[dm-devel] [PATCH v3 10/11] libmultipath: don't try to set hwhandler if it is retained
Hannes Reinecke
hare at suse.de
Thu Jun 22 06:21:58 UTC 2017
On 06/21/2017 05:06 PM, Martin Wilck wrote:
> Setting a device handler only works if retain_attached_hw_handler
> is 'no', or if the kernel didn't auto-assign a handler. If this
> is not the case, don't even attempt to set a different handler.
>
> This requires reading the sysfs "dh_state" path attribute.
> For internal consistency, this attribute must be updated after domap().
>
> Signed-off-by: Martin Wilck <mwilck at suse.com>
> ---
> libmultipath/configure.c | 52 +++++++++++++++++++++++++++++++++++++++++++++++-
> libmultipath/discovery.c | 4 ++++
> libmultipath/propsel.c | 15 ++++++++++++++
> libmultipath/structs.h | 2 ++
> 4 files changed, 72 insertions(+), 1 deletion(-)
>
> diff --git a/libmultipath/configure.c b/libmultipath/configure.c
> index 74b6f52a..55dbb261 100644
> --- a/libmultipath/configure.c
> +++ b/libmultipath/configure.c
> @@ -275,15 +275,21 @@ int setup_map(struct multipath *mpp, char *params, int params_size)
>
> /*
> * properties selectors
> + *
> + * Ordering matters for some properties:
> + * - features after no_path_retry and retain_hwhandler
> + * - hwhandler after retain_hwhandler
> + * No guarantee that this list is complete, check code in
> + * propsel.c if in doubt.
> */
> conf = get_multipath_config();
> select_pgfailback(conf, mpp);
> select_pgpolicy(conf, mpp);
> select_selector(conf, mpp);
> - select_hwhandler(conf, mpp);
> select_no_path_retry(conf, mpp);
> select_retain_hwhandler(conf, mpp);
> select_features(conf, mpp);
> + select_hwhandler(conf, mpp);
> select_rr_weight(conf, mpp);
> select_minio(conf, mpp);
> select_mode(conf, mpp);
> @@ -706,6 +712,48 @@ fail:
> return 1;
> }
>
> +static void
> +check_dh_state_changed(struct multipath *mp)
> +{
> + struct config *conf;
> + struct path newp, *pp;
> + struct pathgroup *pg;
> + int i, j;
> +
> + conf = get_multipath_config();
> +
> + vector_foreach_slot (mp->pg, pg, j) {
> + vector_foreach_slot (pg->paths, pp, i) {
> + if (!pp->udev || !strlen(pp->dh_state) ||
> + (conf->retain_hwhandler == RETAIN_HWHANDLER_ON &&
> + strcmp(pp->dh_state, "detached")))
> + continue;
> +
> + memset(&newp, 0, sizeof(newp));
> + memcpy(newp.dev, pp->dev, sizeof(newp.dev));
> + newp.udev = udev_device_ref(pp->udev);
> +
> + if (pathinfo(&newp, conf, DI_SYSFS) == PATHINFO_OK) {
> + if (strncmp(newp.dh_state, pp->dh_state,
> + SCSI_DH_SIZE)) {
> + condlog(3, "%s: dh_state changed from %s to %s",
> + pp->dev,
> + pp->dh_state,
> + newp.dh_state);
> + memcpy(pp->dh_state, newp.dh_state,
> + SCSI_DH_SIZE);
> + }
> + } else
> + condlog(1, "%s: failed to update dh_state",
> + pp->dev);
> +
> + udev_device_unref(newp.udev);
> + }
> + }
> +
> + put_multipath_config(conf);
> +}
> +
> /*
> * Return value:
> */
I would avoid using a temporary udev device here; either save the
dh_state value and call pathinfo() on the _actual_ device or read
dh_state directly from sysfs without an intermediate udev device.
> @@ -828,6 +876,8 @@ int domap(struct multipath *mpp, char *params, int is_daemon)
> }
> }
> dm_setgeometry(mpp);
> +
> + check_dh_state_changed(mpp);
> return DOMAP_OK;
> }
> return DOMAP_FAIL;
> diff --git a/libmultipath/discovery.c b/libmultipath/discovery.c
> index 663c8eaa..8c0c6a9f 100644
> --- a/libmultipath/discovery.c
> +++ b/libmultipath/discovery.c
> @@ -1188,6 +1188,10 @@ scsi_sysfs_pathinfo (struct path * pp, vector hwtable)
> condlog(3, "%s: tgt_node_name = %s",
> pp->dev, pp->tgt_node_name);
>
> + if (sysfs_attr_get_value(parent, "dh_state",
> + pp->dh_state, sizeof(pp->dh_state)) >= 0)
> + condlog(3, "%s: dh_state = %s", pp->dev, pp->dh_state);
> +
> return 0;
> }
>
> diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c
> index d609394e..d1b3d416 100644
> --- a/libmultipath/propsel.c
> +++ b/libmultipath/propsel.c
> @@ -345,7 +345,22 @@ out:
> int select_hwhandler(struct config *conf, struct multipath *mp)
> {
> char *origin;
> + struct path *pp;
> + char handler[SCSI_DH_SIZE+2];
> + int i;
>
> + if (mp->retain_hwhandler != RETAIN_HWHANDLER_OFF) {
> + vector_foreach_slot(mp->paths, pp, i) {
> + if (strlen(pp->dh_state) &&
> + strcmp(pp->dh_state, "detached")) {
> + snprintf(handler, sizeof(handler),
> + "1 %s", pp->dh_state);
> + mp->hwhandler = handler;
> + origin = "(setting: retained by kernel driver)";
> + goto out;
> + }
> + }
> + }
> mp_set_hwe(hwhandler);
> mp_set_conf(hwhandler);
> mp_set_default(hwhandler, DEFAULT_HWHANDLER);
> diff --git a/libmultipath/structs.h b/libmultipath/structs.h
> index 8ea984d9..4203e2b0 100644
> --- a/libmultipath/structs.h
> +++ b/libmultipath/structs.h
> @@ -22,6 +22,7 @@
> #define SCSI_PRODUCT_SIZE 17
> #define SCSI_REV_SIZE 5
> #define SCSI_STATE_SIZE 19
> +#define SCSI_DH_SIZE 9 /* must hold "detached" */
>
> #define NO_PATH_RETRY_UNDEF 0
> #define NO_PATH_RETRY_FAIL -1
> @@ -205,6 +206,7 @@ struct path {
> char rev[SCSI_REV_SIZE];
> char serial[SERIAL_SIZE];
> char tgt_node_name[NODE_NAME_SIZE];
> + char dh_state[SCSI_DH_SIZE];
> unsigned long long size;
> unsigned int checkint;
> unsigned int tick;
>
Hmm.
Not sure if I fully agree with this.
I do see the need to read 'dh_state' from pathinfo(), just to figure out
if an hardware handler is already loaded.
But once select_hwhandler is done it's quite pointless to update the
dh_state; what exactly _would_ be the error action in this case?
Plus the code detects the failure, but then doesn't do anything with it...
So, please, if you insist on checking dh_state please implement correct
error action here, like updating the 'hwhandler' value to that found in
dh_state or disabling the hardware handler if it's found to be detached.
Cheers,
Hannes
--
Dr. Hannes Reinecke Teamlead Storage & Networking
hare at suse.de +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
More information about the dm-devel
mailing list