linux-kernel - [PATCH] lightnvm: pblk: fix race condition on GC

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-next>] [day] [month] [year] [list]

Date:   Sat, 26 Jan 2019 22:54:21 -0800
From:   Heiner Litz <hlitz@...c.edu>
To:     mb@...htnvm.io
Cc:     javier@...igon.com, hans.holmberg@...xlabs.com,
        linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
        Heiner Litz <hlitz@...c.edu>
Subject: [PATCH] lightnvm: pblk: fix race condition on GC

This patch fixes a race condition where a write is mapped to the last
sectors of a line. The write is synced to the device but the L2P is not
updated yet. When the line is garbage collected before the L2P update is
performed, the sectors are ignored by the GC logic and the line is freed
before all sectors are moved. When the L2P is finally updated, it contains
a mapping to a freed line, subsequent reads of the corresponding LBAs fail.

Note that looking up the L2P and checking the ppa in the write buffer needs
to be performed atomically, hence the refactor of pblk_lookup_l2p_rand.

Signed-off-by: Heiner Litz <hlitz@...c.edu>
---
 drivers/lightnvm/pblk-read.c | 27 +++++++++++++++++++++++++--
 1 file changed, 25 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 3789185144da..7c556b2218e4 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -529,13 +529,35 @@ static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
 	int valid_secs = 0;
 	int i;
 
-	pblk_lookup_l2p_rand(pblk, ppa_list_l2p, lba_list, nr_secs);
-
+	spin_lock(&pblk->trans_lock);
 	for (i = 0; i < nr_secs; i++) {
 		if (lba_list[i] == ADDR_EMPTY)
 			continue;
 
+		ppa_list_l2p[i] = pblk_trans_map_get(pblk, lba_list[i]);
 		ppa_gc = addr_to_gen_ppa(pblk, paddr_list_gc[i], line->id);
+
+		/* Obtain ppa from cache if the sector has been synced to the
+		   device but the L2P has not been updated yet */
+		if(pblk_addr_in_cache(ppa_list_l2p[i])) {
+			struct pblk_rb *rb = &pblk->rwb;
+			struct pblk_rb_entry *entry;
+			struct pblk_w_ctx *w_ctx;
+			u64 pos = pblk_addr_to_cacheline(ppa_list_l2p[i]);
+
+#ifdef CONFIG_NVM_PBLK_DEBUG
+			/* Ensure that the access will not cause an overflow */
+			BUG_ON(pos >= rb->nr_entries);
+#endif
+
+			entry = &rb->entries[pos];
+			w_ctx = &entry->w_ctx;
+			if (pblk_ppa_comp(w_ctx->ppa, ppa_gc)) {
+				rqd->ppa_list[valid_secs++] = ppa_gc;
+				continue;
+			}
+		}
+
 		if (!pblk_ppa_comp(ppa_list_l2p[i], ppa_gc)) {
 			paddr_list_gc[i] = lba_list[i] = ADDR_EMPTY;
 			continue;
@@ -543,6 +565,7 @@ static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
 
 		rqd->ppa_list[valid_secs++] = ppa_list_l2p[i];
 	}
+	spin_unlock(&pblk->trans_lock);
 
 #ifdef CONFIG_NVM_PBLK_DEBUG
 	atomic_long_add(valid_secs, &pblk->inflight_reads);
-- 
2.17.1