On 7/26/19 11:56 AM, Linus Torvalds wrote:

Guys, what happened to the wrong sector boundary and max sector mess?

There are at least two different issues (with one of them having two
proposed fixes);

   https://lore.kernel.org/linux-block/1563896932.3609.15.camel@HansenPartnership.com/ (local)
   https://lore.kernel.org/lkml/1563895995.3609.10.camel@HansenPartnership.com/ (local)
   https://lore.kernel.org/lkml/1563839144.2504.5.camel@HansenPartnership.com/ (local)

but I don't actually seem to have a pull request for any of this.

The

    dma_max_mapping_size(dev) << SECTOR_SHIFT

this in scsi_lib.c is clearly completely wrong, and is still there in my tree.

The virt_boundary_mask thing can apparently be fixed other ways too
(ie questionable whether it should be fixed in block/blk-settings.c or
in drivers/scsi/scsi_lib.c, but I don't see either one. I was
expecting the block/blk-settings.c one to be in this pull request.

Maybe I'm missing some alternate fix? I don't think so, particularly
since I still see the wrong-way shift, at least.

Is this due to some confusion about who is supposed to fix it?
Christoph was involved in both, issues, and the problems came throigh
different trees (ie block tree for the virt_boundary_mask, scsi with
the

The fix is sitting in the SCSI tree:

https://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi.git/commit/?h=fixes&id=1b5d9a6e98350e0713b4faa1b04e8f239f63b581

so hopefully it'll be pushed upstream in time for -rc2...

-- 
Jens Axboe