From: Martin K. Petersen <martin.petersen@oracle.com>
Date: Wed, 8 Aug 2018 11:49:27 +0000 (-0400)
Subject: oracleasm: Don't assume bip was allocated by oracleasm
X-Git-Tag: v4.1.12-124.31.3~584
X-Git-Url: https://www.infradead.org/git/?a=commitdiff_plain;h=95a56c4e00e1a436fb362f5eafab3ad0f7fede08;p=users%2Fjedix%2Flinux-maple.git

oracleasm: Don't assume bip was allocated by oracleasm

Orabug: 28506080

If a user runs an old ASMLIB without integrity support, the bip
attached to a bio is owned by the block layer. We should not attempt
to unmap the pages in question.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Brian Maly <brian.maly@oracle.com>
---

diff --git a/drivers/block/oracleasm/integrity.c b/drivers/block/oracleasm/integrity.c
index 61263f79869f..96ce08c72292 100644
--- a/drivers/block/oracleasm/integrity.c
+++ b/drivers/block/oracleasm/integrity.c
@@ -220,6 +220,13 @@ void asm_integrity_unmap(struct bio *bio)
 	if (!bip)
 		return;
 
+	/*
+	 * We may end up here if the user is running an old ASMLIB
+	 * without integrity support.
+	 */
+	if (bip->bip_flags & BIP_BLOCK_INTEGRITY)
+		return;
+
 	bip_for_each_vec(iv, bio_integrity(bio), iter) {
 		if (bio_data_dir(bio) == READ)
 			set_page_dirty_lock(iv.bv_page);