On Thu, Aug 01, 2019 at 02:20:07PM +0800, Lu Baolu wrote: > Hi Christoph, > > On 8/1/19 2:10 PM, Christoph Hellwig wrote: >> On Thu, Aug 01, 2019 at 02:01:54PM +0800, Lu Baolu wrote: >>> + /* No lock here, assumes no domain exit in normal case */ >> >> s/exit/exists/ ? > > This comment is just moved from one place to another in this patch. > > "no domain exit" means "the domain isn't freed". (my understand)
Maybe we'll get that refconfirmed and can fix up the comment? > >> >>> + info = dev->archdata.iommu; >>> + if (likely(info)) >>> + return info->domain; >> >> But then again the likely would be odd. >> > > Normally there's a domain for a device (default domain or isolation > domain for assignment cases). Makes sense, I just mean to say that the likely was contrary to my understanding of the above comment.
