On Fri, 27 Jun 2014 11:27:12 +0200 Christian Borntraeger <borntrae...@de.ibm.com> wrote:
> On 27/06/14 11:05, Alexander Graf wrote: > > > > > >> Am 27.06.2014 um 09:53 schrieb Christian Borntraeger > >> <borntrae...@de.ibm.com>: > >> > >>> On 26/06/14 16:42, Alexander Graf wrote: > >>> > >>>> On 26.06.14 16:29, Jens Freimann wrote: > >>>> Conny, Alex, Christian, > >>>> > >>>> here are some fixes for the s390-ccw bios. It's a mixture of > >>>> additional features (DASD IPL support for different formats) > >>>> and cleanups. > >>> > >>> From a quick glimpse it looks quite clean and straight forward, but I'd > >>> like to make sure we get rid completely of the static sector size > >>> assumption. > >> > >> Should be. I guess s/SECTOR_SIZE/MAX_SECTOR_SIZE/g would be ok for you > >> then? > > > > I'm not 100% convinced that we're safe on all users of SECTOR_SIZE. So > > please make sure to replace the occasions manually and audit every single > > one. > > Yes, a mindless sed, would also replace VIRTIO_SECTOR_SIZE with > VIRTIO_MAX_SECTOR_SIZE. > Fortunately there are only 3 place in bootmap.c. Should be simple enough to > review. Yes, all places that use it want a MAX_SECTOR_SIZE. All places using the actual sector size are now using the helper function. > >>> > >>> Also, are we guaranteed that virtio always uses 512 byte block size? Or > >>> was that just an internal API thing? > >> > >> The virtio-blk API always talks in 512 byte sectors, no matter the block > >> size. > >> > >> Overall this is a nice improvement of the boot code - if possible I would > >> like to see that in 2.1. > >> > >> Conny, can you carry that in your tree (with > >> s/SECTOR_SIZE/MAX_SECTOR_SIZE/g)? > >> > >> Acked-by: Christian Borntraeger <borntrae...@de.ibm.com> > >> > >> for the series. Will push out shortly. Unless there are objections, I'll send a pull request for this.