[vbox-dev] Changing vboxsf linux mount data argument from vbsf_mount_info_new to a normal string
Hans de Goede
hdegoede at redhat.com
Mon Jan 15 19:33:23 UTC 2018
Now a days the mount syscall usually takes a c-string
with "," separated key=value pair of arguments in plain
The upstream kernel folks have requested that I change vboxsf
to also work this way. My plan for this is for me to not only
modify vboxsf for this, but for me to also provide patches for
mount.vboxsf and VBoxServiceAutoMount.cpp to support this.
My plan is to make the upstream version of the vboxsf kernel-code
still check for the data argument starting with vbsf_mount_info_new
signature and if it sees that return -EINVAL.
Then mount.vboxsf and VBoxServiceAutoMount.cpp can first still
try using vbsf_mount_info_new and if that fails with -EINVAL
then use the new string with "," separated key=value pairs
approach. This way they will keep working unmodified with
the vbox svn vboxsf implementation (which you may not want to
change) and will transparantly do the right thing when running
with an upstream kernel version of vboxsf.
Note that I believe that this will remove the need for a
mount.vboxsf userspace util altogether, once we use the
string method I expect a regular mount command to work for this.
But the proof is in the pudding, so we will see if this is the
case when I actually have code for this.
More information about the vbox-dev