On Mon, Jun 06, 2016 at 11:02:09 +0200, Ján Tomko wrote:
> Filter out some subsystems we are not interested in.
> ---
> After the netdev cleanups, this speeds up the driver initialization from 18 ms to 13 ms,
> which is percentually a lot, but neligible in absolute times.
>
> This patch uses a negative filter because I could not find an exhaustive list of possible subsystems.
> A positive filter could be applied to udev_monitor as well.
>
> src/node_device/node_device_udev.c | 23 ++++++++++++++++++++++-
> 1 file changed, 22 insertions(+), 1 deletion(-)
>
> diff --git a/src/node_device/node_device_udev.c b/src/node_device/node_device_udev.c
> index b46fec6..2e86230 100644
> --- a/src/node_device/node_device_udev.c
> +++ b/src/node_device/node_device_udev.c
> @@ -1167,13 +1167,34 @@ static int udevProcessDeviceListEntry(struct udev *udev,
> }
>
>
I think this deserves a comment explaining the reason and what to add
here.
> +const char *subsystem_blacklist[] = {
> + "acpi", "tty", "vc", "i2c",
> +};
> +
> +static int udevEnumerateAddMatches(struct udev_enumerate *udev_enumerate)
> +{
> + size_t i;
> +
> + for (i = 0; i < ARRAY_CARDINALITY(subsystem_blacklist); i++) {
> + const char *s = subsystem_blacklist[i];
> + if (udev_enumerate_add_nomatch_subsystem(udev_enumerate, s) < 0) {
> + virReportSystemError(errno, "%s", _("failed to add susbsystem filter"));
> + return -1;
Should this be fatal? On the other hand other stuff will probably fail
too.
> + }
> + }
> + return 0;
> +}
> +
> +
> static int udevEnumerateDevices(struct udev *udev)
> {
> struct udev_enumerate *udev_enumerate = NULL;
> struct udev_list_entry *list_entry = NULL;
> - int ret = 0;
> + int ret = -1;
>
> udev_enumerate = udev_enumerate_new(udev);
> + if (udevEnumerateAddMatches(udev_enumerate) < 0)
> + goto cleanup;
>
> ret = udev_enumerate_scan_devices(udev_enumerate);
> if (ret != 0) {
ACK with the comment added.
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list