linux-kernel - [PATCH 10/14] mm: page_vma_mapped_walk: map_pte() use pte_offset_map_maywrite

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 for Android: free password hash cracker in your pocket

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

Message-Id: <e23662d6be464b6ed89d6abde13df1b4694f6583.1724226076.git.zhengqi.arch@bytedance.com>
Date: Wed, 21 Aug 2024 16:18:53 +0800
From: Qi Zheng <zhengqi.arch@...edance.com>
To: david@...hat.com,
	hughd@...gle.com,
	willy@...radead.org,
	muchun.song@...ux.dev,
	vbabka@...nel.org,
	akpm@...ux-foundation.org,
	rppt@...nel.org,
	vishal.moola@...il.com,
	peterx@...hat.com,
	ryan.roberts@....com
Cc: linux-kernel@...r.kernel.org,
	linux-mm@...ck.org,
	linux-arm-kernel@...ts.infradead.org,
	linuxppc-dev@...ts.ozlabs.org,
	Qi Zheng <zhengqi.arch@...edance.com>
Subject: [PATCH 10/14] mm: page_vma_mapped_walk: map_pte() use pte_offset_map_maywrite_nolock()

In the caller of map_pte(), we may modify the pvmw->pte after acquiring
the pvmw->ptl, so convert it to using pte_offset_map_maywrite_nolock(). At
this time, the write lock of mmap_lock is not held, and the pte_same()
check is not performed after the pvmw->ptl held, so we should get pmdval
and do pmd_same() check to ensure the stability of pvmw->pmd.

Signed-off-by: Qi Zheng <zhengqi.arch@...edance.com>
---
 mm/page_vma_mapped.c | 24 ++++++++++++++++++++----
 1 file changed, 20 insertions(+), 4 deletions(-)

diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c
index ae5cc42aa2087..da806f3a5047d 100644
--- a/mm/page_vma_mapped.c
+++ b/mm/page_vma_mapped.c
@@ -13,9 +13,11 @@ static inline bool not_found(struct page_vma_mapped_walk *pvmw)
 	return false;
 }
 
-static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
+static bool map_pte(struct page_vma_mapped_walk *pvmw, pmd_t *pmdvalp,
+		    spinlock_t **ptlp)
 {
 	pte_t ptent;
+	pmd_t pmdval;
 
 	if (pvmw->flags & PVMW_SYNC) {
 		/* Use the stricter lookup */
@@ -25,6 +27,7 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
 		return !!pvmw->pte;
 	}
 
+again:
 	/*
 	 * It is important to return the ptl corresponding to pte,
 	 * in case *pvmw->pmd changes underneath us; so we need to
@@ -32,10 +35,11 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
 	 * proceeds to loop over next ptes, and finds a match later.
 	 * Though, in most cases, page lock already protects this.
 	 */
-	pvmw->pte = pte_offset_map_nolock(pvmw->vma->vm_mm, pvmw->pmd,
-					  pvmw->address, ptlp);
+	pvmw->pte = pte_offset_map_maywrite_nolock(pvmw->vma->vm_mm, pvmw->pmd,
+						   pvmw->address, &pmdval, ptlp);
 	if (!pvmw->pte)
 		return false;
+	*pmdvalp = pmdval;
 
 	ptent = ptep_get(pvmw->pte);
 
@@ -69,6 +73,12 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
 	}
 	pvmw->ptl = *ptlp;
 	spin_lock(pvmw->ptl);
+
+	if (unlikely(!pmd_same(pmdval, pmdp_get_lockless(pvmw->pmd)))) {
+		spin_unlock(pvmw->ptl);
+		goto again;
+	}
+
 	return true;
 }
 
@@ -278,7 +288,7 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
 			step_forward(pvmw, PMD_SIZE);
 			continue;
 		}
-		if (!map_pte(pvmw, &ptl)) {
+		if (!map_pte(pvmw, &pmde, &ptl)) {
 			if (!pvmw->pte)
 				goto restart;
 			goto next_pte;
@@ -307,6 +317,12 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
 		if (!pvmw->ptl) {
 			pvmw->ptl = ptl;
 			spin_lock(pvmw->ptl);
+			if (unlikely(!pmd_same(pmde, pmdp_get_lockless(pvmw->pmd)))) {
+				pte_unmap_unlock(pvmw->pte, pvmw->ptl);
+				pvmw->ptl = NULL;
+				pvmw->pte = NULL;
+				goto restart;
+			}
 		}
 		goto this_pte;
 	} while (pvmw->address < end);
-- 
2.20.1