Re: [PATCH v2] xen/mce: don't issue error message for failed /dev/mcelog registration

[Juergen Gross <jgross@suse.com>]

On 13/06/17 21:30, Joe Perches wrote:
> On Tue, 2017-06-13 at 21:27 +0200, Juergen Gross wrote:
>> On 13/06/17 18:31, Joe Perches wrote:
>>> On Tue, 2017-06-13 at 18:13 +0200, Juergen Gross wrote:
>>>> On 13/06/17 17:20, Ingo Molnar wrote:
>>>>> * Juergen Gross <jgross@suse.com> wrote:
>>>>>
>>>>>> When running under Xen as dom0 /dev/mcelog is being registered by Xen
>>>>>> instead of the normal mcelog driver. Avoid an error message being
>>>>>> issued by the mcelog driver in this case. Instead issue an informative
>>>>>> message that Xen has registered the device.
>>>
>>> []
>>>>>> diff --git a/arch/x86/kernel/cpu/mcheck/dev-mcelog.c b/arch/x86/kernel/cpu/mcheck/dev-mcelog.c
>>>
>>> []
>>>>>> @@ -388,9 +388,16 @@ static __init int dev_mcelog_init_device(void)
>>>>>>  	/* register character device /dev/mcelog */
>>>>>>  	err = misc_register(&mce_chrdev_device);
>>>>>>  	if (err) {
>>>>>> -		pr_err("Unable to init device /dev/mcelog (rc: %d)\n", err);
>>>>>> -		return err;
>>>>>> +		if (err == -EBUSY)
>>>>>> +			/* Xen dom0 might have registered the device already. */
>>>>>> +			pr_info("Unable to init device /dev/mcelog, already registered");
>>>>>> +		else {
>>>>>> +			pr_err("Unable to init device /dev/mcelog (rc: %d)\n",
>>>>>> +			       err);
>>>>>> +			return err;
>>>>>> +		}
>>>>>
>>>>> Please only use balanced curly braces in conditional statements.
>>>>
>>>> Okay, will change.
>>>
>>> Perhaps better is to reverse the test
>>>
>>> 		if (err != -EBUSY) {
>>> 			pr_err("Unable to ....", err);
>>> 			return err;
>>> 		}
>>> 		pr_info("etc...");
>>
>> Okay.
>>
>>>
>>> 		[ rest of code at this indentation ]
>>>
>>> but it looks like you added a logic defect too and
>>> this code should be:
>>>
>>> 	if (err) {
>>> 		if (err == -EBUSY)
>>> 			pr_info(...)
>>> 		else
>>> 			pr_err(...)
>>> 		return err;
>>> 	}
>>>
>>> or less indented using
>>>
>>> 	err = misc_register(&mce_chrdev_device);
>>> 	if (err == -EBUSY) {
>>> 		pr_info(...);
>>> 		return err;
>>> 	} else if (err) {
>>> 		pr_err(...);
>>> 		return err;
>>> 	}
>>
>> I didn't want to omit the call to mce_register_decode_chain() in the Xen
>> case.
> 
> You should definitely mention that behavior change
> in the changelog.

Looking more thoroughly it seems I really don't need the call for Xen.
So thanks for bringing it up. :-)


Juergen