[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH] vfio-pci: Quirk RTL8168 NIC
From: |
Francois Romieu |
Subject: |
Re: [Qemu-devel] [PATCH] vfio-pci: Quirk RTL8168 NIC |
Date: |
Wed, 14 May 2014 00:22:57 +0200 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
Alex Williamson <address@hidden> :
[...]
> > Oh right, I looked for code references to the register but didn't notice
> > that Linux configures it for MSI, not MSI-X. In my brief testing I only
> > saw that Windows generates interrupts on the first vector, so perhaps
> > not much lost without the extra vectors. I guess it's this patch that
> > proves that MSI-X can be configured without this backdoor then. Do you
> > have any insight into why this exists ?
No. I can only speculate that bar registers exhaustion and 64 bits decoders
are not completely alien to this. Hayes may provide some hindsight.
[...]
> So the cycle is:
>
> word read from command register: I/O+ Mem+ BusMaster+ DisINTx+ SERR+
>
> byte read from PCIe link control register[0]: CommClk+
> byte read from PCIe link control register[1]: nothing
>
> I'll be interested if that means anything to you. It's not a very high
> rate access, the command register access is maybe 1Hz. Thanks,
Nothing specific. It reminds me of rtl8168_esd_timer - kind of periodic
sanity check - in Realtek's own 8168 driver but the pattern is a bit
different. I need to check some recent revision of it though.
--
Ueimor