On Sep 3, 2014, at 2:07 AM, Savolainen, Petri (NSN - FI/Espoo) 
<[email protected]> wrote:

> 
> 
>> -----Original Message-----
>> From: ext Anders Roxell [mailto:[email protected]]
>> Sent: Tuesday, September 02, 2014 10:07 PM
>> To: Petri Savolainen
>> Cc: [email protected]
>> Subject: Re: [lng-odp] [PATCH v2 1/2] Use timer resolution
>> 
>> On 2014-09-02 15:47, Petri Savolainen wrote:
>>> Use resolution and min/max tmo values from timer create call.
>>> Use common nanosec time defines.
>>> 
>>> Signed-off-by: Petri Savolainen <[email protected]>
>>> ---
>>> platform/linux-generic/include/api/odp_time.h |  5 ++
>>> platform/linux-generic/odp_timer.c            | 67
>> ++++++++++++++++++++-------
>>> 2 files changed, 55 insertions(+), 17 deletions(-)
>>> 
>>> diff --git a/platform/linux-generic/include/api/odp_time.h
>> b/platform/linux-generic/include/api/odp_time.h
>>> index 188d1fe..1770223 100644
>>> --- a/platform/linux-generic/include/api/odp_time.h
>>> +++ b/platform/linux-generic/include/api/odp_time.h
>>> @@ -21,6 +21,11 @@ extern "C" {
>>> 
>>> #include <odp_std_types.h>
>>> 
>>> +/* Time in nanoseconds */
>>> +#define ODP_TIME_USEC 1000UL       /*< Microsecond in nsec */
>>> +#define ODP_TIME_MSEC 1000000UL    /*< Millisecond in nsec */
>>> +#define ODP_TIME_SEC  1000000000UL /*< Second in nsec */
>> 
>> you need to add "/**<" and not "/*<"
> 
> OK.
> 
>> 
>>> +
>>> 
>>> /**
>>>  * Current time in CPU cycles
>>> diff --git a/platform/linux-generic/odp_timer.c b/platform/linux-
>> generic/odp_timer.c
>>> index 1bf37f9..6e89f5c 100644
>>> --- a/platform/linux-generic/odp_timer.c
>>> +++ b/platform/linux-generic/odp_timer.c
>>> @@ -6,6 +6,7 @@
>>> 
>>> #include <odp_timer.h>
>>> #include <odp_timer_internal.h>
>>> +#include <odp_time.h>
>>> #include <odp_buffer_pool_internal.h>
>>> #include <odp_internal.h>
>>> #include <odp_atomic.h>
>>> @@ -18,9 +19,14 @@
>>> 
>>> #include <string.h>
>>> 
>>> +#define USEC ODP_TIME_USEC
>>> +#define MSEC ODP_TIME_MSEC
>>> +#define SEC  ODP_TIME_SEC
>> 
>> no, why duplicate it?
> 
> What's the problem? 

The only problem I see with using the shorting names is possible name space 
collision with pre-existing code. As much as I like the shorter names the 
ODP_TIME_XXX prefix is safer and the one I would have picked. If we already 
have ODP_TIME_XXX defines then using them instead of creating new defines just 
to save a few keystrokes is not worth it IMO. As for readability they both are 
readable and when I see ODP_TIME_XXX I know for a fact it is an ODP define.
> 
> For example this is shorter and more readable...
> sec  = res / SEC;
> nsec = res - sec*SEC;
> 
> than this.
> sec  = res / ODP_TIME_SEC;
> nsec = res - sec*ODP_TIME_SEC;
> 
> 
> SEC is too generic definition for ODP API, so ODP_TIME_SEC was used (for 
> nanoseconds, time bases could be added also but nanoseconds is the current 
> default).
> 
> -Petri
> 
> 
> _______________________________________________
> lng-odp mailing list
> [email protected]
> http://lists.linaro.org/mailman/listinfo/lng-odp

Keith Wiles, Principal Technologist with CTO office, Wind River mobile 
972-213-5533


_______________________________________________
lng-odp mailing list
[email protected]
http://lists.linaro.org/mailman/listinfo/lng-odp

Reply via email to