Re: [PATCH 2/2] mm: drop PF_MEMALLOC_NORECLAIM

[Michal Hocko <mhocko@suse.com>]

On Mon 26-08-24 18:49:23, Matthew Wilcox wrote:
> On Mon, Aug 26, 2024 at 06:51:55PM +0200, Michal Hocko wrote:
[...]
> > If a plan revert is preferably, I will go with it.
> 
> There aren't any other users of PF_MEMALLOC_NOWARN and it definitely
> seems like something you want at a callsite rather than blanket for every
> allocation below this point.  We don't seem to have many PF_ flags left,
> so let's not keep it around if there's no immediate plans for it.

Good point. What about this?
--- 
From 923cd429d4b1a3520c93bcf46611ae74a3158865 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.com>
Date: Mon, 26 Aug 2024 21:15:02 +0200
Subject: [PATCH] Revert "mm: introduce PF_MEMALLOC_NORECLAIM,
 PF_MEMALLOC_NOWARN"

This reverts commit eab0af905bfc3e9c05da2ca163d76a1513159aa4.

There is no existing user of those flags. PF_MEMALLOC_NOWARN is
dangerous because a nested allocation context can use GFP_NOFAIL which
could cause unexpected failure. Such a code would be hard to maintain
because it could be deeper in the call chain.

PF_MEMALLOC_NORECLAIM has been added even when it was pointed out [1]
that such a allocation contex is inherently unsafe if the context
doesn't fully control all allocations called from this context.

While PF_MEMALLOC_NOWARN is not dangerous the way PF_MEMALLOC_NORECLAIM
is it doesn't have any user and as Matthew has pointed out we are
running out of those flags so better reclaim it without any real users.

[1] https://lore.kernel.org/all/ZcM0xtlKbAOFjv5n@tiehlicka/

Signed-off-by: Michal Hocko <mhocko@suse.com>
---
 include/linux/sched.h    |  4 ++--
 include/linux/sched/mm.h | 17 ++++-------------
 2 files changed, 6 insertions(+), 15 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index f8d150343d42..731ff1078c9e 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1657,8 +1657,8 @@ extern struct pid *cad_pid;
 						 * I am cleaning dirty pages from some other bdi. */
 #define PF_KTHREAD		0x00200000	/* I am a kernel thread */
 #define PF_RANDOMIZE		0x00400000	/* Randomize virtual address space */
-#define PF_MEMALLOC_NORECLAIM	0x00800000	/* All allocation requests will clear __GFP_DIRECT_RECLAIM */
-#define PF_MEMALLOC_NOWARN	0x01000000	/* All allocation requests will inherit __GFP_NOWARN */
+#define PF__HOLE__00800000	0x00800000
+#define PF__HOLE__01000000	0x01000000
 #define PF__HOLE__02000000	0x02000000
 #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_mask */
 #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
diff --git a/include/linux/sched/mm.h b/include/linux/sched/mm.h
index 91546493c43d..07c4fde32827 100644
--- a/include/linux/sched/mm.h
+++ b/include/linux/sched/mm.h
@@ -258,25 +258,16 @@ static inline gfp_t current_gfp_context(gfp_t flags)
 {
 	unsigned int pflags = READ_ONCE(current->flags);
 
-	if (unlikely(pflags & (PF_MEMALLOC_NOIO |
-			       PF_MEMALLOC_NOFS |
-			       PF_MEMALLOC_NORECLAIM |
-			       PF_MEMALLOC_NOWARN |
-			       PF_MEMALLOC_PIN))) {
+	if (unlikely(pflags & (PF_MEMALLOC_NOIO | PF_MEMALLOC_NOFS | PF_MEMALLOC_PIN))) {
 		/*
-		 * Stronger flags before weaker flags:
-		 * NORECLAIM implies NOIO, which in turn implies NOFS
+		 * NOIO implies both NOIO and NOFS and it is a weaker context
+		 * so always make sure it makes precedence
 		 */
-		if (pflags & PF_MEMALLOC_NORECLAIM)
-			flags &= ~__GFP_DIRECT_RECLAIM;
-		else if (pflags & PF_MEMALLOC_NOIO)
+		if (pflags & PF_MEMALLOC_NOIO)
 			flags &= ~(__GFP_IO | __GFP_FS);
 		else if (pflags & PF_MEMALLOC_NOFS)
 			flags &= ~__GFP_FS;
 
-		if (pflags & PF_MEMALLOC_NOWARN)
-			flags |= __GFP_NOWARN;
-
 		if (pflags & PF_MEMALLOC_PIN)
 			flags &= ~__GFP_MOVABLE;
 	}
-- 
2.46.0


-- 
Michal Hocko
SUSE Labs