From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37417) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1WPqOU-0004N4-UG for qemu-devel@nongnu.org; Tue, 18 Mar 2014 05:33:16 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1WPqOM-0005Cl-IS for qemu-devel@nongnu.org; Tue, 18 Mar 2014 05:33:10 -0400 Received: from mx1.redhat.com ([209.132.183.28]:56116) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1WPqOM-0005Bn-Ar for qemu-devel@nongnu.org; Tue, 18 Mar 2014 05:33:02 -0400 Date: Tue, 18 Mar 2014 10:32:48 +0100 From: Kevin Wolf Message-ID: <20140318093248.GF4607@noname.str.redhat.com> References: <1395093892-29271-1-git-send-email-mreitz@redhat.com> <1395093892-29271-3-git-send-email-mreitz@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1395093892-29271-3-git-send-email-mreitz@redhat.com> Subject: Re: [Qemu-devel] [PATCH v2 2/2] qcow2: Fix fail path in realloc_refcount_block() List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Max Reitz Cc: Laszlo Ersek , qemu-devel@nongnu.org, Stefan Hajnoczi , =?iso-8859-1?Q?Beno=EEt?= Canet Am 17.03.2014 um 23:04 hat Max Reitz geschrieben: > If qcow2_alloc_clusters() fails, new_offset and ret will both be > negative after the fail label, thus passing the first if condition and > subsequently resulting in a call of qcow2_free_clusters() with an > invalid (negative) offset parameter. Fix this by introducing a new label > "fail_free_cluster" which is only invoked if new_offset is indeed > pointing to a newly allocated cluster that should be cleaned up by > freeing it. > > While we're at it, clean up the whole fail path. qcow2_cache_put() > should (and actually can) never fail, hence the return value can safely > be ignored (aside from asserting that it indeed did not fail). > > Furthermore, there is no reason to give QCOW2_DISCARD_ALWAYS to > qcow2_free_clusters(), a mere QCOW2_DISCARD_OTHER will suffice. > > Ultimately, rename the "fail" label to "done", as it is invoked both on > failure and success. > > Signed-off-by: Max Reitz > Suggested-by: Laszlo Ersek > --- > block/qcow2-refcount.c | 34 ++++++++++++++++++---------------- > 1 file changed, 18 insertions(+), 16 deletions(-) > > diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c > index 3f2ed08..4a2df5f 100644 > --- a/block/qcow2-refcount.c > +++ b/block/qcow2-refcount.c > @@ -1399,14 +1399,14 @@ static int64_t realloc_refcount_block(BlockDriverState *bs, int reftable_index, > fprintf(stderr, "Could not allocate new cluster: %s\n", > strerror(-new_offset)); > ret = new_offset; > - goto fail; > + goto done; > } > > /* fetch current refcount block content */ > ret = qcow2_cache_get(bs, s->refcount_block_cache, offset, &refcount_block); > if (ret < 0) { > fprintf(stderr, "Could not fetch refcount block: %s\n", strerror(-ret)); > - goto fail; > + goto fail_free_cluster; > } > > /* new block has not yet been entered into refcount table, therefore it is > @@ -1417,8 +1417,7 @@ static int64_t realloc_refcount_block(BlockDriverState *bs, int reftable_index, > "check failed: %s\n", strerror(-ret)); > /* the image will be marked corrupt, so don't even attempt on freeing > * the cluster */ > - new_offset = 0; > - goto fail; > + goto done; > } > > /* write to new block */ > @@ -1426,7 +1425,7 @@ static int64_t realloc_refcount_block(BlockDriverState *bs, int reftable_index, > s->cluster_sectors); > if (ret < 0) { > fprintf(stderr, "Could not write refcount block: %s\n", strerror(-ret)); > - goto fail; > + goto fail_free_cluster; > } > > /* update refcount table */ > @@ -1436,24 +1435,27 @@ static int64_t realloc_refcount_block(BlockDriverState *bs, int reftable_index, > if (ret < 0) { > fprintf(stderr, "Could not update refcount table: %s\n", > strerror(-ret)); > - goto fail; > + goto fail_free_cluster; > } > > -fail: > - if (new_offset && (ret < 0)) { > - qcow2_free_clusters(bs, new_offset, s->cluster_size, > - QCOW2_DISCARD_ALWAYS); > - } > + goto done; I generally prefer an explicit ret = 0 in such places, so that we don't have to rely on the next person adding new code before the goto to understand the dependency here. But it's correct code, so I won't block it because of this. Kevin