On 03/10/2015 04:39 AM, Richard W.M. Jones wrote:
On Tue, Mar 10, 2015 at 02:32:04AM -0400, Laine Stump wrote:
> Commit 4bbe1029f fixed a problem in commit f7afeddc by moving the call
> to virNetDevGetIndex() to a location common to all interface types (so
> that the niceindex array would be filled in for macvtap as well as tap
> interfaces), but the location was *too* common, as the original call
> to virNetDevGetIndex() had been in a section qualified by "if
> (cfg->privileged)". The result was that the "fixed" libvirtd would
try
> to call virNetDevGetIndex() even for session mode libvirtd, and end up
> failing with the log message:
>
> Unable to open control socket: Operation not permitted
>
> To remedy that, this patch qualifies the call to virNetDevGetIndex()
> in its new location with cfg->privileged.
>
> This resolves
https://bugzilla.redhat.com/show_bug.cgi?id=1198244
> ---
>
> If someone (Rich?) needs this pushed before I am awake, please feel
> free to push it. (also push to the 1.2.13-maint branch if you do)
>
> src/qemu/qemu_command.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
> index 1510797..3d1483e 100644
> --- a/src/qemu/qemu_command.c
> +++ b/src/qemu/qemu_command.c
> @@ -7861,6 +7861,7 @@ qemuBuildInterfaceCommandLine(virCommandPtr cmd,
> char **tapfdName = NULL;
> char **vhostfdName = NULL;
> int actualType = virDomainNetGetActualType(net);
> + virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
> virNetDevBandwidthPtr actualBandwidth;
> size_t i;
>
> @@ -7936,7 +7937,7 @@ qemuBuildInterfaceCommandLine(virCommandPtr cmd,
> /* network and bridge use a tap device, and direct uses a
> * macvtap device
> */
> - if (nicindexes && nnicindexes && net->ifname) {
> + if (cfg->privileged && nicindexes && nnicindexes
&& net->ifname) {
> if (virNetDevGetIndex(net->ifname, &nicindex) < 0 ||
> VIR_APPEND_ELEMENT(*nicindexes, *nnicindexes, nicindex) < 0)
> goto cleanup;
Looks sensible, ACK.
As Laine says, please make sure this gets into 1.2.13-maint because it
currently affects all 1.2.13 / Rawhide users.
I just pushed it to master and to v1.2.13-maint.
Also, why isn't there a regression test that would have picked this
up? A trivial reproducer is:
$ guestfish -a /dev/null --network run
but any test case that launches a guest with a network interface as
non-root would have caught this.
This fails only if a tap device is used, which requires a properly
configured qemu-bridge-helper and a bridge device with a particular name
(and that no other session mode libvirtd be currently running for the
user running the test), and we can't require that for the unit tests in
" make check", so it would need to be done in the tck tests.