[OpenIndiana-discuss] JMB363 Chipset support in OI_148

[Sean O'Brien]

On Sun, May 8, 2011 at 1:50 PM, Albert Lee <trisk at opensolaris.org> wrote:

> On Sun, May 8, 2011 at 5:33 AM, Sean O'Brien <upinthecloudz at gmail.com>
> wrote:
> > I have a pci-express card based on JMB363 chipset, because it is listed
> in
> > HCL. The card is recognized by the ahci driver.
> >
> > model:  'SATA AHCI 1.0 Interface'
> >                power-consumption:  00000001.00000001
> >                devsel-speed:  00000000
> >                interrupts:  00000001
> >                subsystem-vendor-id:  0000197b
> >                subsystem-id:  00002363
> >                unit-address:  '0'
> >                class-code:  00010601
> >                revision-id:  00000002
> >                vendor-id:  0000197b
> >                device-id:  00002363
> >                name:  'pci197b,2363'
> >
> > The device is seen by cfgadm as well, but an attempt to connect a drive
> or
> > to reset the controller results in the following in /var/adm/messages:
> >
> > May  8 01:54:36 fileserv ahci: [ID 860969 kern.warning] WARNING: ahci1:
> > ahci_port_reset port 0 the device hardware has been initialized and the
> > power-up diagnostics failed
> > May  8 01:54:59 fileserv ahci: [ID 860969 kern.warning] WARNING: ahci1:
> > ahci_port_reset port 1 the device hardware has been initialized and the
> > power-up diagnostics failed
> >
> >
> > There are some patches posted for an old version of OpenSolaris to
> resolve
> > issues with the ahci driver and this card.
> > http://web.archiveorange.com/archive/v/1sSOwqLUl8mdFzhkydYU
> >
> > Would it be possible to apply these patches to the kernel source for
> oi_148?
> > Any assistance or guidance on compiling this would be much appreciated.
>
> Hi Sean,
>
> Thanks for the pointer, unfortunately both of those bugs were fixed in 2008
> as:
> 6645543 relax AHCI checks violated by JMicron JMB363 controller
> 6648246 AHCI driver looks for its registers wrongly, blocking support
> for JMicron JMB363
>
> This is not the first time problems have been mentioned for JMB363
> controllers recently, though.
>
> If you're interested in looking into this problem, you should sign up
> on the illumos developer mailing list.
> http://www.illumos.org/projects/site/wiki/Mailing_Lists
> Also, the build instructions are here:
> http://www.illumos.org/projects/illumos-gate/wiki/How_To_Build_Illumos
>
> -Albert
>


I found another SATA controller (sil3132 based) which is working properly.
However, I upgraded ZFS while in OpenIndiana before testing the
suspend-to-ram functionality with the new controller, and it seems that the
system is no longer able to suspend in oi_148 as it was under snv_134. It
seems that there are two basic reasons for this which are documented on the
Oracle side.

First, there is a feature added to the kernel around build 147 which
prevents the system from suspending properly. This is reportedly fixed, but
without Oracle access it's not clear how exactly it was fixed or which
version it was fixed in.
https://defect.opensolaris.org/bz/show_bug.cgi?id=17093


Second, there is a bug which prevents the kernel from suspending while SMB
service is active, which has been fixed in development build 156, according
to this thread.
http://forums.oracle.com/forums/thread.jspa?threadID=2160046
Unfortunately I couldn't find this bug in the defect tracker.

I have seen other users post complaints about S3 functionality in OI_148,
and I'm hoping that in future builds of OpenIndiana these fixes can be
included. I'm going to try building Illumos, and see if a newer kernel will
help.

Thanks
Sean