In one of my latest patch (v6.6.0~30) I was trying to remove
libdevmapper use in favor of our own implementation. However, the
code did not take into account that device mapper can be not
compiled into the kernel (e.g. be a separate module that's not
loaded) in which case /proc/devices won't have the device-mapper
major number and thus virDevMapperGetTargets() and/or
virIsDevMapperDevice() fails.

Fixes: 22494556542c676d1b9e7f1c1f2ea13ac17e1e3e
Reported-by: Andrea Bolognani <abologna@redhat.com>
Reported-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
---
 src/util/virdevmapper.c | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/src/util/virdevmapper.c b/src/util/virdevmapper.c
index b24d9377f8..3cc399f382 100644
--- a/src/util/virdevmapper.c
+++ b/src/util/virdevmapper.c
@@ -54,6 +54,9 @@ virDevMapperGetMajor(unsigned int *major)
     VIR_AUTOSTRINGLIST lines = NULL;
     size_t i;
 
+    if (!virFileExists(CONTROL_PATH))
+        return -2;
+
     if (virFileReadAll(PROC_DEVICES, BUF_SIZE, &buf) < 0)
         return -1;
 
@@ -126,8 +129,13 @@ virDMOpen(void)
 
     memset(&dm, 0, sizeof(dm));
 
-    if ((controlFD = open(CONTROL_PATH, O_RDWR)) < 0)
+    if ((controlFD = open(CONTROL_PATH, O_RDWR)) < 0) {
+        if (errno == ENOENT)
+            return -2;
+
+        virReportSystemError(errno, _("Unable to open %s"), CONTROL_PATH);
         return -1;
+    }
 
     if (!virDMIoctl(controlFD, DM_VERSION, &dm, &tmp)) {
         virReportSystemError(errno, "%s",
@@ -299,8 +307,16 @@ virDevMapperGetTargets(const char *path,
      * consist of devices or yet another targets. If that's the
      * case, we have to stop recursion somewhere. */
 
-    if ((controlFD = virDMOpen()) < 0)
+    if ((controlFD = virDMOpen()) < 0) {
+        if (controlFD == -2) {
+            /* The CONTROL_PATH doesn't exist. Probably the
+             * module isn't loaded, yet. Don't error out, just
+             * exit. */
+            return 0;
+        }
+
         return -1;
+    }
 
     return virDevMapperGetTargetsImpl(controlFD, path, devPaths, ttl);
 }
-- 
2.26.2

On Tue, Aug 18, 2020 at 11:30:25 +0200, Michal Privoznik wrote:
> In one of my latest patch (v6.6.0~30) I was trying to remove
> libdevmapper use in favor of our own implementation. However, the
> code did not take into account that device mapper can be not
> compiled into the kernel (e.g. be a separate module that's not
> loaded) in which case /proc/devices won't have the device-mapper
> major number and thus virDevMapperGetTargets() and/or
> virIsDevMapperDevice() fails.

I'd probably mention the fact that all callers just configure some
aspects of the DM devices thus it's guaranteed that there's nothing to
do in this case and it's safe to ignore the error. Just I can't seem to
come up with a nice formulation.

> Fixes: 22494556542c676d1b9e7f1c1f2ea13ac17e1e3e
> Reported-by: Andrea Bolognani <abologna@redhat.com>
> Reported-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
> Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
> ---

Regardless of the above:

Reviewed-by: Peter Krempa <pkrempa@redhat.com>

>  src/util/virdevmapper.c | 20 ++++++++++++++++++--
>  1 file changed, 18 insertions(+), 2 deletions(-)