svn commit: r274739 - head/sys/mips/conf
Warner Losh
imp at bsdimp.com
Mon Nov 24 23:22:29 UTC 2014
> On Nov 24, 2014, at 4:16 PM, Dag-Erling Smørgrav <des at des.no> wrote:
>
> Mark R V Murray <mark at grondar.org> writes:
>
>>> On 24 Nov 2014, at 15:01, Ian Lepore <ian at FreeBSD.org> wrote:
>>>
>>>
>>> The logging change was pretty simple:
>>>
>>> Index: kern/subr_bus.c
>>
>> Thanks.
>>
>>> The low number of devices is pretty typical. Sometimes there'll be
>>> another uart. Sometimes an i2c eeprom.
>>
>> OK - a good example of a low-entropy system then.
>>
>>> I have no idea what's up with the first 3 unchanging numbers, but I
>>> suspect a big part of the explanation for the other numbers is the 32KHz
>>> clock it's getting the numbers from. There used to be a clock running
>>> at about 2.8MHz, but the source code for that driver seems to have
>>> disappeared from FreeBSD at some point between 8.x and -current.
>>
>> What really bothers me is that these should be the difference between
>> 2 essentially similar numbers; times not all that far apart, yet some
>> of the numbers are truly massive.
>
> They're not just massive, they're preposterous, as is the fact that most
> of them are absolutely identical, or just one bit off, from one run to
> the next. My best guess is that get_cyclecount() is broken.
What’s the minimum specs required for get_cyclecount()? Not all of the
boxes that Ian was posting about have high-resolution time-keeping counters
in hardware… Maybe there’s some underlying expectation for this function
that these systems either aren’t providing or can’t provide.
Warner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.freebsd.org/pipermail/freebsd-arch/attachments/20141124/8dc0a375/attachment.sig>
More information about the freebsd-arch
mailing list