Re: [Qemu-devel] [PATCH v2 0/3] SD emulation fixes for Pi2 Tianocore EDK2 UEFI

[Andrew Baumann <Andrew.Baumann@microsoft.com>]

> From: Peter Maydell [mailto:peter.maydell@linaro.org]
> Sent: Monday, 25 January 2016 05:57
> 
> On 16 December 2015 at 19:02, Andrew Baumann
> <Andrew.Baumann@microsoft.com> wrote:
> > This series contains fixes to the SD card emulation that are needed to
> > unblock Tianocore EDK2 UEFI (specifically, the bootloader for Windows
> > on Raspberry Pi 2).
> >
> > Changes in v2, based on feedback from Peter Crosthwaite:
> >  * correct implementation of CMD23 to switch to transfer state on
> completion
> >  * use an actual timer for the power-up delay, rather than relying on
> >    the guest polling ACMD41 twice
> >  * added patch 3: replace fprintfs with guest error logging
> >
> > (I'm guessing at the CC list here, since this code appears to be
> > unmaintained. Apologies if I guessed wrong!)
> 
> I still have this patchset lurking on my to-review list, but I have
> a feeling it's been superseded by other patches or possibly even
> patches committed to master. Am I right, or does it still need review?

This is the most recent version of the patch series. However, there was an unresolved question about migration compatibility for the vmstate layout (patch 2/3). First, it's not clear to me that we care about migrating SD instances across qemu versions -- is this important? If not, it seems that the simple fix is to bump the vmstate version number and be done with it. If we do need to support migration, I'm very hazy on the correct way to handle this, or how to test it. I posted a suggestion here:
https://www.mail-archive.com/qemu-devel@nongnu.org/msg343178.html
... it would be helpful to get some guidance on that, if you want me to pursue this path.

The other two patches in the series already got a Reviewed-By from Peter C. Once we nail the migration issue, I can rebase the patches and they should be ready to go.

Thanks,
Andrew