6:50 p.m.
On 04/02/2015 07:44 PM, Christoph Hellwig wrote:
On Thu, Apr 02, 2015 at 09:01:14AM -0700, Dan Williams wrote:
>>> If anything I think these should be dev_dbg().
>>
>> We do not have a dev at any of this point, and it does not
>> belong to any specific device.
>
> Ah, true this is prior to the driver attaching... that said it seems
> more relevant to print from probe() (where we do have a device) than
> init where the device may remain idle due to some other policy.
>
>> Also I would like this
>> _info and not _dbg so to always have it, also for production.
>> See the chatter for a single SCSI disk the minimum we need
>> is just the small print that tells all that we need (for now)
>
> Not sure we want to follow so closely in the footsteps of SCSI's chattiness.
Defintively not! A single dev_info at ->probe time sounds ok, something
like:
dev_info(dev, "registering region [0x%pa:0x%zx]\n",
&pmem->phys_addr, pmem->size);
but there are plenty other drivers not that chatty, e.g. virtio and we're
doing just fine for them.
For me, my print is just the exact amount.
So I will see in my dmesg:
[ +0.000000] user: [mem 0x0000000100000000-0x000000015fffffff] persistent (type 12)
[ +0.000000] user: [mem 0x0000000160000000-0x00000001dfffffff] persistent (type 12)
...
[ +0.000537] pmem: init 2 devices => 0
So I have all the information. And I know the driver was actually loaded
successfully on the expected two regions.
Your print above will give me two prints with information I already have.
All I want to know from users logs is that the driver was actually loaded
successful or not. And when it was unloaded. I think this is the bear minimum
that gives me all the information I need. Less then this I would be missing
a very important event.
Thanks
Boaz