Bug 13398

(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).

On Thu, 28 May 2009 16:26:36 GMT
bugzilla-daemon@bugzilla.kernel.org wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=13398
> 
>            Summary: sky2 occasionally fails to recognize device on cold
>                     boot
>            Product: Drivers
>            Version: 2.5
>     Kernel Version: 2.6.29
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: Network
>         AssignedTo: drivers_network@kernel-bugs.osdl.org
>         ReportedBy: thomas.burdick@gmail.com
>         Regression: No
> 
> 
> occasionally the sky2 module will report in dmesg
> 
> sky2 0000:02:00.0: unsupported chip type 0xff
> 
> on a cold boot, unloading and reloading the module gives the same error.
> 
> the network adapter is unusable. A warm reboot and the network adapter is
> usable and the sky2 module reports success.
> 
> This is in a hp 1035nr netbook with, according to lspci, 
> Marvell 88E8040 PCE-E Fast Ethernet Controller
>

That indicates the hardware problem. The driver is failing on reading the PCI identification information, this indicates a BIOS or hardware problem, not something the driver can handle. Could also be a PCI-E power management issue.

I'd argue that while it may be a hardware issue, the windows driver handles this case just fine so there is something the driver can do though it may not be pretty.

My NDA with Marvell seems to have dried up, and never had access to windows driver or any errata's or related startup issues.  You could try the vendor sk98lin driver and if it works the initialization timing could be changed to match. If the vendor driver is broken, then you might be able to get them to look at the errata to figure out why it is broken.

I investigated a bit more about this laptop and possible hardware issues, it appears by default the bios does *not* enable the internel ethernet driver and there is an obscure option (I feel like all bios options are obscure honestly...) to enable it. There also appears to have been some bios updates that improve some issues with the hardware in general. It appears to work fine now so all is good. Thats pretty lousy that they require an NDA at all though, I was intersted in possibly fixing the problem myself as I've been learning quite a bit about kernel driver development only to see that they have NDA requirements to obtain documentation. I personally will avoid buying hardware with marvell adapters if there happens to be adapters with real documentation out there I'd like to know.

I appreciate the time and effort you put in to at least looking at this bug report.

Sorry it turned out to be something truely related to the laptop and not the driver, poor bug reporting on my part!