Subject: Re: CVS commit: src
To: None <rmind@NetBSD.org>
From: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
List: source-changes
Date: 02/09/2008 21:29:32
> Hello,
> 
> Sorry, had a problems with free-time.

no problem.

> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> > > cvs rdiff -r1.44 -r1.45 src/sys/sys/sched.h
> > 
> > the change of struct sched_param broke the ABI.  please revert.
> 
> Sad thing that I have separately asked about this point in <tech-kern>...
> 
> http://www.netbsd.org/~rmind/sched_fixes_1.diff
> 
> Can you please take a look?

i'll take a look, thanks.

> yamt@mwd.biglobe.ne.jp (YAMAMOTO Takashi) wrote:
> > while you can make it compatible, is there any reason to
> > make our native one (_sched_setaffinity) inherit the limitations?
> > 
> > iirc recent linux has a set of macros to allocate dynamic-sized cpusets.
> > can you check?
> 
> I have checked - yes (but no manual pages). CPU_ALLOC_SIZE(), CPU_ALLOC(),
> CPU_FREE() and a bunch of CPU_*_S() macros are used, but as I understand they
> still use static cpu_set_t (?).

my guess is that they are trying to migrate from static cpu_set_t by
introducing these macros.

YAMAMOTO Takashi

> 
> Not sure what solution we should pick.
> 
> -- 
> Best regards,
> Mindaugas
> www.NetBSD.org