r8169 vpd r/w failed

r8169 vpd r/w failed

[Dave Jones]

Francois,
 We occasionally see reports from r8169 where it seems to hang
reading vpd data from the NIC. The printk in drivers/pci/
suggests it's likely a firmware bug on the device, and recommends the user
contact the vendor for an update (does Realtek even do firmware updates?)

Is this as the message suggests a firmware bug, or should r8169 be
avoiding trying to read vpd ?

https://bugzilla.redhat.com/show_bug.cgi?id=827137 is one example.

thanks,

	Dave

Re: r8169 vpd r/w failed

[Francois Romieu]

Dave Jones <davej@redhat.com> :
[...]
>  We occasionally see reports from r8169 where it seems to hang
> reading vpd data from the NIC. The printk in drivers/pci/
> suggests it's likely a firmware bug on the device, and recommends the user
> contact the vendor for an update (does Realtek even do firmware updates?)
> 
> Is this as the message suggests a firmware bug, or should r8169 be
> avoiding trying to read vpd ?

I have not practiced vpd with the 816x a lot but I have experienced an
unlimited read behavior with the 8168 while trying to access the vpd back
in march. The output exhibited a periodic pattern, so I got rid of
it with something similar to drivers/pci/quirks.c::quirk_brcm_570x_limit_vpd.

I can resurrect it but I'd rather give the pending softirq stuff a try
first and I must test davandra netif_napi_del patch as well. Btw I have
just upgraded my workstation to F17 and it's time for some sleep.

> https://bugzilla.redhat.com/show_bug.cgi?id=827137 is one example.

(990FX)

Adding a RTL_GIGA_MAC_VER_34 case similar to RTL_GIGA_MAC_VER_24 in
rtl_init_rxcfg may temporarily help if netdev watchdog triggers (see
the mess known as https://bugzilla.kernel.org/show_bug.cgi?id=42899).

-- 
Ueimor