linux-kernel - [PATCH 7/7] uprobes: deny mremap(xol

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: <20240929144258.GA9492@redhat.com>
Date: Sun, 29 Sep 2024 16:42:58 +0200
From: Oleg Nesterov <oleg@...hat.com>
To: Andrii Nakryiko <andrii@...nel.org>, Jiri Olsa <jolsa@...nel.org>,
	Masami Hiramatsu <mhiramat@...nel.org>,
	Peter Zijlstra <peterz@...radead.org>
Cc: Liao Chang <liaochang1@...wei.com>, linux-kernel@...r.kernel.org,
	linux-trace-kernel@...r.kernel.org
Subject: [PATCH 7/7] uprobes: deny mremap(xol_vma)

kernel/events/uprobes.c assumes that xol_area->vaddr is always correct but
a malicious application can remap its "[uprobes]" vma to another adress to
confuse the kernel. Introduce xol_mremap() to make this impossible.

With this change utask->xol_vaddr in xol_free_insn_slot() can't be invalid,
we can turn the offset check into WARN_ON_ONCE(offset >= PAGE_SIZE).

Signed-off-by: Oleg Nesterov <oleg@...hat.com>
---
 kernel/events/uprobes.c | 30 +++++++++++++++++-------------
 1 file changed, 17 insertions(+), 13 deletions(-)

diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index da45d0e5bcf4..20c58b6ee1ad 100644
--- a/kernel/events/uprobes.c
+++ b/kernel/events/uprobes.c
@@ -1475,9 +1475,15 @@ static vm_fault_t xol_fault(const struct vm_special_mapping *sm,
 	return 0;
 }
 
+static int xol_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma)
+{
+	return -EPERM;
+}
+
 static const struct vm_special_mapping xol_mapping = {
 	.name = "[uprobes]",
 	.fault = xol_fault,
+	.mremap = xol_mremap,
 };
 
 /* Slot allocation for XOL */
@@ -1670,21 +1676,19 @@ static void xol_free_insn_slot(struct uprobe_task *utask)
 {
 	struct xol_area *area = current->mm->uprobes_state.xol_area;
 	unsigned long offset = utask->xol_vaddr - area->vaddr;
+	unsigned int slot_nr;
 
 	utask->xol_vaddr = 0;
-	/*
-	 * xol_vaddr must fit into [area->vaddr, area->vaddr + PAGE_SIZE).
-	 * This check can only fail if the "[uprobes]" vma was mremap'ed.
-	 */
-	if (offset < PAGE_SIZE) {
-		int slot_nr = offset / UPROBE_XOL_SLOT_BYTES;
-
-		clear_bit(slot_nr, area->bitmap);
-		atomic_dec(&area->slot_count);
-		smp_mb__after_atomic(); /* pairs with prepare_to_wait() */
-		if (waitqueue_active(&area->wq))
-			wake_up(&area->wq);
-	}
+	/* xol_vaddr must fit into [area->vaddr, area->vaddr + PAGE_SIZE) */
+	if (WARN_ON_ONCE(offset >= PAGE_SIZE))
+		return;
+
+	slot_nr = offset / UPROBE_XOL_SLOT_BYTES;
+	clear_bit(slot_nr, area->bitmap);
+	atomic_dec(&area->slot_count);
+	smp_mb__after_atomic(); /* pairs with prepare_to_wait() */
+	if (waitqueue_active(&area->wq))
+		wake_up(&area->wq);
 }
 
 void __weak arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
-- 
2.25.1.362.g51ebf55