commit 0188171d757596637de6ba02762c94a27518905a
author Alexei Starovoitov <ast@kernel.org> Sun Apr 06 15:22:23 2025 -0700
committer Alexei Starovoitov <ast@kernel.org> Wed Apr 30 20:09:11 2025 -0700
tree 1532d7db47d1ca1b80a475b40f714d9934acd947
parent 0693046ce49132d88e8bb968896533167cd9a0de

mm: Allow GFP_ACCOUNT and GFP_COMP to be used in alloc_pages_nolock().

Allow __GFP_ACCOUNT and __GFP_COMP flags to be specified when calling
alloc_pages_nolock(), since upcoming reentrant alloc_slab_page() needs
to allocate __GFP_COMP pages while BPF infra needs __GFP_ACCOUNT.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>

include/linux/gfp.h

diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index be160e8..9afbe5b 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -354,7 +354,7 @@ static inline struct page *alloc_page_vma_noprof(gfp_t gfp,
 }
 #define alloc_page_vma(...)			alloc_hooks(alloc_page_vma_noprof(__VA_ARGS__))
 
-struct page *alloc_pages_nolock_noprof(int nid, unsigned int order);
+struct page *alloc_pages_nolock_noprof(gfp_t gfp_flags, int nid, unsigned int order);
 #define alloc_pages_nolock(...)			alloc_hooks(alloc_pages_nolock_noprof(__VA_ARGS__))
 
 extern unsigned long get_free_pages_noprof(gfp_t gfp_mask, unsigned int order);

kernel/bpf/syscall.c

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index d0ddba2..83af8fa 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -578,7 +578,7 @@ static bool can_alloc_pages(void)
 static struct page *__bpf_alloc_page(int nid)
 {
 	if (!can_alloc_pages())
-		return alloc_pages_nolock(nid, 0);
+		return alloc_pages_nolock(__GFP_ACCOUNT, nid, 0);
 
 	return alloc_pages_node(nid,
 				GFP_KERNEL | __GFP_ZERO | __GFP_ACCOUNT

mm/page_alloc.c

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 1d77a07..303df20 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -7379,6 +7379,7 @@ static bool __free_unaccepted(struct page *page)
 
 /**
  * alloc_pages_nolock - opportunistic reentrant allocation from any context
+ * @gfp_flags: GFP flags. Only __GFP_ACCOUNT, __GFP_COMP allowed.
  * @nid: node to allocate from
  * @order: allocation order size
  *
@@ -7392,7 +7393,7 @@ static bool __free_unaccepted(struct page *page)
  * Return: allocated page or NULL on failure. NULL does not mean EBUSY or EAGAIN.
  * It means ENOMEM. There is no reason to call it again and expect !NULL.
  */
-struct page *alloc_pages_nolock_noprof(int nid, unsigned int order)
+struct page *alloc_pages_nolock_noprof(gfp_t gfp_flags, int nid, unsigned int order)
 {
 	/*
 	 * Do not specify __GFP_DIRECT_RECLAIM, since direct claim is not allowed.
@@ -7415,11 +7416,12 @@ struct page *alloc_pages_nolock_noprof(int nid, unsigned int order)
 	 * doesn't want to deplete reserves.
 	 */
 	gfp_t alloc_gfp = __GFP_NOWARN | __GFP_ZERO | __GFP_NOMEMALLOC
-			| __GFP_ACCOUNT;
+			| gfp_flags;
 	unsigned int alloc_flags = ALLOC_TRYLOCK;
 	struct alloc_context ac = { };
 	struct page *page;
 
+	VM_WARN_ON_ONCE(gfp_flags & ~(__GFP_ACCOUNT | __GFP_COMP));
 	/*
 	 * In PREEMPT_RT spin_trylock() will call raw_spin_lock() which is
 	 * unsafe in NMI. If spin_trylock() is called from hard IRQ the current
@@ -7462,7 +7464,7 @@ struct page *alloc_pages_nolock_noprof(int nid, unsigned int order)
 	if (page)
 		set_page_refcounted(page);
 
-	if (memcg_kmem_online() && page &&
+	if (memcg_kmem_online() && page && (gfp_flags & __GFP_ACCOUNT) &&
 	    unlikely(__memcg_kmem_charge_page(page, alloc_gfp, order) != 0)) {
 		free_pages_nolock(page, order);
 		page = NULL;