On Thu, Sep 08, 2011 at 02:59:56PM +0800, Hu Tao wrote:
> This patch adds a parameter --weight-device to virsh command
> blkiotune for setting/getting blkio.weight_device.
> ---
> daemon/remote.c | 5 +
> include/libvirt/libvirt.h.in | 9 ++
> src/conf/domain_conf.c | 142 ++++++++++++++++++++++++++++++++++-
> src/conf/domain_conf.h | 15 ++++
> src/libvirt_private.syms | 1 +
> src/qemu/qemu_cgroup.c | 22 ++++++
> src/qemu/qemu_driver.c | 170 +++++++++++++++++++++++++++++++++++++++++-
> src/util/cgroup.c | 33 ++++++++
> src/util/cgroup.h | 3 +
> tools/virsh.c | 31 ++++++++
> tools/virsh.pod | 5 +-
> 11 files changed, 430 insertions(+), 6 deletions(-)
>
> diff --git a/daemon/remote.c b/daemon/remote.c
> index a9d0daa..ec91526 100644
> --- a/daemon/remote.c
> +++ b/daemon/remote.c
> @@ -1503,6 +1503,7 @@ remoteDispatchDomainGetBlkioParameters(virNetServerPtr server ATTRIBUTE_UNUSED,
> int nparams = args->nparams;
> unsigned int flags;
> int rv = -1;
> + int i;
> struct daemonClientPrivate *priv =
> virNetServerClientGetPrivateData(client);
>
> @@ -1547,6 +1548,10 @@ success:
> cleanup:
> if (rv < 0)
> virNetMessageSaveError(rerr);
> + for (i = 0; i < nparams; i++) {
> + if (params[i].type == VIR_TYPED_PARAM_STRING)
> + VIR_FREE(params[i].value.s);
> + }
> VIR_FREE(params);
> if (dom)
> virDomainFree(dom);
> diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in
> index e57241c..c65d8f7 100644
> --- a/include/libvirt/libvirt.h.in
> +++ b/include/libvirt/libvirt.h.in
> @@ -1137,6 +1137,15 @@ char * virDomainGetSchedulerType(virDomainPtr domain,
>
> #define VIR_DOMAIN_BLKIO_WEIGHT "weight"
>
> +/**
> + * VIR_DOMAIN_BLKIO_WEIGHT_DEVICE:
> + *
> + * Macro for the blkio tunable weight_device: it represents the
> + * per device weight.
> + */
> +
> +#define VIR_DOMAIN_BLKIO_WEIGHT_DEVICE "weight_device"
> +
> /* Set Blkio tunables for the domain*/
> int virDomainSetBlkioParameters(virDomainPtr domain,
> virTypedParameterPtr params,
> diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
> index 74f8d6a..d10e30c 100644
> --- a/src/conf/domain_conf.c
> +++ b/src/conf/domain_conf.c
> @@ -565,6 +565,108 @@ VIR_ENUM_IMPL(virDomainNumatuneMemMode, VIR_DOMAIN_NUMATUNE_MEM_LAST,
> #define VIR_DOMAIN_XML_WRITE_FLAGS VIR_DOMAIN_XML_SECURE
> #define VIR_DOMAIN_XML_READ_FLAGS VIR_DOMAIN_XML_INACTIVE
>
> +/**
> + * virBlkioWeightDeviceToStr:
> + *
> + * This function returns a string representing device weights that is
> + * suitable for writing to /cgroup/blkio/blkio.weight_device, given
> + * a list of weight devices.
> + */
> +int virBlkioWeightDeviceToStr(virBlkioWeightDevicePtr weightdevices,
> + int ndevices,
> + char **result)
> +{
> + int len = 0;
> + int ret = -1;
> + int i, j;
> + char **weight_devices;
> + char *str;
> +
> + if (VIR_ALLOC_N(weight_devices, ndevices) < 0) {
> + goto fail_nomem1;
> + }
> + for (i = 0; i < ndevices; i++) {
> + int tmp;
> + tmp = virAsprintf(&weight_devices[i], "%d:%d %d",
> + weightdevices[i].major,
> + weightdevices[i].minor,
> + weightdevices[i].weight);
> + if (tmp < 0) {
> + goto fail_nomem2;
> + }
> + len += tmp + 1; /* 1 for '\n' and the trailing '\0' */
> + }
> +
> + if (VIR_ALLOC_N(str, len) < 0) {
> + goto fail_nomem2;
> + }
> + for (i = 0; i < ndevices; i++) {
> + strcat(str, weight_devices[i]);
> + strcat(str, "\n");
> + }
> + str[len-1] = '\0';
> +
> + *result = str;
> +
> + ret = 0;
> +
> +fail_nomem2:
> + for (j = 0; j < i; j++)
> + VIR_FREE(weight_devices[i]);
> + VIR_FREE(weight_devices);
> +fail_nomem1:
> + if (ret != 0)
> + virReportOOMError();
> + return ret;
> +}
I think this method would become alot simpler if you switch
over to use virBufferPtr for all the string concatenation
and formatting.
> +
> +/**
> + * virDomainBlkioWeightDeviceParseXML
> + *
> + * this function parses a XML node:
> + *
> + * <device>
> + * <major>major</major>
> + * <minor>minor</minor>
> + * <weight>weight</weight>
> + * </device>
> + *
> + * and fills a virBlkioWeightDevice struct.
> + */
I'm not really seeing the benefit in using major, minor in the XML for
this. The <disk> element is using the /dev/hda1 path for the host
device, so I'd expect the same path to be usable for the block I/O
tuning.
How does the scope work here, does major,minor have to refer to a block
device, or can it refer to a partition ? If we have multiple <device>
elements, each giving a different partition on the same device can we
set different weight for each partition ?
Daniel
--
|: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org -o- http://virt-manager.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list