables IOMMU. If that is the case 1 
doesnt apply. In that case the BIOS is broken and you have to disable 
the IOMMU with the following:

   Append 
        -B immu-enable=false
   to the end of the kernel$ line in the GRUB menu.

3. From what I have seen, even disabling IOMMU will not still allow the 
system to boot because there is another problem with ehci (USB 
controller driver).
The BIOS in my ACER is crippled and will not allow any USB settings to 
be toggled, so until we have ehci driver fixes, we cant boot opensolaris.

Let me know how it goes.

Vikram

On Saturday 20 February 2010 13:42:05 Gary Mills wrote:
> > Still the machine panics early on boot (page fault).
> > Has the option "intel-iommu=no" been gone in the
> > rewrite? Even with this option set (via "-B
> > intel-iommu=no" on the GrUB command line) the machine
> > panics.
> 
> I just tried OS133 too on my AS3810TH.  With no boot option
> changes, it displays the Solaris banner and then reboots into 
Windows!
> 
> When I boot with `-kdv', things get a bit more interesting.  I used the
> same breakpoints as before, but this time I only got to enter `:c' four
> times...
> 
> [0]> prom_debug/W 1
> [0]> ::bp acpica`acpica_init                    2 3
> [0]> ::bp pci_autoconfig`pci_enumerate          1 4
> [0]> ::bp isa`isa_attach
> [0]> post_startup_cpu_fixups:b
> [0]> vfs_mountroot:b
> 
> Processing DMAR ACPI table
> NOTICE: agptarget0: IMMU: Mobile 4 chipset detected ...
> Creating state structures for Intel IOMMU units
> WARNING: Calculated AGAW(39) is outside valid limits [30,12] 
specified by
>  Vt-d spec and magaw trap: Unknown trap type 8 in user mode
> panic[cpu0]/thread-...
> 
> Is this enough information to isolate the problem?
> 

Reply via email to