------- Comment From <email address hidden> 2020-11-20 09:50 EDT-------
(In reply to comment #14)
> suggested bionic backport for /mm/page_idle.c
Hmm, not sure if I read this diff correcty, but it seems to remove struct zone and spinlock, which would not be right, and introduce a new bug.
The reason why the upstream commit does not apply directly to bionic, is because of conflicting context. The "struct zone" was replaced with "pg_data_t *pgdat" by (another) commit 92fb1db26eef, but that change (or any change to the struct zone) would not be necessary to fix the uninitialized struct page access.
So, I would suggest simply replacing the line
pg_data_t *pgdat;
from the upstream commit context, with this line to fix the context for bionic
struct zone *zone;
Removing struct zone and the spinlock would certainly introduce a new bug, because the spinlock is necessary.
------- Comment From <email address hidden> 2020-11-20 09:50 EDT-------
(In reply to comment #14)
> suggested bionic backport for /mm/page_idle.c
Hmm, not sure if I read this diff correcty, but it seems to remove struct zone and spinlock, which would not be right, and introduce a new bug.
The reason why the upstream commit does not apply directly to bionic, is because of conflicting context. The "struct zone" was replaced with "pg_data_t *pgdat" by (another) commit 92fb1db26eef, but that change (or any change to the struct zone) would not be necessary to fix the uninitialized struct page access.
So, I would suggest simply replacing the line
pg_data_t *pgdat;
from the upstream commit context, with this line to fix the context for bionic
struct zone *zone;
Removing struct zone and the spinlock would certainly introduce a new bug, because the spinlock is necessary.