[PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Stan C. Smith]

Contain pthread definitions with ifdef HAVE_LIBPTHREAD

Signed-off-by: stan smith <stan.smith-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

diff --git a/opensm/include/vendor/osm_vendor_ibumad.h b/opensm/include/vendor/osm_vendor_ibumad.h
index 0a4692d..d523412 100644
--- a/opensm/include/vendor/osm_vendor_ibumad.h
+++ b/opensm/include/vendor/osm_vendor_ibumad.h
@@ -160,8 +160,13 @@ typedef struct _osm_vendor {
 	char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
 	vendor_match_tbl_t mtbl;
 	umad_port_t umad_port;
+#ifdef HAVE_LIBPTHREAD
 	pthread_mutex_t cb_mutex;
 	pthread_mutex_t match_tbl_mutex;
+#else
+	cl_mutex_t cb_mutex;
+	cl_mutex_t match_tbl_mutex;
+#endif
 	int umad_port_id;
 	void *receiver;
 	int issmfd;



--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Sean Hefty]

>Contain pthread definitions with ifdef HAVE_LIBPTHREAD
>
>Signed-off-by: stan smith <stan.smith@intel.com>
>
>diff --git a/opensm/include/vendor/osm_vendor_ibumad.h
>b/opensm/include/vendor/osm_vendor_ibumad.h
>index 0a4692d..d523412 100644
>--- a/opensm/include/vendor/osm_vendor_ibumad.h
>+++ b/opensm/include/vendor/osm_vendor_ibumad.h
>@@ -160,8 +160,13 @@ typedef struct _osm_vendor {
> 	char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
> 	vendor_match_tbl_t mtbl;
> 	umad_port_t umad_port;
>+#ifdef HAVE_LIBPTHREAD
> 	pthread_mutex_t cb_mutex;
> 	pthread_mutex_t match_tbl_mutex;
>+#else
>+	cl_mutex_t cb_mutex;
>+	cl_mutex_t match_tbl_mutex;
>+#endif

complib is available on both platforms, why not just use it?

Alternately, #define cl_mutex_t pthread_mutex_t and avoid the #ifdef's

RE: RE: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Sean Hefty]

>Alternately, #define cl_mutex_t pthread_mutex_t and avoid the #ifdef's

switch that define around...

RE: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Smith, Stan]

Hefty, Sean wrote:
>> Contain pthread definitions with ifdef HAVE_LIBPTHREAD
>>
>> Signed-off-by: stan smith <stan.smith-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
>>
>> diff --git a/opensm/include/vendor/osm_vendor_ibumad.h
>> b/opensm/include/vendor/osm_vendor_ibumad.h
>> index 0a4692d..d523412 100644
>> --- a/opensm/include/vendor/osm_vendor_ibumad.h
>> +++ b/opensm/include/vendor/osm_vendor_ibumad.h
>> @@ -160,8 +160,13 @@ typedef struct _osm_vendor {
>>      char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
>>      vendor_match_tbl_t mtbl; umad_port_t umad_port;
>> +#ifdef HAVE_LIBPTHREAD
>>      pthread_mutex_t cb_mutex;
>>      pthread_mutex_t match_tbl_mutex;
>> +#else
>> +    cl_mutex_t cb_mutex;
>> +    cl_mutex_t match_tbl_mutex;
>> +#endif
>
> complib is available on both platforms, why not just use it?
>
> Alternately, #define cl_mutex_t pthread_mutex_t and avoid the #ifdef's

Pthreads have been workin just fine, no need fix something that is not broken.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Smith, Stan]

Stan C. Smith wrote:
> Contain pthread definitions with ifdef HAVE_LIBPTHREAD
>
> Signed-off-by: stan smith <stan.smith-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
>
> diff --git a/opensm/include/vendor/osm_vendor_ibumad.h
> b/opensm/include/vendor/osm_vendor_ibumad.h index 0a4692d..d523412
> 100644 --- a/opensm/include/vendor/osm_vendor_ibumad.h
> +++ b/opensm/include/vendor/osm_vendor_ibumad.h
> @@ -160,8 +160,13 @@ typedef struct _osm_vendor {
>       char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
>       vendor_match_tbl_t mtbl;
>       umad_port_t umad_port;
> +#ifdef HAVE_LIBPTHREAD
>       pthread_mutex_t cb_mutex;
>       pthread_mutex_t match_tbl_mutex;
> +#else
> +     cl_mutex_t cb_mutex;
> +     cl_mutex_t match_tbl_mutex;
> +#endif
>       int umad_port_id;
>       void *receiver;
>       int issmfd;

The use of ifdef HAVE_LIBPTHREAD was based on existing HAVE_LIBPTHREAD usage in opensm/include/opensm/osm_stats.h @ line 97.
If the community prefers the use of macros then these two patches can be skipped.
Please advise on path to travel.

Stan.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Sasha Khapyorsky]

On 09:07 Thu 05 Nov     , Smith, Stan wrote:
> 
> The use of ifdef HAVE_LIBPTHREAD was based on existing HAVE_LIBPTHREAD usage in opensm/include/opensm/osm_stats.h @ line 97.

Right, but it was used due to differences in the functionality between
cl_event_t and pthread stuff. I would be happy to remove those #ifdef's
upon such differences elimination.

> If the community prefers the use of macros then these two patches can be skipped.

I think that macros are nicer.

Sasha
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Sasha Khapyorsky]

On 13:05 Wed 04 Nov     , Sean Hefty wrote:
> >Contain pthread definitions with ifdef HAVE_LIBPTHREAD
> >
> >Signed-off-by: stan smith <stan.smith@intel.com>
> >
> >diff --git a/opensm/include/vendor/osm_vendor_ibumad.h
> >b/opensm/include/vendor/osm_vendor_ibumad.h
> >index 0a4692d..d523412 100644
> >--- a/opensm/include/vendor/osm_vendor_ibumad.h
> >+++ b/opensm/include/vendor/osm_vendor_ibumad.h
> >@@ -160,8 +160,13 @@ typedef struct _osm_vendor {
> > 	char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
> > 	vendor_match_tbl_t mtbl;
> > 	umad_port_t umad_port;
> >+#ifdef HAVE_LIBPTHREAD
> > 	pthread_mutex_t cb_mutex;
> > 	pthread_mutex_t match_tbl_mutex;
> >+#else
> >+	cl_mutex_t cb_mutex;
> >+	cl_mutex_t match_tbl_mutex;
> >+#endif
> 
> complib is available on both platforms, why not just use it?

We don't have cl_mutex_t in linux's implementation.

I didn't know that complib on windows already implements semantics which
are similar to most useful pthread things. That is good, and seems that
an easy porting could be done by using macros like:

#define pthread_* cl_thread_*

Sasha

RE: [PATCH 1/2] contain pthreads defs with ifdef HAVE_LIBPTHREAD

[Smith, Stan]

Sasha Khapyorsky wrote:
> On 09:07 Thu 05 Nov     , Smith, Stan wrote:
>>
>> The use of ifdef HAVE_LIBPTHREAD was based on existing
>> HAVE_LIBPTHREAD usage in opensm/include/opensm/osm_stats.h @ line
>> 97.
>
> Right, but it was used due to differences in the functionality between
> cl_event_t and pthread stuff. I would be happy to remove those
> #ifdef's upon such differences elimination.
>
>> If the community prefers the use of macros then these two patches
>> can be skipped.
>
> I think that macros are nicer.
>
> Sasha

OK, will proceed with macros. Skip previously submitted patch set.
I will submit patches to remove relevant HAVE_LIBPTHREAD as not all can/should be removed.