While running the Avocado test suite on my laptop/host (f26), I ran
into a few instances where if the timing is just right the libvirtd
restart will "hang" and ends up not being stoppable - leaving the
libvirtd <defunct> eventually once a kill -9 is run.
I tracked this into a pci_system_init() call that never returns.
Since I was mucking in the code - I also figured we could move
device enumeration into its own thread to "speed" startup.
John Ferlan (3):
nodedev: Move device enumumeration out of nodeStateInitialize
nodedev: Move the udevPCITranslate{Init|Deinit}
nodedev: Move the udevPCITranslateInit call
src/conf/virnodedeviceobj.h | 1 +
src/node_device/node_device_udev.c | 107 ++++++++++++++++++++++++-------------
2 files changed, 70 insertions(+), 38 deletions(-)
--
2.13.6
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list
