mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Brendan Jackman <jackmanb@google.com>
To: Borislav Petkov <bp@alien8.de>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	 Peter Zijlstra <peterz@infradead.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	 David Hildenbrand <david@kernel.org>,
	Vlastimil Babka <vbabka@kernel.org>,
	Mike Rapoport <rppt@kernel.org>,  Wei Xu <weixugc@google.com>,
	Johannes Weiner <hannes@cmpxchg.org>, Zi Yan <ziy@nvidia.com>,
	 Lorenzo Stoakes <ljs@kernel.org>
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org, x86@kernel.org,
	 rppt@kernel.org, Sumit Garg <sumit.garg@oss.qualcomm.com>,
	 Will Deacon <will@kernel.org>,
	rientjes@google.com,  "Kalyazin, Nikita" <kalyazin@amazon.co.uk>,
	patrick.roy@linux.dev,  "Itazuri, Takahiro" <itazur@amazon.co.uk>,
	Andy Lutomirski <luto@kernel.org>,
	 David Kaplan <david.kaplan@amd.com>,
	Thomas Gleixner <tglx@kernel.org>, Yosry Ahmed <yosry@kernel.org>,
	 Patrick Bellasi <derkling@google.com>,
	Reiji Watanabe <reijiw@google.com>,
	 Sean Christopherson <seanjc@google.com>,
	Brendan Jackman <jackmanb@google.com>
Subject: [PATCH v3 20/26] mm/page_alloc: introduce ALLOC_NOBLOCK
Date: Sun, 26 Jul 2026 22:22:53 +0000	[thread overview]
Message-ID: <20260726-page_alloc-unmapped-v3-20-6f5729aa9832@google.com> (raw)
In-Reply-To: <20260726-page_alloc-unmapped-v3-0-6f5729aa9832@google.com>

This flag is set unless we can be sure the caller isn't in an atomic
context.

The allocator will soon start needing to call set_direct_map_* APIs
which cannot be called with IRQs off. It will need to do this even
before direct reclaim is possible.

Despite the fact that, in principle, ALLOC_NOBLOCK is distinct from
__GFP_DIRECT_RECLAIM, in order to avoid introducing a GFP flag, just
infer the former based on whether the caller set the latter. This means
that, in practice, ALLOC_NOBLOCK is just !__GFP_DIRECT_RECLAIM, except
that it is not influenced by gfp_allowed_mask. This requires some rather
ugly plumbing to get it into the slowpath without clobbering it.

Call it ALLOC_NOBLOCK in order to try and mitigate confusion vs the
recently-removed ALLOC_NON_BLOCK, which meant something different.

Signed-off-by: Brendan Jackman <jackmanb@google.com>
---
 mm/page_alloc.c | 23 +++++++++++++++++++----
 mm/page_alloc.h |  1 +
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index fb522a09f2e60..ecdd3780192e8 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5286,6 +5286,19 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	return page;
 }
 
+/* Set alloc flags before applying gfp_allowed_mask. */
+static inline unsigned int init_alloc_flags(gfp_t gfp_mask, unsigned int alloc_flags)
+{
+	/*
+	 * If the caller allowed __GFP_DIRECT_RECLAIM, they can't be atomic.
+	 * Note this is a separate determination from whether direct reclaim is
+	 * actually allowed, it must happen before applying gfp_allowed_mask.
+	 */
+	if (!(gfp_mask & __GFP_DIRECT_RECLAIM))
+		alloc_flags |= ALLOC_NOBLOCK;
+	return alloc_flags;
+}
+
 static inline bool prepare_alloc_pages(gfp_t gfp_mask, unsigned int order,
 		int preferred_nid, nodemask_t *nodemask,
 		struct alloc_context *ac, gfp_t *alloc_gfp,
@@ -5364,8 +5377,10 @@ unsigned long alloc_pages_bulk_noprof(gfp_t gfp, int preferred_nid,
 	struct zoneref *z;
 	struct per_cpu_pages *pcp;
 	struct list_head *pcp_list;
-	struct alloc_context ac;
-	unsigned int alloc_flags = ALLOC_WMARK_LOW;
+	struct alloc_context ac = {
+		.alloc_flags = init_alloc_flags(gfp, ALLOC_DEFAULT),
+	};
+	unsigned int alloc_flags = ac.alloc_flags | ALLOC_WMARK_LOW;
 	int nr_populated = 0, nr_account = 0;
 
 	/*
@@ -5584,9 +5599,9 @@ struct page *__alloc_frozen_pages_noprof(gfp_t gfp, unsigned int order,
 	struct page *page;
 	gfp_t alloc_gfp; /* The gfp_t that was actually used for allocation */
 	struct alloc_context ac = {
-		.alloc_flags = alloc_flags,
+		.alloc_flags = init_alloc_flags(gfp, alloc_flags),
 	};
-	unsigned int fastpath_alloc_flags = alloc_flags;
+	unsigned int fastpath_alloc_flags = ac.alloc_flags;
 
 	/* Other flags could be supported later if needed. */
 	if (WARN_ON(alloc_flags & ~(ALLOC_NOLOCK | ALLOC_NO_CODETAG)))
diff --git a/mm/page_alloc.h b/mm/page_alloc.h
index 2307b173a459f..02db12c9a1dc2 100644
--- a/mm/page_alloc.h
+++ b/mm/page_alloc.h
@@ -75,6 +75,7 @@
  */
 #define ALLOC_UNMAPPED	       0x2000
 #endif
+#define ALLOC_NOBLOCK	       0x4000 /* Caller may be atomic */
 
 /* Flags that allow allocations below the min watermark. */
 #define ALLOC_RESERVES (ALLOC_HARDER|ALLOC_MIN_RESERVE|ALLOC_HIGHATOMIC|ALLOC_OOM)

-- 
2.54.0


  parent reply	other threads:[~2026-07-26 22:23 UTC|newest]

Thread overview: 119+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-26 22:22 [PATCH v3 00/26] mm: Add ALLOC_UNMAPPED and AS_NO_DIRECT_MAP Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 01/26] set_memory: add folio_{zap,restore}_direct_map helpers Brendan Jackman
2026-07-27 10:33   ` Mike Rapoport
2026-07-29 11:42     ` Brendan Jackman
2026-07-30 20:34   ` Yosry Ahmed
2026-07-31  5:21     ` Mike Rapoport
2026-07-31 11:57       ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 02/26] mm/secretmem: make use of folio_{zap,restore}_direct_map Brendan Jackman
2026-07-27 10:40   ` Mike Rapoport
2026-07-26 22:22 ` [PATCH v3 03/26] mm: introduce AS_NO_DIRECT_MAP Brendan Jackman
2026-07-30 21:06   ` Yosry Ahmed
2026-07-31 12:15     ` Brendan Jackman
2026-07-31 19:28       ` Yosry Ahmed
2026-08-07  0:02         ` Sean Christopherson
2026-08-07  0:13           ` Yosry Ahmed
2026-08-07  0:19             ` Sean Christopherson
2026-08-07  0:29               ` Yosry Ahmed
2026-08-07 14:26                 ` Sean Christopherson
2026-08-07 18:12                   ` Yosry Ahmed
2026-08-07 18:49                     ` Sean Christopherson
2026-08-07 19:39                       ` Yosry Ahmed
2026-08-07 22:44                         ` Sean Christopherson
2026-08-07 22:48                           ` Yosry Ahmed
2026-09-03  7:26                             ` Takahiro Itazuri
2026-09-03 14:06                               ` Yosry Ahmed
2026-09-03 14:25                                 ` Takahiro Itazuri
2026-09-15 16:01                                 ` Gregory Price
2026-08-08 13:56         ` Brendan Jackman
2026-08-10 21:39           ` Yosry Ahmed
2026-08-10 21:46             ` Sean Christopherson
2026-08-13 15:32         ` Brendan Jackman
2026-08-13 17:25           ` Ackerley Tng
2026-08-18  0:32           ` Yosry Ahmed
2026-08-18 10:41             ` Brendan Jackman
2026-08-02 16:10   ` Mike Rapoport
2026-08-08  0:19   ` Yosry Ahmed
2026-07-26 22:22 ` [PATCH v3 04/26] x86/mm: split out preallocate_sub_pgd() Brendan Jackman
2026-07-31 22:10   ` Yosry Ahmed
2026-08-13 15:42     ` Brendan Jackman
2026-08-18  0:30       ` Yosry Ahmed
2026-08-02 16:13   ` Mike Rapoport
2026-08-13 15:46     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 05/26] x86: move PAE PMD preallocation defines to header Brendan Jackman
2026-07-31 23:59   ` Yosry Ahmed
2026-08-13 15:49     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 06/26] x86/tlb: Expose some flush function declarations to modules Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 07/26] x86/mm: introduce mm-local region Brendan Jackman
2026-08-02 16:27   ` Mike Rapoport
2026-08-13 16:10     ` Brendan Jackman
2026-08-03 22:29   ` Yosry Ahmed
2026-08-13 16:23     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 08/26] x86/mm: move LDT remap into " Brendan Jackman
2026-08-03 22:33   ` Yosry Ahmed
2026-07-26 22:22 ` [PATCH v3 09/26] mm: Create flags arg for __apply_to_page_range() Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 10/26] mm: Add more flags " Brendan Jackman
2026-08-04  0:08   ` Yosry Ahmed
2026-08-13 16:40     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 11/26] x86/mm: introduce the mermap Brendan Jackman
2026-08-02 16:40   ` Mike Rapoport
2026-08-13 16:44     ` Brendan Jackman
2026-08-04 18:38   ` Yosry Ahmed
2026-07-26 22:22 ` [PATCH v3 12/26] mm: KUnit tests for " Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 13/26] mm: introduce freetype_t Brendan Jackman
2026-08-04 22:23   ` Yosry Ahmed
2026-08-14 10:37     ` Brendan Jackman
2026-08-18  0:35       ` Yosry Ahmed
2026-09-03  8:00         ` Vlastimil Babka (SUSE)
2026-09-03 13:30           ` Yosry Ahmed
2026-08-04 23:02   ` Yosry Ahmed
2026-08-14 10:48     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 14/26] mm: move migratetype definitions to freetype.h Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 15/26] mm/page_alloc: add support for freetypes with no freelist Brendan Jackman
2026-07-31 14:13   ` Vlastimil Babka (SUSE)
2026-07-26 22:22 ` [PATCH v3 16/26] mm: add definitions for allocating unmapped pages Brendan Jackman
2026-08-04 19:53   ` Yosry Ahmed
2026-08-14 11:22     ` Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 17/26] mm: encode freetype flags in pageblock flags Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 18/26] mm/page_alloc: separate pcplists by freetype flags Brendan Jackman
2026-07-26 22:22 ` [PATCH v3 19/26] mm/page_alloc: rename ALLOC_NON_BLOCK back to _HARDER Brendan Jackman
2026-07-31 14:52   ` Vlastimil Babka (SUSE)
2026-08-03  9:20     ` Vlastimil Babka (SUSE)
2026-08-04 21:50     ` Yosry Ahmed
2026-08-14 12:09       ` Brendan Jackman
2026-08-18  0:45         ` Yosry Ahmed
2026-08-18 10:58           ` Brendan Jackman
2026-07-26 22:22 ` Brendan Jackman [this message]
2026-07-26 22:22 ` [PATCH v3 21/26] mm/page_alloc: implement FREETYPE_UNMAPPED allocations Brendan Jackman
2026-08-03  9:18   ` Vlastimil Babka (SUSE)
2026-08-15 14:12     ` Brendan Jackman
2026-08-04 23:41   ` Yosry Ahmed
2026-08-07  0:05     ` Yosry Ahmed
2026-08-14 12:31     ` Brendan Jackman
2026-08-04 23:53   ` Yosry Ahmed
2026-08-05 16:13     ` Yosry Ahmed
2026-08-15 14:28     ` Brendan Jackman
2026-08-18  0:55       ` Yosry Ahmed
2026-08-18 11:03         ` Brendan Jackman
2026-08-18 17:51           ` Yosry Ahmed
2026-08-25 20:01             ` Yosry Ahmed
2026-08-07  0:16   ` Yosry Ahmed
2026-08-15 14:30     ` Brendan Jackman
2026-08-18  0:50       ` Yosry Ahmed
2026-08-12 21:26   ` Yosry Ahmed
2026-08-15 14:43     ` Brendan Jackman
2026-08-18  0:49       ` Yosry Ahmed
2026-08-18  1:01   ` Yosry Ahmed
2026-07-26 22:22 ` [PATCH v3 22/26] mm: Minimal KUnit tests for some new page_alloc logic Brendan Jackman
2026-08-03  9:30   ` Vlastimil Babka (SUSE)
2026-07-26 22:22 ` [PATCH v3 23/26] mm: Split out NR_FREE_PAGES_BLOCKS_[UN]MAPPED Brendan Jackman
2026-08-03  9:32   ` Vlastimil Babka (SUSE)
2026-07-26 22:22 ` [PATCH v3 24/26] mm/page_alloc: always direct compact for unmapped allocs Brendan Jackman
2026-08-03  9:44   ` Vlastimil Babka (SUSE)
2026-08-15 14:44     ` Brendan Jackman
2026-08-06 23:29   ` Yosry Ahmed
2026-07-26 22:22 ` [PATCH v3 25/26] mm: plumb alloc flags into some alloc funcs Brendan Jackman
2026-08-03  9:52   ` Vlastimil Babka (SUSE)
2026-07-26 22:22 ` [PATCH v3 26/26] mm: add fast path for AS_NO_DIRECT_MAP Brendan Jackman
2026-08-08  0:06   ` Yosry Ahmed
2026-07-29 11:52 ` [PATCH v3 00/26] mm: Add ALLOC_UNMAPPED and AS_NO_DIRECT_MAP Brendan Jackman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260726-page_alloc-unmapped-v3-20-6f5729aa9832@google.com \
    --to=jackmanb@google.com \
    --cc=akpm@linux-foundation.org \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=david.kaplan@amd.com \
    --cc=david@kernel.org \
    --cc=derkling@google.com \
    --cc=hannes@cmpxchg.org \
    --cc=itazur@amazon.co.uk \
    --cc=kalyazin@amazon.co.uk \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=ljs@kernel.org \
    --cc=luto@kernel.org \
    --cc=patrick.roy@linux.dev \
    --cc=peterz@infradead.org \
    --cc=reijiw@google.com \
    --cc=rientjes@google.com \
    --cc=rppt@kernel.org \
    --cc=seanjc@google.com \
    --cc=sumit.garg@oss.qualcomm.com \
    --cc=tglx@kernel.org \
    --cc=vbabka@kernel.org \
    --cc=weixugc@google.com \
    --cc=will@kernel.org \
    --cc=x86@kernel.org \
    --cc=yosry@kernel.org \
    --cc=ziy@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®