fixing ACPI wakealarm test?

Jeffrey Hugo jhugo at codeaurora.org
Wed Sep 27 17:41:26 UTC 2017


On 9/26/2017 5:33 PM, Colin Ian King wrote:
> On 27/09/17 00:27, Al Stone wrote:
>> So, some platforms don't really have an ACPI wake alarm, but the test is set to
>> essentially always be run:
>>
>> FWTS_REGISTER("wakealarm", &wakealarm_ops, FWTS_TEST_ANYTIME, FWTS_FLAG_BATCH |
>> FWTS_FLAG_ROOT_PRIV)
>>
>> Is there a recommended way to make a test more hardware specific?  Granted, if
>> I run FWTS on the platform, I can double check and make sure /sys/class/rtc/rtc0
>> exists (or something like that); if I run FWTS elsewhere, though, that's clearly
>> not going to work.  I know I've added options in the past for architectures, but
>> this is specific to an individual machine and it would seem like overkill to
>> put in options for specific machines.
>>
>> Suggestions?
> 
> Perhaps the simplest approach is to skip all the subsequent tests if
> wakealarm_test1() fails to find a working interface.

Its been a while since I've looked at this, but if I recall correctly, 
wakealarm_tes1() always finds a "working" interface due to how the Linux 
kernel abstracts RTC.

What I've seen on our platform is that UEFI does not implement the wake 
alarm functionality, however the kernel is happy to "fake" it, until you 
try setting an actual alarm, then you start getting failures.

I'd be interested in determining a way to address this, since it would 
alleviate the burden of blacklisting the test and explaining why.

> 
>>
>> BTW, I was not able to file a bug for this without registering with Ubuntu One;
>> is that really necessary?
>>
> Afraid so :-(
> 
> Colin
> 
> 


-- 
Jeffrey Hugo
Qualcomm Datacenter Technologies as an affiliate of Qualcomm 
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.



More information about the fwts-devel mailing list