[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH 3/3] util: storage: Improve error message when requesting image above 'start'



On Wed, Apr 22, 2015 at 02:15:21PM +0200, Peter Krempa wrote:
On Wed, Apr 22, 2015 at 14:09:18 +0200, Martin Kletzander wrote:
On Tue, Apr 21, 2015 at 05:52:14PM +0200, Peter Krempa wrote:
>When a user would specify a backing chain index that is above the start
>point libvirt would report a rather unhelpful error:
>
>invalid argument: could not find backing store 1 in chain for 'sub/link2'
>
>This patch adds an explicit check that the index is below start point in
>the backing store and reports the following error if not:
>
>invalid argument: requested backing store index 1 is above 'sub/../qcow2' in chain for 'sub/link2'
>
>Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1177062
>---
> src/util/virstoragefile.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
>diff --git a/src/util/virstoragefile.c b/src/util/virstoragefile.c
>index c9d3977..2a2f238 100644
>--- a/src/util/virstoragefile.c
>+++ b/src/util/virstoragefile.c
>@@ -1341,6 +1341,15 @@ virStorageFileChainLookup(virStorageSourcePtr chain,
>         *parent = startFrom;
>     }
>
>+    if (idx && idx < i) {
>+        virReportError(VIR_ERR_INVALID_ARG,
>+                       _("requested backing store index %u is above '%s' "
>+                         "in chain for '%s'"),
>+                       idx, NULLSTR(startFrom->path), NULLSTR(start));

Possible crasher: startFrom can be NULL here, rather move the check
into the previous block where the parent is being set when found (idx
can't be < i when there is no startFrom anyway).  You can then avoid
cleaning up the *parent here -+
                              |
                  +-----------+
                  |
                  v

Actually right at the beginning of the function @parent gets filled by a
pointer to a local virStorageSource pointer (stack allocated) in case
the caller passed NULL as @parent. This allows to use the variable since
it is used in the lookup loop even if the user doesn't pass it.


Yeah, but I was talking about startFrom->path sigsegv-ing if
startFrom == NULL.

Don't worry, I had the same thought when I wrote the condition :)


Don't worry, I know people sometime misunderstand me ;)

Attachment: signature.asc
Description: PGP signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]