On Thu, Jul 20, 2017 at 10:08:15AM -0400, John Ferlan wrote: > Since virnodedeviceobj now has a self-lockable hash table, there's no > need to lock the table from the driver for processing. Thus remove the > locks from the driver for NodeDeviceObjList mgmt. > > This includes the test driver as well. > > Signed-off-by: John Ferlan <jferlan@xxxxxxxxxx> ... > > /* Populate with known devices */ ^This comment should come after the unlock, so it's 100% clear it's linked to the udevEnumerateDevices call. > - > + nodeDeviceUnlock(); > if (udevEnumerateDevices(udev) != 0) > goto cleanup; > > - ret = 0; > + return 0; > > cleanup: > nodeDeviceUnlock(); > - > - if (ret == -1) > - nodeStateCleanup(); > - return ret; > + nodeStateCleanup(); > + return -1; > } > The patch is straightforward, if there is a spot where you forgot to handle the locks in a similar manner and I missed it, either I will hit it with my follow-up "mdev nodedev-list kernel race" series or someone else eventually will :). ACK to 1,2 and this one, I had some comments to 3, but ACK'd it in principle. Erik -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list