Re: [Powerpc-utils-devel] lparstat man page, smt entry

[Nathan F. <nf...@li...>]

I think there are a couple of things that have led to this. First the lparstat
command was first done when you had either smt off or on, where smt on == smt 2.
The later addition of multiple smt modes has confused this. The other reason
is that the lparstat command on Linux was originally written to be a clone of
aix command.

I think at this point we should drop the output that indicates smt is 'on'.
Instead we should just print the actual smt mode, or off.

-Nathan

On 11/17/2014 05:26 PM, Thomas Falcon wrote:
> The lparstat man page specifies "if there are two SMT threads, the row is displayed as 'on.' However, if there are more than two SMT threads,  the number of SMT threads is displayed."  The actual SMT output would be Dedicated, which I think is because of a bug.  I sent a patch for that earlier today, but the documentation does not match the new output. 
> 
> For example, the machine that I tested on gave an output of "on" when SMT was set to four and "2" when SMT was set to two. 
> 
> # ppc64_cpu --smt=4
> # ./lparstat
> 
> System Configuration
> type=Shared mode=Capped smt=On lcpu=4 mem=3899904 kB cpus=16 ent=2.00
> 
> %user  %sys %wait %idle physc %entc lbusy  vcsw phint
> ----- ----- ----- ----- ----- ----- ----- ----- -----
>  0.00  0.01  0.00 99.99  0.00  0.00  0.01 9744131     0
> ltcbrazos2-lp08:~/powerpc-utils/src # ppc
> ppc        ppc32      ppc64      ppc64_cpu 
> 
> # ppc64_cpu --smt=2
> # ./lparstat
> 
> System Configuration
> type=Shared mode=Capped smt=2 lcpu=4 mem=3899904 kB cpus=16 ent=2.00
> 
> %user  %sys %wait %idle physc %entc lbusy  vcsw phint
> ----- ----- ----- ----- ----- ----- ----- ----- -----
>  0.00  0.00  0.00 99.99  0.00  0.00  0.00 9749696     0
> 
> After some testing, it looks to me that the output is "on" when using the maximum possible SMT value.  I created a patch to change the documentation, but I wanted to be sure that I was correct before sending it. 
>