Re: [RFC PATCH v4] ndctl: monitor: add ndctl monitor daemon

From: "Qi, Fuli" <qi.fuli@jp.fujitsu.com>
To: Dan Williams <dan.j.williams@intel.com>
Cc: linux-nvdimm <linux-nvdimm@lists.01.org>
Subject: Re: [RFC PATCH v4] ndctl: monitor: add ndctl monitor daemon
Date: Tue, 20 Mar 2018 10:03:44 +0900	[thread overview]
Message-ID: <8862cf22-1e78-2bdf-0abe-4403164d3240@jp.fujitsu.com> (raw)
In-Reply-To: <CAPcyv4jZ6zYWEdAeVM2DQKyroCSVc+-x4kHY-9wtLPX=ZdN9Tg@mail.gmail.com>


>>>> -                       if (!fctx->filter_region(region, fctx))
>>>> +                       if (fctx->filter_region &&
>>>> +                                       !fctx->filter_region(region,
>>>> fctx))
>>> No, I don't agree with these change. The filter_bus() and
>>> filter_region() callbacks are mandatory because, like list, you need
>>> to be prepared to handle dimms underneath the bus, and namespaces
>>> underneath a region.
>>
>> The filter_bus() and filter_region() callbacks are mandatory in ndctl list,
>> but not in ndctl monitor.
>> Current design only monitors the events coming from DIMMs, does not monitor
>> events coming from buses yet. When monitoring the events coming from DIMMs,
>> we just need to put the DIMMs which match filters into a linklist using
>> filter_dimm(). No additional actions will be taken on buses. It is the
>> similar
>> case as in ndctl list, when the command does not contain [--buses] option,
>> filter_bus() will always return true.
> Right, I'd prefer that you create a ->filter_bus() implementation for
> your use case that always returns true rather than change the core to
> allow a ->filter_bus() routine to be omitted. This way the core stays
> generic for all use cases and the bus monitoring can be added without
> touching the core.
>
>
OK, I see.
Thank you very much.


_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm