
On 8/26/20 3:31 AM, Daniel P. Berrangé wrote:
On Thu, Aug 20, 2020 at 05:03:44PM -0600, Jim Fehlig wrote:
Add a configuration option for specifying location of the qemu modules directory, defaulting to /usr/lib64/qemu. Then use this location to check for changes in the directory, indicating that a qemu module has changed and capabilities need to be reprobed.
Signed-off-by: Jim Fehlig <jfehlig@suse.com> ---
This is a mildly tested impl of the idea discussed here (and elsewhere in the thread)
https://www.redhat.com/archives/libvir-list/2020-August/msg00684.html
I've checked that both system and session caches are updated when a qemu modules package is updated. I'll do some more testing but wanted to send the patch for early comments. Testing in other environments would be much appreciated. Thanks!
meson.build | 6 ++++++ meson_options.txt | 1 + src/qemu/qemu_capabilities.c | 35 +++++++++++++++++++++++++++++++++++ 3 files changed, 42 insertions(+)
diff --git a/meson.build b/meson.build index a72d0c0e85..16928482aa 100644 --- a/meson.build +++ b/meson.build @@ -1745,6 +1745,12 @@ if not get_option('driver_qemu').disabled() if use_qemu conf.set('WITH_QEMU', 1)
+ qemu_moddir = get_option('qemu_moddir') + if qemu_moddir == '' + qemu_moddir = '/usr/lib64/qemu'
We sholdn't hardcode lib64 - use libdir variable, eg
"/usr" / libdir / "qemu"
Good point. I've squashed in the below diff. While testing it I remembered to also test specifying the option, e.g. meson configure -Dqemu_moddir=/var/run/qemu/modules diff --git a/meson.build b/meson.build index 928f6a0bad..70a33cf5e4 100644 --- a/meson.build +++ b/meson.build @@ -1760,7 +1760,7 @@ if not get_option('driver_qemu').disabled() qemu_moddir = get_option('qemu_moddir') if qemu_moddir == '' - qemu_moddir = '/usr/lib64/qemu' + qemu_moddir = '/usr' / libdir / 'qemu' endif conf.set_quoted('QEMU_MODDIR', qemu_moddir)
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Thanks. I see we've entered freeze for 6.7.0. I'll wait until after the release before pushing this and the xen PCI permissive patches you reviewed. Regards, Jim