Optimizing universe somewhat

2003-02-20 Thread Ruslan Ermilov
On Wed, Feb 19, 2003 at 07:40:19AM -0800, Ruslan Ermilov wrote:
 ru  2003/02/19 07:40:19 PST
 
   Modified files:
 .Makefile 
   Log:
   Fixed universe.
   
   Folded pc98 into the common case.
   Retired ${JFLAG} (``make -jX universe'' should work).
   
   Revision  ChangesPath
   1.276 +30 -34src/Makefile
 
Would it be too bad (in anyone's opinion) if we optimize this
a bit to build modules only once for each architecture, with
buildworld (-DMODULES_WITH_WORLD)?  That would speed-up the
creation of universe somewhat, but has one bad side effect of
polluting userland build with kernel stuff, but is easiest
to implement.

Another option would be to build modules only for the first
kernel for a given arch, whatever it happens to be.  This is
still not quite good as kernel/modules may or may not be
independently broken.

Yet another option would be to still build modules once for
a given architecture, but independently of kernels and world.

Before I go for implementing this or that, I'd like people's
opinion on that.


Cheers,
-- 
Ruslan Ermilov  Sysadmin and DBA,
[EMAIL PROTECTED]   Sunbay Software AG,
[EMAIL PROTECTED]  FreeBSD committer,
+380.652.512.251Simferopol, Ukraine

http://www.FreeBSD.org  The Power To Serve
http://www.oracle.com   Enabling The Information Age



msg52758/pgp0.pgp
Description: PGP signature


Re: Optimizing universe somewhat

2003-02-20 Thread phk
In message [EMAIL PROTECTED], Ruslan Ermilov writes:

--/9DWx/yDrRhgMJTb
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Wed, Feb 19, 2003 at 07:40:19AM -0800, Ruslan Ermilov wrote:
 ru  2003/02/19 07:40:19 PST
=20
   Modified files:
 .Makefile=20
   Log:
   Fixed universe.
  =20
   Folded pc98 into the common case.
   Retired ${JFLAG} (``make -jX universe'' should work).
  =20
   Revision  ChangesPath
   1.276 +30 -34src/Makefile
=20
Would it be too bad (in anyone's opinion) if we optimize this
a bit to build modules only once for each architecture, with
buildworld (-DMODULES_WITH_WORLD)?  That would speed-up the
creation of universe somewhat, but has one bad side effect of
polluting userland build with kernel stuff, but is easiest
to implement.

I think we should build the modules as specified by the kernels.

Nothing prevents you from adding

makeoptions MODULES_OVERRIDE=acpi linux

or similar to your kernels.

Universe just takes time, and that's it.  Don't try to optimize it
if the result is less coverage.

-- 
Poul-Henning Kamp   | UNIX since Zilog Zeus 3.20
[EMAIL PROTECTED] | TCP/IP since RFC 956
FreeBSD committer   | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message



Re: Optimizing universe somewhat

2003-02-20 Thread Ruslan Ermilov
On Thu, Feb 20, 2003 at 10:44:50AM +0100, [EMAIL PROTECTED] wrote:
 In message [EMAIL PROTECTED], Ruslan Ermilov writes:
 
 --/9DWx/yDrRhgMJTb
 Content-Type: text/plain; charset=us-ascii
 Content-Disposition: inline
 Content-Transfer-Encoding: quoted-printable
 
 On Wed, Feb 19, 2003 at 07:40:19AM -0800, Ruslan Ermilov wrote:
  ru  2003/02/19 07:40:19 PST
 =20
Modified files:
  .Makefile=20
Log:
Fixed universe.
   =20
Folded pc98 into the common case.
Retired ${JFLAG} (``make -jX universe'' should work).
   =20
Revision  ChangesPath
1.276 +30 -34src/Makefile
 =20
 Would it be too bad (in anyone's opinion) if we optimize this
 a bit to build modules only once for each architecture, with
 buildworld (-DMODULES_WITH_WORLD)?  That would speed-up the
 creation of universe somewhat, but has one bad side effect of
 polluting userland build with kernel stuff, but is easiest
 to implement.
 
 I think we should build the modules as specified by the kernels.
 
 Nothing prevents you from adding
 
   makeoptions MODULES_OVERRIDE=acpi linux
 
 or similar to your kernels.
 
 Universe just takes time, and that's it.  Don't try to optimize it
 if the result is less coverage.
 
Okay, and this _is_ the easiest to implement, though I've found
some bogons with putting ``makeoptions NO_MODULES=yes'' that
need to be addressed.


Cheers,
-- 
Ruslan Ermilov  Sysadmin and DBA,
[EMAIL PROTECTED]   Sunbay Software AG,
[EMAIL PROTECTED]  FreeBSD committer,
+380.652.512.251Simferopol, Ukraine

http://www.FreeBSD.org  The Power To Serve
http://www.oracle.com   Enabling The Information Age



msg52762/pgp0.pgp
Description: PGP signature


Re: Optimizing universe somewhat

2003-02-20 Thread Dag-Erling Smorgrav
Ruslan Ermilov [EMAIL PROTECTED] writes:
 Okay, and this _is_ the easiest to implement, though I've found
 some bogons with putting ``makeoptions NO_MODULES=yes'' that
 need to be addressed.

makeoptions MODULES_OVERRIDE= should work fine.

DES
-- 
Dag-Erling Smorgrav - [EMAIL PROTECTED]

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message


Re: Optimizing universe somewhat

2003-02-20 Thread Ruslan Ermilov
On Thu, Feb 20, 2003 at 10:33:21PM +0100, Dag-Erling Smorgrav wrote:
 Ruslan Ermilov [EMAIL PROTECTED] writes:
  Okay, and this _is_ the easiest to implement, though I've found
  some bogons with putting ``makeoptions NO_MODULES=yes'' that
  need to be addressed.
 
 makeoptions MODULES_OVERRIDE= should work fine.
 
I haven't looked in-depth (yet), but I recall that NO_MODULES
passed in makeoptions doesn't have the immediate effect (i.e.,
it still causes ``make obj'' to be run for modules).

I'm likely to look at this now, if I don't fall asleep before.


Cheers,
-- 
Ruslan Ermilov  Sysadmin and DBA,
[EMAIL PROTECTED]   Sunbay Software AG,
[EMAIL PROTECTED]   FreeBSD committer,
+380.652.512.251Simferopol, Ukraine

http://www.FreeBSD.org  The Power To Serve
http://www.oracle.com   Enabling The Information Age


pgp0.pgp
Description: PGP signature