clk: timestamps

clk: timestamps

[Subhashini Rao Beerisetty]

Hi All,

In the kernel code I see it supports CLOCK_REALTIME \ CLOCK_MONOTONIC  \
CLOCK_MONOTONIC_RAW timestamps. Can someone explain what?s major difference
between those three modes and when to use which one?


I?ve N number of Linux machines in the network with the same software
running. Basically I need to collect the timestamps in kernel mode in all
the machines and then compare, which(either CLOCK_REALTIME or
CLOCK_MONOTIC_RAW or MONOTONIC) one would be the correct way?

Thanks,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20180626/40f5b091/attachment.html>

where and when the kernel destroy devices memory if there were no matched driver for it?

[kipade]

Hi, all

As my question, would kernel destroy deivces struct if there were no matched
driver attached on it? If so, what's a suite time to drop them, and where?

Thanks.

where and when the kernel destroy devices memory if there were no matched driver for it?

[Greg KH]

On Wed, Jun 27, 2018 at 01:52:55PM +0800, kipade wrote:
> Hi, all
> 
> As my question, would kernel destroy deivces struct if there were no matched
> driver attached on it?

No, only when the device is removed from the system does that structure
get destroyed.

greg k-h

clk: timestamps

[Yann Droneaud]

Hi,

Le mardi 26 juin 2018 ? 19:44 +0530, Subhashini Rao Beerisetty a
?crit :
> 
> In the kernel code I see it supports CLOCK_REALTIME \ CLOCK_MONOTONIC
>  \ CLOCK_MONOTONIC_RAW timestamps. Can someone explain what?s major
> difference between those three modes and when to use which one?
> 

A generic answer could be found here:
http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/time.h.html

> I?ve N number of Linux machines in the network with the same software
> running. Basically I need to collect the timestamps in kernel mode in
> all the machines and then compare, which(either CLOCK_REALTIME or
> CLOCK_MONOTIC_RAW or MONOTONIC) one would be the correct way?
> 

Probably none. Anyway, please find some hints below:

None of the monotonic clocks can be used because they're not supposed
to be synchronized accross the network

You're left with CLOCK_REALTIME, which is not synchronized accross the
network by default.

So before doing timestamp comparison, you would need to setup time
synchronisation between hosts on your network: NTP to synchronise hosts
 to the same seconds with couple of millisecond precision, PTP for sub
millisecond precision.

Time synchronisation is a tough problem but it's required if you want
to be able to compare timestamp accross hosts in a network.

Regards.

-- 
Yann Droneaud
OPTEYA

clk: timestamps

[Subhashini Rao Beerisetty]

On Wed, Jun 27, 2018 at 1:17 PM, Yann Droneaud <ydroneaud@opteya.com> wrote:

> Hi,
>
> Le mardi 26 juin 2018 ? 19:44 +0530, Subhashini Rao Beerisetty a
> ?crit :
> >
> > In the kernel code I see it supports CLOCK_REALTIME \ CLOCK_MONOTONIC
> >  \ CLOCK_MONOTONIC_RAW timestamps. Can someone explain what?s major
> > difference between those three modes and when to use which one?
> >
>
> A generic answer could be found here:
> http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/time.h.html
>
> > I?ve N number of Linux machines in the network with the same software
> > running. Basically I need to collect the timestamps in kernel mode in
> > all the machines and then compare, which(either CLOCK_REALTIME or
> > CLOCK_MONOTIC_RAW or MONOTONIC) one would be the correct way?
> >
>
> Probably none. Anyway, please find some hints below:
>
> None of the monotonic clocks can be used because they're not supposed
> to be synchronized accross the network
>
> You're left with CLOCK_REALTIME, which is not synchronized accross the
> network by default.
>


Does printk log timestamp by default uses the CLOCK_REALTIME?

>
> So before doing timestamp comparison, you would need to setup time
> synchronisation between hosts on your network: NTP to synchronise hosts
>  to the same seconds with couple of millisecond precision, PTP for sub
> millisecond precision.
>
Can you point me on how to setup time synchronisation between hosts on
network.


>
> Time synchronisation is a tough problem but it's required if you want
> to be able to compare timestamp accross hosts in a network.
>
> Regards.
>
> --
> Yann Droneaud
> OPTEYA
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20180627/794dba3f/attachment.html>

where and when the kernel destroy devices memory if there were no matched driver for it?

[kipade]

If so, I make a driver was compiled as module and inserted by insmod
after some time the system booted, the driver could find the device if it
did not compiled as module. But, the fact is, my driver could not find
its device which was created during booting from device tree.

On 2018?06?27? 13:57, Greg KH wrote:
> On Wed, Jun 27, 2018 at 01:52:55PM +0800, kipade wrote:
>> Hi, all
>>
>> As my question, would kernel destroy deivces struct if there were no matched
>> driver attached on it?
> No, only when the device is removed from the system does that structure
> get destroyed.
>
> greg k-h
>
> _______________________________________________
> Kernelnewbies mailing list
> Kernelnewbies at kernelnewbies.org
> https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies

where and when the kernel destroy devices memory if there were no matched driver for it?

[Greg KH]

A: http://en.wikipedia.org/wiki/Top_post
Q: Were do I find info about this thing called top-posting?
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

A: No.
Q: Should I include quotations after my reply?

http://daringfireball.net/2007/07/on_top

On Wed, Jun 27, 2018 at 04:57:56PM +0800, kipade wrote:
> If so, I make a driver was compiled as module and inserted by insmod
> after some time the system booted, the driver could find the device if it
> did not compiled as module. But, the fact is, my driver could not find
> its device which was created during booting from device tree.

I don't know, sorry, you are going to have to be a lot more specific.
Do you see the device in /sys/devices/ ?  If so, are you sure your
driver probe function is called?  If not, then work on fixing that.

Do you have a pointer to your driver code anywhere we can review it to
see what you are trying to do?

thanks,

greg k-h

clk: timestamps

[valdis.kletnieks at vt.edu]

On Tue, 26 Jun 2018 19:44:21 +0530, Subhashini Rao Beerisetty said:

> I???ve N number of Linux machines in the network with the same software
> running. Basically I need to collect the timestamps in kernel mode in all
> the machines and then compare, which(either CLOCK_REALTIME or
> CLOCK_MONOTIC_RAW or MONOTONIC) one would be the correct way?

That depends a lot on what question you're trying to answer by
comparing timestamps.  Read up on the behavior of all three (as
they have different semantics), and see which one is the best fit
for your specific problem.