Re: [PATCH v7 2/3] kho: fix deferred init of kho scratch

[SeongJae Park <sj@kernel.org>]

On Tue, 17 Mar 2026 16:23:07 -0700 "Vishal Moola (Oracle)" <vishal.moola@gmail.com> wrote:

> On Tue, Mar 17, 2026 at 03:15:33PM +0100, Michal Clapinski wrote:
> > Currently, if DEFERRED is enabled, kho_release_scratch will initialize
> > the struct pages and set migratetype of kho scratch. Unless the whole
> > scratch fit below first_deferred_pfn, some of that will be overwritten
> > either by deferred_init_pages or memmap_init_reserved_pages.
> > 
> > To fix it, I modified kho_release_scratch to only set the migratetype
> > on already initialized pages. Then, modified init_pageblock_migratetype
> > to set the migratetype to CMA if the page is located inside scratch.
> > 
> > Signed-off-by: Michal Clapinski <mclapinski@google.com>
> > ---
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index ee81f5c67c18..5ca078dde61d 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -55,6 +55,7 @@
> >  #include <linux/cacheinfo.h>
> >  #include <linux/pgalloc_tag.h>
> >  #include <linux/mmzone_lock.h>
> > +#include <linux/kexec_handover.h>
> >  #include <asm/div64.h>
> >  #include "internal.h"
> >  #include "shuffle.h"
> > @@ -549,6 +550,12 @@ void __meminit init_pageblock_migratetype(struct page *page,
> >  		     migratetype < MIGRATE_PCPTYPES))
> >  		migratetype = MIGRATE_UNMOVABLE;
> >  
> > +	/*
> > +	 * Mark KHO scratch as CMA so no unmovable allocations are made there.
> > +	 */
> > +	if (unlikely(kho_scratch_overlap(page_to_phys(page), PAGE_SIZE)))
> > +		migratetype = MIGRATE_CMA;
> > +
> >  	flags = migratetype;
> >  
> >  #ifdef CONFIG_MEMORY_ISOLATION
> 
> I've just tried to build the current mm-new tree. I'm getting this
> error:
> 
>  error: ‘MIGRATE_CMA’ undeclared (first use in this function); did you mean ‘MIGRATE_SYNC’?
>   557 |                 migratetype = MIGRATE_CMA;
>       |                               ^~~~~~~~~~~
>       |                               MIGRATE_SYNC
> 
> >From what I can tell, MIGRATE_CMA is only defined if CONFIG_CMA is
> enabled (for x86 defconfig its disabled).

I also got the same issue, and was about to report.  I added below workaround
patch to my test setup.  It is just a temporal quick fix for only my setup.
Michal or others may find a better fix.

=== >8 ===
From 9d89a12e1a17edd68750e97b0c8b1970e3adc648 Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Tue, 17 Mar 2026 16:56:17 -0700
Subject: [PATCH] mm/page_alloc: enclose kho-specific code with
 CONFIG_KEXEC_HANDOVER

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/page_alloc.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 5ca078dde61d6..ed4d585f46202 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -550,11 +550,13 @@ void __meminit init_pageblock_migratetype(struct page *page,
 		     migratetype < MIGRATE_PCPTYPES))
 		migratetype = MIGRATE_UNMOVABLE;
 
+#ifdef CONFIG_KEXEC_HANDOVER
 	/*
 	 * Mark KHO scratch as CMA so no unmovable allocations are made there.
 	 */
 	if (unlikely(kho_scratch_overlap(page_to_phys(page), PAGE_SIZE)))
 		migratetype = MIGRATE_CMA;
+#endif
 
 	flags = migratetype;
 
-- 
2.47.3