Date:   Fri, 20 Oct 2017 11:13:39 -0400
From:   Steve Lin <steven.lin1@...adcom.com>
To:     Jiri Pirko <jiri@...nulli.us>
Cc:     Linux Netdev List <netdev@...r.kernel.org>,
        Jiri Pirko <jiri@...lanox.com>,
        "David S . Miller" <davem@...emloft.net>,
        Michael Chan <michael.chan@...adcom.com>,
        John Linville <linville@...driver.com>,
        Andy Gospodarek <gospo@...adcom.com>
Subject: Re: [PATCH net-next v2 1/6] devlink: Add permanent config parameter
 get/set operations

On Fri, Oct 20, 2017 at 10:39 AM, Jiri Pirko <jiri@...nulli.us> wrote:
> Thu, Oct 19, 2017 at 09:17:05PM CEST, steven.lin1@...adcom.com wrote:
>>Add support for permanent config parameter get/set commands. Used
>>for parameters held in NVRAM, persistent device configuration.
>>
>>Signed-off-by: Steve Lin <steven.lin1@...adcom.com>
>>Acked-by: Andy Gospodarek <gospo@...adcom.com>
>>---
>> include/net/devlink.h        |   3 +
>> include/uapi/linux/devlink.h |  11 ++
>> net/core/devlink.c           | 234 +++++++++++++++++++++++++++++++++++++++++++
>> 3 files changed, 248 insertions(+)
>>
>>diff --git a/include/net/devlink.h b/include/net/devlink.h
>>index b9654e1..bd64623 100644
>>--- a/include/net/devlink.h
>>+++ b/include/net/devlink.h
>>@@ -270,6 +270,9 @@ struct devlink_ops {
>>       int (*eswitch_inline_mode_set)(struct devlink *devlink, u8 inline_mode);
>>       int (*eswitch_encap_mode_get)(struct devlink *devlink, u8 *p_encap_mode);
>>       int (*eswitch_encap_mode_set)(struct devlink *devlink, u8 encap_mode);
>>+      int (*perm_config_get)(struct devlink *devlink, u32 param, u32 *value);
>>+      int (*perm_config_set)(struct devlink *devlink, u32 param, u32 value,
>
> Please use enum instead of "u32 param". Also, what would happen if the
> value is >u32, like string for example? I believe we need to take it into
> the consideration for the UAPI sake.
>
>

Using enum instead of u32 param: ok, will do in v3, thanks.

Value > u32:  In the RFC and v1 versions of the patch, each parameter
was its own attribute, so could have its own type (u32, string,
whatever).  In v2, trying to move to nested parameters w/ parameter
being an enum, as requested, it seems to mean that the parameter value
now must be defined as a specific type, so I went with u32.

If we need to support strings or other types > u32, then the
perm_config_value attribute will not be a fixed type, so can't be
policy checked.  Or, I could go back to non-nested as in RFC/v1 case
and have each parameter with its own type.

> [...]
>
>
>>+
>>+static int devlink_nl_single_param_set(struct sk_buff *msg,
>>+                                     struct devlink *devlink,
>>+                                     u32 param, u32 value)
>>+{
>>+      u32 orig_value;
>>+      u8 need_restart;
>>+      int err;
>>+      const struct devlink_ops *ops = devlink->ops;
>>+      struct nlattr *cfgparam_attr;
>
> Reverse christmas tree please (this applies to all functions)

Will do in v3, thanks.

>
>
>>+
>>+      /* First get current value of parameter */
>>+      err = ops->perm_config_get(devlink, param, &orig_value);
>
> I'm missing why this is needed.
>
>
>>+      if (err)
>>+              return err;
>>+
>>+      /* Now set parameter */
>>+      err = ops->perm_config_set(devlink, param, value, &need_restart);
>>+      if (err)
>>+              return err;
>>+
>>+      cfgparam_attr = nla_nest_start(msg, DEVLINK_ATTR_PERM_CONFIG);
>>+      /* Update restart reqd - if any param needs restart, should be set */
>>+      if (need_restart)
>>+              err = nla_put_u8(msg,
>>+                               DEVLINK_ATTR_PERM_CONFIG_RESTART_REQUIRED, 1);
>>+
>>+      /* Since set was successful, write attr back to msg with orig val */
>>+      err = nla_put_u32(msg, DEVLINK_ATTR_PERM_CONFIG_PARAMETER, param);
>>+      err = nla_put_u32(msg, DEVLINK_ATTR_PERM_CONFIG_VALUE, orig_value);
>
> Why to write it back?

In a response to the "RFC" version of this patch, you wrote:  "Also,
we need to expose to the user the original value (currently being
used) and the new one (to be used after driver re-instatiation)".

I understood that to mean that we need to return the current/original
value of the parameter (and the user knows the new value, since they
are setting it).

If I mis-interpreted that comment, then I'm happy to remove returning
the original value to the user; it wasn't in there originally.

>
>
>>+
>>+      nla_nest_end(msg, cfgparam_attr);
>>+
>>+      return 0;
>>+}
>>+
>>+static int devlink_nl_cmd_perm_config_set_doit(struct sk_buff *skb,
>>+                                             struct genl_info *info)
>>+{
>>+      struct devlink *devlink = info->user_ptr[0];
>>+      struct sk_buff *msg;
>>+      void *hdr;
>>+      struct nlattr *attr;
>>+      int rem;
>>+      int err;
>>+      u8 restart_reqd = 0;
>>+      struct nlattr *cfgparam_attr;
>>+      struct nlattr *tb[DEVLINK_ATTR_MAX + 1];
>>+      u32 param;
>>+      u32 value;
>>+
>>+      if (!devlink->ops || !devlink->ops->perm_config_get ||
>>+          !devlink->ops->perm_config_set)
>>+              return -EOPNOTSUPP;
>>+
>>+      msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
>>+      if (!msg)
>>+              return -ENOMEM;
>>+
>>+      hdr = genlmsg_put(msg, info->snd_portid, info->snd_seq,
>>+                        &devlink_nl_family, 0, DEVLINK_CMD_PERM_CONFIG_SET);
>>+      if (!hdr) {
>>+              err = -EMSGSIZE;
>>+              goto nla_msg_failure;
>>+      }
>>+
>>+      err = devlink_nl_put_handle(msg, devlink);
>>+      if (err)
>>+              goto nla_put_failure;
>>+
>>+      cfgparam_attr = nla_nest_start(msg, DEVLINK_ATTR_PERM_CONFIGS);
>>+
>>+      nla_for_each_nested(attr, info->attrs[DEVLINK_ATTR_PERM_CONFIGS], rem) {
>>+              err = nla_parse_nested(tb, DEVLINK_ATTR_MAX, attr,
>>+                                     devlink_nl_policy, NULL);
>>+              if (err)
>>+                      goto nla_nest_failure;
>>+
>>+              if (!tb[DEVLINK_ATTR_PERM_CONFIG_PARAMETER] ||
>>+                  !tb[DEVLINK_ATTR_PERM_CONFIG_VALUE])
>>+                      continue;
>>+
>>+              param = nla_get_u32(tb[DEVLINK_ATTR_PERM_CONFIG_PARAMETER]);
>
> You should check it the "param" value is withing the enum boundary.

Will do in v3, I'll add a DEVLINK_PERM_CONFIG_MAX enum and check against that.

>
>
>>+              value = nla_get_u32(tb[DEVLINK_ATTR_PERM_CONFIG_VALUE]);
>>+              err = devlink_nl_single_param_set(msg, devlink, param,
>>+                                                value);
>>+              if (err)
>>+                      goto nla_nest_failure;
>
> Wouldn't it make sense to rollback to old values if any of the config
> parameters set would fail?

Hmmmm....  We could do a rollback on failure of any of the set cmds,
but I feel like that's more useful when the user doesn't know which
sets failed. (i.e. if the response was just a boolean ok/not-ok
response).

If the user knows which sets worked and which didn't, then maybe no
rollback is necessary.  So perhaps I change the code so that rather
than falling out on error, it continues to try all the set cmds, but
just returns the parameters which were successful.

Is there a convention for this?  I'm fine either way.

>
>
>>+      }
>>+
>>+      nla_nest_end(msg, cfgparam_attr);
>>+
>>+      if (restart_reqd) {
>>+              err = nla_put_u8(msg, DEVLINK_ATTR_PERM_CONFIG_RESTART_REQUIRED,
>>+                               restart_reqd);
>>+              if (err)
>>+                      goto nla_put_failure;
>>+      }
>>+
>>+      genlmsg_end(msg, hdr);
>>+      return genlmsg_reply(msg, info);
>>+
>>+nla_nest_failure:
>>+      nla_nest_cancel(msg, cfgparam_attr);
>>+nla_put_failure:
>>+      genlmsg_cancel(msg, hdr);
>>+nla_msg_failure:
>>+      return err;
>>+}
>>+
>> int devlink_dpipe_match_put(struct sk_buff *skb,
>>                           struct devlink_dpipe_match *match)
>> {
>>@@ -2291,6 +2509,8 @@ static const struct nla_policy devlink_nl_policy[DEVLINK_ATTR_MAX + 1] = {
>>       [DEVLINK_ATTR_ESWITCH_ENCAP_MODE] = { .type = NLA_U8 },
>>       [DEVLINK_ATTR_DPIPE_TABLE_NAME] = { .type = NLA_NUL_STRING },
>>       [DEVLINK_ATTR_DPIPE_TABLE_COUNTERS_ENABLED] = { .type = NLA_U8 },
>>+      [DEVLINK_ATTR_PERM_CONFIG_PARAMETER] = { .type = NLA_U32 },
>>+      [DEVLINK_ATTR_PERM_CONFIG_VALUE] = { .type = NLA_U32 },
>> };
>>
>> static const struct genl_ops devlink_nl_ops[] = {
>>@@ -2451,6 +2671,20 @@ static const struct genl_ops devlink_nl_ops[] = {
>>               .flags = GENL_ADMIN_PERM,
>>               .internal_flags = DEVLINK_NL_FLAG_NEED_DEVLINK,
>>       },
>>+      {
>>+              .cmd = DEVLINK_CMD_PERM_CONFIG_GET,
>>+              .doit = devlink_nl_cmd_perm_config_get_doit,
>>+              .policy = devlink_nl_policy,
>>+              .flags = GENL_ADMIN_PERM,
>>+              .internal_flags = DEVLINK_NL_FLAG_NEED_DEVLINK,
>>+      },
>>+      {
>>+              .cmd = DEVLINK_CMD_PERM_CONFIG_SET,
>>+              .doit = devlink_nl_cmd_perm_config_set_doit,
>>+              .policy = devlink_nl_policy,
>>+              .flags = GENL_ADMIN_PERM,
>>+              .internal_flags = DEVLINK_NL_FLAG_NEED_DEVLINK,
>>+      },
>> };
>>
>> static struct genl_family devlink_nl_family __ro_after_init = {
>>--
>>2.7.4
>>

Powered by blists - more mailing lists