2006-03-09 23:33:51

by Roland Dreier

Subject: Re: [PATCH 10 of 20] ipath - support for userspace apps using core driver

> + yield(); /* don't hog the cpu */

You probably don't want to do this -- yield() means "put me at the
tail of the runqueue." I think cond_resched() is more likely what you
want.

> +#endif
> +/* END_NOSHIP_TO_OPENIB */

uhh... and I don't see an #if to match that #endif.

- R.

2006-03-09 23:56:44

by Bryan O'Sullivan

Subject: Re: [PATCH 10 of 20] ipath - support for userspace apps using core driver

On Thu, 2006-03-09 at 15:33 -0800, Roland Dreier wrote:
> > + yield(); /* don't hog the cpu */
>
> You probably don't want to do this -- yield() means "put me at the
> tail of the runqueue." I think cond_resched() is more likely what you
> want.

OK. I don't think it much matters either way.

> > +#endif
> > +/* END_NOSHIP_TO_OPENIB */
>
> uhh... and I don't see an #if to match that #endif.

The code got drain bamaged by the patch mangling script. The real code
contains a mess of crap to support kernels back to 2.6.9, which gets
automatically stripped, except when it gets broken as above.

Next rev will be clean in this regard.

<b