On Thu, May 23, 2019 at 10:56:09AM -0700, Mike Kravetz wrote: > On 5/22/19 4:51 PM, Jason Gunthorpe wrote: > > On Wed, May 22, 2019 at 01:23:22PM -0700, Andrew Morton wrote: > >> > >> Also fair enough. But why the heck is huge_page_shift() a macro? We > >> keep doing that and it bites so often :( > > > > Let's fix it, with the below? (compile tested) > > > > Note __alloc_bootmem_huge_page was returning null but the signature > > was unsigned int. > > > > From b5e2ff3c88e6962d0e8297c87af855e6fe1a584e Mon Sep 17 00:00:00 2001 > > From: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > > Date: Wed, 22 May 2019 20:45:59 -0300 > > Subject: [PATCH] mm: Make !CONFIG_HUGE_PAGE wrappers into static inlines > > > > Instead of using defines, which looses type safety and provokes unused > > variable warnings from gcc, put the constants into static inlines. > > > > Suggested-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > > Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > > Thanks for doing this Jason. > > I do not see any issues unless there is some weird arch specific usage which > would be caught by zero day testing. Agreed. I did a couple quick searches and I don't see any such issues. I was thinking the same thing WRT zero day. Reviewed-by: Ira Weiny <ira.weiny@xxxxxxxxx> > > Reviewed-by: Mike Kravetz <mike.kravetz@xxxxxxxxxx> > > -- > Mike Kravetz >
