linux-kernel - [RFC PATCH 4/5] mm/madvise: add PMADV_SET_FORK_EXEC_DEFAULT process

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <617413860ff194dfb1afedb175b2d84a457e449d.1747686021.git.lorenzo.stoakes@oracle.com>
Date: Mon, 19 May 2025 21:52:41 +0100
From: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: "Liam R . Howlett" <Liam.Howlett@...cle.com>,
        David Hildenbrand <david@...hat.com>, Vlastimil Babka <vbabka@...e.cz>,
        Jann Horn <jannh@...gle.com>, Arnd Bergmann <arnd@...db.de>,
        Christian Brauner <brauner@...nel.org>, linux-mm@...ck.org,
        linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org,
        SeongJae Park <sj@...nel.org>, Usama Arif <usamaarif642@...il.com>
Subject: [RFC PATCH 4/5] mm/madvise: add PMADV_SET_FORK_EXEC_DEFAULT process_madvise() flag

It's useful in certain cases to be able to default-enable an madvise() flag
for all newly mapped VMAs, and for that to survive fork/exec.

The natural place to specify something like this is in an madvise()
invocation, and thus providing this functionality as a flag to
process_madvise() makes sense.

We intentionally limit this only to flags that we know should function
correctly without issue, and to be conservative about this, so we initially
limit ourselves only to MADV_HUGEPAGE, MADV_NOHUGEPAGE, that is - setting
the VM_HUGEPAGE, VM_NOHUGEPAGE VMA flags.

We implement this functionality by using the mm_struct->def_flags field.

Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
---
 include/uapi/asm-generic/mman-common.h |  1 +
 mm/madvise.c                           | 43 +++++++++++++++++++++++++-
 2 files changed, 43 insertions(+), 1 deletion(-)

diff --git a/include/uapi/asm-generic/mman-common.h b/include/uapi/asm-generic/mman-common.h
index 58c8a3fadf99..6998ea0ecc6d 100644
--- a/include/uapi/asm-generic/mman-common.h
+++ b/include/uapi/asm-generic/mman-common.h
@@ -94,5 +94,6 @@
 /* process_madvise() flags */
 #define PMADV_SKIP_ERRORS (1U << 0) /* Skip VMAs on errors, but carry on. Implies no error on unmapped. */
 #define PMADV_NO_ERROR_ON_UNMAPPED (1U << 1) /* Never report an error on unmapped ranges. */
+#define PMADV_SET_FORK_EXEC_DEFAULT (1U << 2) /* Set the behavior as a default that survives fork/exec. */
 
 #endif /* __ASM_GENERIC_MMAN_COMMON_H */
diff --git a/mm/madvise.c b/mm/madvise.c
index fd94ef27f909..9ea36800de3c 100644
--- a/mm/madvise.c
+++ b/mm/madvise.c
@@ -1869,6 +1869,40 @@ SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
 	return do_madvise(current->mm, start, len_in, behavior);
 }
 
+/*
+ * Are we permitted to set an madvise() behavior by default across the virtual
+ * address space, surviving fork/exec?
+ */
+static bool can_set_default_behavior(int behavior)
+{
+	switch (behavior) {
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+	case MADV_HUGEPAGE:
+	case MADV_NOHUGEPAGE:
+		return true;
+#endif
+	default:
+		return false;
+	}
+}
+
+static void set_default_behavior(struct mm_struct *mm, int behavior)
+{
+	switch (behavior) {
+	case MADV_HUGEPAGE:
+		mm->def_flags &= ~VM_NOHUGEPAGE;
+		mm->def_flags |= VM_HUGEPAGE;
+		break;
+	case MADV_NOHUGEPAGE:
+		mm->def_flags &= ~VM_HUGEPAGE;
+		mm->def_flags |= VM_NOHUGEPAGE;
+		break;
+	default:
+		WARN_ON(1);
+		break;
+	}
+}
+
 /* Perform an madvise operation over a vector of addresses and lengths. */
 static ssize_t vector_madvise(struct mm_struct *mm, struct iov_iter *iter,
 			      int behavior, int flags)
@@ -1882,8 +1916,12 @@ static ssize_t vector_madvise(struct mm_struct *mm, struct iov_iter *iter,
 		.flags = flags,
 	};
 	bool can_skip = flags & PMADV_SKIP_ERRORS;
+	bool set_default = flags & PMADV_SET_FORK_EXEC_DEFAULT;
 	size_t skipped = 0;
 
+	if (set_default && !can_set_default_behavior(behavior))
+		return -EINVAL;
+
 	total_len = iov_iter_count(iter);
 
 	ret = madvise_lock(mm, behavior);
@@ -1931,6 +1969,8 @@ static ssize_t vector_madvise(struct mm_struct *mm, struct iov_iter *iter,
 		if (can_skip && madv_behavior.saw_error) {
 			skipped++;
 			madv_behavior.saw_error = false;
+		} else if (set_default) {
+			set_default_behavior(mm, behavior);
 		}
 
 		iov_iter_advance(iter, iter_iov_len(iter));
@@ -1951,7 +1991,8 @@ static ssize_t vector_madvise(struct mm_struct *mm, struct iov_iter *iter,
 
 static bool check_process_madvise_flags(unsigned int flags)
 {
-	unsigned int mask = PMADV_SKIP_ERRORS | PMADV_NO_ERROR_ON_UNMAPPED;
+	unsigned int mask = PMADV_SKIP_ERRORS | PMADV_NO_ERROR_ON_UNMAPPED |
+		PMADV_SET_FORK_EXEC_DEFAULT;
 
 	if (flags & ~mask)
 		return false;
-- 
2.49.0