On Tue, Jan 29, 2019 at 08:43:31AM -0500, John Ferlan wrote: > > > On 1/29/19 6:44 AM, Daniel P. Berrangé wrote: > > On Thu, Jan 17, 2019 at 04:22:07PM -0500, John Ferlan wrote: > >> https://bugzilla.redhat.com/show_bug.cgi?id=1584663 > >> > >> Modify the command generation to add some default options to > >> an NFS Storage Pool based on the OS type. For Linux, it'll be > >> the "nodev, nosuid, noexec". For FreeBSD, it'll be "nosuid, noexec". > >> For others, just leave the options alone. > >> > >> Modify the storagepoolxml2argvtest to handle the fact that the > >> same input XML could generate different output XML based on whether > >> Linux, FreeBSD, or other was being built. > >> > >> Signed-off-by: John Ferlan <jferlan@xxxxxxxxxx> > >> --- > >> src/storage/storage_util.c | 16 ++++++++ > >> .../pool-netfs-auto-freebsd.argv | 1 + > >> .../pool-netfs-auto-linux.argv | 1 + > >> .../pool-netfs-freebsd.argv | 1 + > >> .../pool-netfs-linux.argv | 1 + > >> tests/storagepoolxml2argvtest.c | 40 +++++++++++++++---- > >> 6 files changed, 53 insertions(+), 7 deletions(-) > >> create mode 100644 tests/storagepoolxml2argvdata/pool-netfs-auto-freebsd.argv > >> create mode 100644 tests/storagepoolxml2argvdata/pool-netfs-auto-linux.argv > >> create mode 100644 tests/storagepoolxml2argvdata/pool-netfs-freebsd.argv > >> create mode 100644 tests/storagepoolxml2argvdata/pool-netfs-linux.argv > >> > >> diff --git a/src/storage/storage_util.c b/src/storage/storage_util.c > >> index a84ee5b600..44a95d9fab 100644 > >> --- a/src/storage/storage_util.c > >> +++ b/src/storage/storage_util.c > >> @@ -34,6 +34,11 @@ > >> # ifndef FS_NOCOW_FL > >> # define FS_NOCOW_FL 0x00800000 /* Do not cow file */ > >> # endif > >> +# define default_nfs_mount_opts "nodev,nosuid,noexec" > >> +#elif defined(__FreeBSD__) > >> +# define default_nfs_mount_opts "nosuid,noexec" > >> +#else > >> +# define default_nfs_mount_opts "" > >> #endif > >> > >> #if WITH_BLKID > >> @@ -4261,12 +4266,21 @@ virStorageBackendFileSystemGetPoolSource(virStoragePoolObjPtr pool) > >> } > >> > >> > >> +static void > >> +virStorageBackendFileSystemMountNFSAddOptions(virCommandPtr cmd) > >> +{ > >> + if (*default_nfs_mount_opts != '\0') > >> + virCommandAddArgList(cmd, "-o", default_nfs_mount_opts, NULL); > >> +} > >> + > >> + > >> static void > >> virStorageBackendFileSystemMountNFSArgs(virCommandPtr cmd, > >> const char *src, > >> virStoragePoolDefPtr def) > >> { > >> virCommandAddArgList(cmd, src, def->target.path, NULL); > >> + virStorageBackendFileSystemMountNFSAddOptions(cmd); > >> } > >> > >> > >> @@ -4308,6 +4322,8 @@ virStorageBackendFileSystemMountDefaultArgs(virCommandPtr cmd, > >> else > >> fmt = virStoragePoolFormatFileSystemNetTypeToString(def->source.format); > >> virCommandAddArgList(cmd, "-t", fmt, src, def->target.path, NULL); > >> + if (def->type == VIR_STORAGE_POOL_NETFS) > >> + virStorageBackendFileSystemMountNFSAddOptions(cmd); > > > > This doesn't need to be restricted to just NFS. A kind of filesystem we mount > > as a storage pool should use these extra options, as they are general purpose > > mount options, not FS specific. > > > > Just s/nfs//i in all the method names / constants. > > > > I can modify default_nfs_mount_opts to be just default_mount_opts; > however, as you see going forward the new method ends up being used by > patch4 for "nfsvers=%u," and patch9 for Namespace options, so I thinking > keeping the name virStorageBackendFileSystemMountNFSAddOptions is better. I mentioned later for patch 9 that the namespace options should apply for any filesystem mounts we do IOW, both pool type=fs and type=netfs. In fact even for type=netfs, we can't assume it is NFS - it could be a CIFS or GlusterFS pool, rather than NFS so the name is a bit misleading in that respect anyway. For patch 4, we just need to push the check for whether it is NFS down into the virStorageBackendFileSystemMountAddOptions method, so it only formats the nfsvers for that specific pool config. Presumably the XML parser stops us even parsing the protocol out of the XML unless type == netfs + format==nfs, so we'd be ok even without the check, but its less suprising to have the explicit check anyway. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :| -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
