Linux-2.4.1-pre11

Linux-2.4.1-pre11

[Linus Torvalds]

I just uploaded it to kernel.org, and I expect that I'll do the final
2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
pre-kernel works for you..

The main noticeable things in pre11 are fixing some bugs that crept in
after 2.4.0 - the block device queuing improvements could lose wakeups
under extreme load by multiple clients, and the vmscanning "get rid of
special return codes for shared memory" thing had missed a bit.

This should also fix the VIA IDE driver issues (if you want safe, do NOT
enable auto-dma), and the reported problems with hpt366 controllers and
IBM drives. Hopefully these were the last major IDE issues for a while.

Also, can people who have had unhappy relationships with their eepro100
please try to cuddle and make up again? The eepro100 changes should fix
the problem of having posted writes that basically made some of the timing
not work out.

		Linus

-----
pre11:
 - Trond Myklebust: NFS/RPC client SMP fixes
 - rth: alpha pyxis and cabriolet fixes
 - remove broken sys_wait4() declarations
 - disable radeon debugging code
 - VIA IDE driver should not enable autodma unless asked for
 - Andrey Savochkin: eepro100 update. Should fix the resource timing problems.
 - Jeff Garzik: via82cxxx_audio update
 - YMF7xx PCI audio update: get rid of old broken driver, make new
   driver handle legacy control too. 
 - fix missed wakeup on block device request list
 - hpt366 controller doesn't play nice with some IBM harddisks
 - remove inode pages from the page cache only after having removed them
   from the page tables.
 - shared memory out-of-swap writepage() fixup (no more magic return)

pre10:
 - got a few too-new R128 #defines in the Radeon merge. Fix.
 - tulip driver update from Jeff Garzik
 - more cpq and DAC elevator fixes from Jens. Looks good.
 - Petr Vandrovec: nicer ncpfs behaviour
 - Andy Grover: APCI update
 - Cort Dougan: PPC update
 - David Miller: sparc updates
 - David Miller: networking updates
 - Neil Brown: RAID5 fixes

pre9:
 - cpq array driver elevator fixes 
 - merge radeon driver from X CVS tree
 - ispnp cleanups
 - emu10k unlock on error fixes
 - hpfs doesn't allow truncate to larger

pre8:
 - Don't drop a megabyte off the old-style memory size detection
 - remember to UnlockPage() in ramfs_writepage()
 - 3c59x driver update from Andrew Morton
 - egcs-1.1.2 miscompiles depca: workaround by Andrew Morton
 - dmfe.c module init fix: Andrew Morton
 - dynamic XMM support. Andrea Arkangeli.
 - ReiserFS merge
 - USB hotplug updates/fixes
 - boots on real i386 machines
 - blk-14 from Jens Axboe
 - fix DRM R128/AGP dependency
 - fix n_tty "canon" mode SMP race
 - ISDN fixes
 - ppp UP deadlock attack fix
 - FAT fat_cache SMP race fix
 - VM balancing tuning
 - Locked SHM segment deadlock fix
 - fork() page table copy race fix

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Dieter Nützel]

> I just uploaded it to kernel.org, and I expect that I'll do the final
> 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> pre-kernel works for you..

Hello Linus,

can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release 
updated: 1-25-2001)  in 2.4.1 final?
ftp://download.intel.com/technology/iapc/acpi/downloads/acpica-linux-20010125.tar.gz)

Thanks,	
	Dieter

BTW Have a nice trip.
-- 
Dieter Nützel
Graduate Student, Computer Science

University of Hamburg
Department of Computer Science
Cognitive Systems Group
Vogt-Kölln-Straße 30
D-22527 Hamburg, Germany

email: nuetzel@kogs.informatik.uni-hamburg.de
@home: Dieter.Nuetzel@hamburg.de
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Linus Torvalds]

[ Cc'd to linux-mm, because I'm trying to also explain what kind of code I
  like, and because the more people who understand how I judge the
  "goodness" of code, the happier I hope I will be. ]

On Sun, 28 Jan 2001, Marcelo Tosatti wrote:
> 
> Why you have not applied the conditional background page aging patch?

Because it's only performance, not stability, and every time I look at the
patch it looks ugly to me. Let me explain.

Quite frankly, I suspect I'd like the thing much more if it didn't have a
"background" flag, but a more specific thing. I hate overloading functions
with two things and then having flags that change the behaviour of them. I
like arguments that say "do THIS". I do not like arguments that say "I'm
THIS, do whatever you want that you think will satisfy me".

For example, I don't like how the current one has the "oneshot" parameter.
I'd much rather just have

	deactivated = refill_inactive_scan(count);

where the "count" would be the thing we ask for, and "deactivated" would
obviously be how many we got. No "background" or "oneshot". The "oneshot"
stuff in particular makes the

	while (refill_inactive_scan(DEF_PRIORITY, one)) ...

logic look completely buggered: first we ask for it to exit after having
found _one_, and then we have a loop that does this "count" times. Where's
the logic?

But at the same time, the "oneshot" parameter _does_ fulfill my need for
well-defined behaviour. It says "DO THIS!". Or at least it tries to. It
doesn't say "I like the color blue, so try to take that into account when
you do whatever you like to do".

Your "background" paramater makes the whole parameter pretty fluffy, in my
opinion.. Why do we exit early when '!background'? Where's the
_philosophy_ of the function? What do the arguments really mean? You can't
read the callers and understand what the callers really are trying to do..

Now, to me the "background" check should be in the caller. Th ecaller
knows that it's doing background activity, so the _caller_ should be the
one that say "when in the background, DO THIS!". Instead of letting the
function try to decide on its own what it is we want when we're in the
background.

And conversely the "how many pages should be try to de-activate" logic
should be in "refill_inactive_scan". So to me, the following calling
convetion would make some amount of sense:

 - refill_inactive_scan() calling convention change: it should be

	int refill_inactive_scan(int nr_to_deactivate);

   and basically return the number of pages it de-activated, with the
   parameter being how many pages we _want_ to de-activate. We've already
   stopped using the priority (it's always DEF_PROPROTY), so we can drop
   that. And I'd like the other parameter to _mean_ something, if you see
   what I'm saying.

   "Try to scan for THIS many" is a meaning. "Try to scan in the
   background" doesn't really mean anything. What does "background" mean
   to the scanning logic? It means something to the caller, but not to the
   scanner.

 - kswapd, before it calls refill_inactive_scan(), would basically do
   something like

	nr = free_shortage() + inactive_shortage();
	if (nr > MAXSCAN)
		nr = MAXSCAN;
	if (nr || bg_page_aging)
		refill_inactive_scan(nr);

   which again has some _meaning_. You can point to it and say: we want to
   de-activate "nr" pages because we're short on memory. But even if the
   answer is "zero", we may want to do some aging in the background, so
   "refill_inactive_scan()" can know that a zero argument means that we
   don't _really_ want to deactivate anything.

   See how you can _read_ the calling code directly? Show the above five
   lines to a programmer who hasn't even seen what it is that
   "refill_inactive_scan()" actually does, and I bet he can guess what
   we're trying to do.

   That's what I mean with _meaning_. Which the current code lacks, and
   which your change makes even less of.

   And note how "refill_inactive_scan(0)" is automatically a logical
   special case. It tells refill_inactive_scan() that we don't actually
   want to _really_ deactivate anything, so we're obviously just doing the
   aging. 

 - refill_inactive() can do

	count -= refill_inactive_scan(count);

   instead of the current while-loop. Again, you can pretty much see from
   that one line what it tries to do.

Now, I'm not saying the above is how it must be done. The above is meant
more as an example of an interface and a logic that I can follow, and that
I think is more appropriate for programming. Programming is never about
asking the computer to do something and telling it what the constraints
are (a constraint would be "do this, but remember that we're a background
process"). Programming is about telling the computer what to do. You
should NOT say:

 "please scan some pages in the ackground"

but instead say

 "scan and age the active list, and deactive up to 5 pages"

You're the captain. Don't say "Please". Say "Make it so".

		Linus

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Linux-2.4.1-pre11

[Linus Torvalds]

On Sun, 28 Jan 2001, Dieter Nützel wrote:

> > I just uploaded it to kernel.org, and I expect that I'll do the final
> > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> > pre-kernel works for you..
> 
> Hello Linus,
> 
> can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release 
> updated: 1-25-2001)  in 2.4.1 final?

Does it fix stuff? Andrew?

		Linus

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Jeff Garzik]

Linus Torvalds wrote:
> On Sun, 28 Jan 2001, Dieter Nützel wrote:
> > > I just uploaded it to kernel.org, and I expect that I'll do the final
> > > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> > > pre-kernel works for you..
> >
> > Hello Linus,
> >
> > can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release
> > updated: 1-25-2001)  in 2.4.1 final?
> 
> Does it fix stuff? Andrew?

I'm running it here..  No problems yet on my Toshiba test laptop, which
is the same behavior (for me) on 2.4.0-pre10 vanilla.

ACPI changelog, from
http://developer.intel.com/technology/IAPC/acpi/index.htm follows...


> Summary of changes for this label: 01_25_01
> 
> Core ACPI CA Subsystem:
> Restructured the implementation of object store support within the 
> interpreter.  This includes support for the Store operator as well
> as any ASL operators that include a target operand.
> 
> Partially implemented support for Implicit Result-to-Target conversion.
> This is when a result object is converted on the fly to the type of
> an existing target object.  Completion of this support is pending
> further analysis of the ACPI specification concerning this matter.
> 
> CPU-specific code has been removed from the subsystem (hardware directory).
> 
> New Power Management Timer functions added
> 
> Linux OS Services Layer (OSL):
> Moved system state transition code to the core, fixed it, and modified
> Linux OSL accordingly.
> 
> Fixed C2 and C3 latency calculations.
> 
> We no longer use the compilation date for the version message on
> initialization, but retrieve the version from AcpiGetSystemInfo().
> 
> Incorporated for fix Sony VAIO machines.
> 
> Documentation:
> The Programmer Reference has been updated and reformatted.
> 
> ASL Compiler:
> Version X2013:
> Fixed a problem where the line numbering and error reporting could get out
> of sync in the presence of multiple include files.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Derek Wildstar]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, 28 Jan 2001, Linus Torvalds wrote:

> On Sun, 28 Jan 2001, Dieter Nützel wrote:
>
> > > I just uploaded it to kernel.org, and I expect that I'll do the final
> > > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> > > pre-kernel works for you..
> >
> > Hello Linus,
> >
> > can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release
> > updated: 1-25-2001)  in 2.4.1 final?
>
> Does it fix stuff? Andrew?
>
I just tried adding this to 2.4.1-pre11 and the compile failed, the
problem i've been having with ACPI is the kernel soft-hangs just after
loading the tables.  Using APM or no power management at all doesn't hang.

Hardware: Dell Inspiron 5000e, bios A04 (latest provided by Dell)

I have heard of some bugs in Dell's ACPI implementation, but since so many
people have dell machines it may be worth trying to work around, or even
detect the buggy implementation and disable ACPI with an error printed.

I can donate time if needed, just let me know what needs to be tested.

Thanks,
dwild
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.4 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iEYEARECAAYFAjp0mHAACgkQhazASHM/AFMr0ACgjPE3+hzS05N5gt1qvl5Pgue7
smcAoIITSnkaawBXj+zToaajc9NgfrlK
=n4fr
-----END PGP SIGNATURE-----


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: linux-2.4.1-pre11

[Arnaldo Carvalho de Melo]

Em Sun, Jan 28, 2001 at 06:55:25PM -0500, Louis Garcia escreveu:
> I am getting messages everytime I use the network from my RH7 + 
> kernel-2.4.1-pre11 system:
> 
> modprobe: modprobe: Can't locate module net-pf-10
> 
> I have checked my .config  and can't find that modules. This does not 
> happen with 2.4.0 kernel, only with the latest pre series maybe pre7 on.

you haven't included support for IPv6 and your distribution initscripts is
trying to load it for some reason, two solutions:

1. enable IPv6 in your kernel build
2. disable it in your /etc/modules.conf file, like this:

alias net-pf-10 off

- Arnaldo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Drew Bertola]

Andrew's latest ACPI fixes (acpica-linux-20000125 patched against
2.4.0) compile fine here and don't hang on my Vaio after loading
tables.

That's a start.  I'll play around some more.

Jeff Garzik writes:
> Linus Torvalds wrote:
> > On Sun, 28 Jan 2001, Dieter Nützel wrote:
> > > > I just uploaded it to kernel.org, and I expect that I'll do the final
> > > > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> > > > pre-kernel works for you..
> > >
> > > Hello Linus,
> > >
> > > can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release
> > > updated: 1-25-2001)  in 2.4.1 final?
> > 
> > Does it fix stuff? Andrew?
> 
> I'm running it here..  No problems yet on my Toshiba test laptop, which
> is the same behavior (for me) on 2.4.0-pre10 vanilla.
> 
> ACPI changelog, from
> http://developer.intel.com/technology/IAPC/acpi/index.htm follows...
> 
> 
> > Summary of changes for this label: 01_25_01
> > 
> > Core ACPI CA Subsystem:
> > Restructured the implementation of object store support within the 
> > interpreter.  This includes support for the Store operator as well
> > as any ASL operators that include a target operand.
> > 
> > Partially implemented support for Implicit Result-to-Target conversion.
> > This is when a result object is converted on the fly to the type of
> > an existing target object.  Completion of this support is pending
> > further analysis of the ACPI specification concerning this matter.
> > 
> > CPU-specific code has been removed from the subsystem (hardware directory).
> > 
> > New Power Management Timer functions added
> > 
> > Linux OS Services Layer (OSL):
> > Moved system state transition code to the core, fixed it, and modified
> > Linux OSL accordingly.
> > 
> > Fixed C2 and C3 latency calculations.
> > 
> > We no longer use the compilation date for the version message on
> > initialization, but retrieve the version from AcpiGetSystemInfo().
> > 
> > Incorporated for fix Sony VAIO machines.
> > 
> > Documentation:
> > The Programmer Reference has been updated and reformatted.
> > 
> > ASL Compiler:
> > Version X2013:
> > Fixed a problem where the line numbering and error reporting could get out
> > of sync in the presence of multiple include files.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/

-- 
Drew Bertola  | Send a text message to my pager or cell ... 
              |   http://jpager.com/Drew

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Drew Bertola]

Drew Bertola writes:
> Andrew's latest ACPI fixes (acpica-linux-20000125 patched against
> 2.4.0) compile fine here and don't hang on my Vaio after loading
> tables.
> 
> That's a start.  I'll play around some more.

Unfortunately, pcmcia modules fail to load.  I can't understand the
interaction.  

The message displayed on boot when starting the service says:

ds: no socket drivers loaded

-- 
Drew Bertola  | Send a text message to my pager or cell ... 
              |   http://jpager.com/Drew

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

linux-2.4.1-pre11

[Louis Garcia]

I am getting messages everytime I use the network from my RH7 + 
kernel-2.4.1-pre11 system:

modprobe: modprobe: Can't locate module net-pf-10

I have checked my .config  and can't find that modules. This does not 
happen with 2.4.0 kernel, only with the latest pre series maybe pre7 on.

Lou

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Arnaldo Carvalho de Melo]

Em Sun, Jan 28, 2001 at 08:10:41PM -0500, Michael H. Warfield escreveu:
> 	Damn...  So much for typing too fast...  Screwed it up...

toooo fast 8)
 
> > 	Patch was in /pub/linux/kernel/v2.4/test/patch-2.4.1-pre11.gz
> 
>  	Patch was in /pub/linux/kernel/test/patch-2.4.1-pre11.gz
> 
> 	Cut and paste screwup.  Sorry.

screwed again, there's no test directory, testing is the right one:

http://www.kernel.org/pub/linux/kernel/testing

- Arnaldo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Drew Bertola]

Drew Bertola writes:
> Drew Bertola writes:
> > Andrew's latest ACPI fixes (acpica-linux-20000125 patched against
> > 2.4.0) compile fine here and don't hang on my Vaio after loading
> > tables.
> > 
> > That's a start.  I'll play around some more.
> 
> Unfortunately, pcmcia modules fail to load.  I can't understand the
> interaction.  
> 
> The message displayed on boot when starting the service says:
> 
> ds: no socket drivers loaded

I resolved this issue by using yenta_socket.

For my RH7.0 system, /etc/sysconfig/pcmcia needs to be edited to show:

PCIC=yenta_socket

-- 
Drew Bertola  | Send a text message to my pager or cell ... 
              |   http://jpager.com/Drew

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[David D.W. Downey]

Hi Linus,
	Sorry to bother you. I'm trying to find where you uploaded
linux-2.4.1-pre11.

I was on ftp.kernel.org and ftp.us.kernel.org and could not find it in the
/pub/linux/kernel/v2.4 or /pub/linux/kernel/v2.4/test-kernels/
directories. Is it somewhere different?

Also, Alan, I grabbed your patch-2.4.0-ac11.gz file from your directory on
ftp.kernel.org. Does this contain the updated VIA IDE support that Linus
was talking about in the 2.4.1-pre11?

I'm thinking either kernel.org hasn't posted the 2.4.1-pre11 or I totally
misunderstand the directory layout on kernel.org.

A URL to the right patch or, preferably, full source for 2.4.1-pre11 would
be great.

Thanks,

David D.W. Downey


On Sun, 28 Jan 2001, Linus Torvalds wrote:

> 
> I just uploaded it to kernel.org, and I expect that I'll do the final
> 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> pre-kernel works for you..
> 
> The main noticeable things in pre11 are fixing some bugs that crept in
> after 2.4.0 - the block device queuing improvements could lose wakeups
> under extreme load by multiple clients, and the vmscanning "get rid of
> special return codes for shared memory" thing had missed a bit.
> 
> This should also fix the VIA IDE driver issues (if you want safe, do NOT
> enable auto-dma), and the reported problems with hpt366 controllers and
> IBM drives. Hopefully these were the last major IDE issues for a while.
> 
> Also, can people who have had unhappy relationships with their eepro100
> please try to cuddle and make up again? The eepro100 changes should fix
> the problem of having posted writes that basically made some of the timing
> not work out.
> 
> 		Linus
> 
> -----
> pre11:
>  - Trond Myklebust: NFS/RPC client SMP fixes
>  - rth: alpha pyxis and cabriolet fixes
>  - remove broken sys_wait4() declarations
>  - disable radeon debugging code
>  - VIA IDE driver should not enable autodma unless asked for
>  - Andrey Savochkin: eepro100 update. Should fix the resource timing problems.
>  - Jeff Garzik: via82cxxx_audio update
>  - YMF7xx PCI audio update: get rid of old broken driver, make new
>    driver handle legacy control too. 
>  - fix missed wakeup on block device request list
>  - hpt366 controller doesn't play nice with some IBM harddisks
>  - remove inode pages from the page cache only after having removed them
>    from the page tables.
>  - shared memory out-of-swap writepage() fixup (no more magic return)
> 
> pre10:
>  - got a few too-new R128 #defines in the Radeon merge. Fix.
>  - tulip driver update from Jeff Garzik
>  - more cpq and DAC elevator fixes from Jens. Looks good.
>  - Petr Vandrovec: nicer ncpfs behaviour
>  - Andy Grover: APCI update
>  - Cort Dougan: PPC update
>  - David Miller: sparc updates
>  - David Miller: networking updates
>  - Neil Brown: RAID5 fixes
> 
> pre9:
>  - cpq array driver elevator fixes 
>  - merge radeon driver from X CVS tree
>  - ispnp cleanups
>  - emu10k unlock on error fixes
>  - hpfs doesn't allow truncate to larger
> 
> pre8:
>  - Don't drop a megabyte off the old-style memory size detection
>  - remember to UnlockPage() in ramfs_writepage()
>  - 3c59x driver update from Andrew Morton
>  - egcs-1.1.2 miscompiles depca: workaround by Andrew Morton
>  - dmfe.c module init fix: Andrew Morton
>  - dynamic XMM support. Andrea Arkangeli.
>  - ReiserFS merge
>  - USB hotplug updates/fixes
>  - boots on real i386 machines
>  - blk-14 from Jens Axboe
>  - fix DRM R128/AGP dependency
>  - fix n_tty "canon" mode SMP race
>  - ISDN fixes
>  - ppp UP deadlock attack fix
>  - FAT fat_cache SMP race fix
>  - VM balancing tuning
>  - Locked SHM segment deadlock fix
>  - fork() page table copy race fix
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/
> 

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Michael H. Warfield]

On Sun, Jan 28, 2001 at 04:55:51PM -0800, David D.W. Downey wrote:

> Hi Linus,
> 	Sorry to bother you. I'm trying to find where you uploaded
> linux-2.4.1-pre11.

> I was on ftp.kernel.org and ftp.us.kernel.org and could not find it in the
> /pub/linux/kernel/v2.4 or /pub/linux/kernel/v2.4/test-kernels/
> directories. Is it somewhere different?

	Patch was in /pub/linux/kernel/v2.4/test/patch-2.4.1-pre11.gz

> Also, Alan, I grabbed your patch-2.4.0-ac11.gz file from your directory on
> ftp.kernel.org. Does this contain the updated VIA IDE support that Linus
> was talking about in the 2.4.1-pre11?

> I'm thinking either kernel.org hasn't posted the 2.4.1-pre11 or I totally
> misunderstand the directory layout on kernel.org.

	Close...

> A URL to the right patch or, preferably, full source for 2.4.1-pre11 would
> be great.

> Thanks,

> David D.W. Downey

	[...]

	Mike
-- 
 Michael H. Warfield    |  (770) 985-6132   |  mhw@WittsEnd.com
  (The Mad Wizard)      |  (678) 463-0932   |  http://www.wittsend.com/mhw/
  NIC whois:  MHW9      |  An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471    |  possible worlds.  A pessimist is sure of it!

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[David D.W. Downey]

> 	Patch was in /pub/linux/kernel/v2.4/test/patch-2.4.1-pre11.gz
> 



I'm on ftp.kernel.org right this second in /pub/linux/kernel/v2.4/

There is only a test-kernels/ subdir there, not a test/

test-kernels/ does not contain the patch.


David


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[David D.W. Downey]

Nevermind. I found it.

It's actually residing in /pub/linux/kernel/testing/ and NOT in
/pub/linux/kernel/v2.4/ or it's subdirs.


David D.W. Downey


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Michael H. Warfield]

	Damn...  So much for typing too fast...  Screwed it up...

On Sun, Jan 28, 2001 at 07:59:47PM -0500, Michael H. Warfield wrote:
> On Sun, Jan 28, 2001 at 04:55:51PM -0800, David D.W. Downey wrote:

> > Hi Linus,
> > 	Sorry to bother you. I'm trying to find where you uploaded
> > linux-2.4.1-pre11.

> > I was on ftp.kernel.org and ftp.us.kernel.org and could not find it in the
> > /pub/linux/kernel/v2.4 or /pub/linux/kernel/v2.4/test-kernels/
> > directories. Is it somewhere different?

> 	Patch was in /pub/linux/kernel/v2.4/test/patch-2.4.1-pre11.gz

 	Patch was in /pub/linux/kernel/test/patch-2.4.1-pre11.gz

	Cut and paste screwup.  Sorry.

	[...]

	Mike
-- 
 Michael H. Warfield    |  (770) 985-6132   |  mhw@WittsEnd.com
  (The Mad Wizard)      |  (678) 463-0932   |  http://www.wittsend.com/mhw/
  NIC whois:  MHW9      |  An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471    |  possible worlds.  A pessimist is sure of it!

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Luc de Louw]

Hi Linus

On Sun, 28 Jan 2001, Linus Torvalds wrote:


> 
> I just uploaded it to kernel.org, and I expect that I'll do the final
> 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
> pre-kernel works for you..

yes, it works :-)

> 
> The main noticeable things in pre11 are fixing some bugs that crept in
> after 2.4.0 - the block device queuing improvements could lose wakeups
> under extreme load by multiple clients, and the vmscanning "get rid of
> special return codes for shared memory" thing had missed a bit.
> 
> This should also fix the VIA IDE driver issues (if you want safe, do NOT
> enable auto-dma), and the reported problems with hpt366 controllers and
> IBM drives. Hopefully these were the last major IDE issues for a while.

It works fine for me

> 
> Also, can people who have had unhappy relationships with their eepro100
> please try to cuddle and make up again? The eepro100 changes should fix
> the problem of having posted writes that basically made some of the timing
> not work out.

I'll try that at monday ( In a couple of hours I'm at work) .....

> 
> 		Linus
> 

<snip>

Have a nice trip and enjoy NYC! weatherforcast for monday and wednesday
looks great, visit the empire state buiöding its gread (if you get the
time) I wish I could at LinuxWold too .... Have fun

rgds

Luc de Louw



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Bob Chiodini]

How about:

http://www.us.kernel.org/pub/linux/kernel/testing/patch-2.4.1-pre11.gz

Bob...

"David D.W. Downey" wrote:

> >       Patch was in /pub/linux/kernel/v2.4/test/patch-2.4.1-pre11.gz
> >
>
> I'm on ftp.kernel.org right this second in /pub/linux/kernel/v2.4/
>
> There is only a test-kernels/ subdir there, not a test/
>
> test-kernels/ does not contain the patch.
>
> David
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/

--

rchiodini@cfl.rr.com



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Derek Wildstar]

On Sun, 28 Jan 2001, Derek Wildstar wrote:

> OK, tried the patch and it worked, don't remember the exact errors with
> the .tar.gz, there were 7 or so undefined references.
>
> ACPI soft-hangs one step before (after looking at the non-debug source
> it may be the same place) it did last time, right after it prints:
>
> ACPI: Core Subsystem version [20010125]
>
> grabbing the debug version now to see if i can get more info.

OK, the debug version printed the following:

 tbxface-0089: ACPI Tables successfully loaded
Parsing Methods:................(more dots, i can count if needed)
173 Control Methods found and parsed (602 nodes total)
ACPI Namespace successfully loaded at root c042f718
ACPI: Core Subsystem version [20010125]
evxfevnt-0082: Transition to ACPI mode successful
Executing device _INI methods:........

The cursor stays at the end of the last line.

Let me know if there is anything you would like me to try.

-dwild

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: linux-2.4.1-pre11

[Louis Garcia]

Arnaldo Carvalho de Melo wrote:

> Em Sun, Jan 28, 2001 at 06:55:25PM -0500, Louis Garcia escreveu:
> 
>> I am getting messages everytime I use the network from my RH7 + 
>> kernel-2.4.1-pre11 system:
>> 
>> modprobe: modprobe: Can't locate module net-pf-10
>> 
>> I have checked my .config  and can't find that modules. This does not 
>> happen with 2.4.0 kernel, only with the latest pre series maybe pre7 on.
> 
> 
> you haven't included support for IPv6 and your distribution initscripts is
> trying to load it for some reason, two solutions:
> 
> 1. enable IPv6 in your kernel build
> 2. disable it in your /etc/modules.conf file, like this:
> 
> alias net-pf-10 off
> 
> - Arnaldo
> 
> 
> 
Anyone have an idea where in the initscripts does this happen?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Dieter Nützel]

Am Sonntag, 28. Januar 2001 22:46 schrieb Linus Torvalds:
> On Sun, 28 Jan 2001, Dieter Nützel wrote:
> > > I just uploaded it to kernel.org, and I expect that I'll do the final
> > > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that
> > > the pre-kernel works for you..
> >
> > Hello Linus,
> >
> > can we please see Andrew's latest ACPI fixes ([Acpi] ACPI source release
> > updated: 1-25-2001)  in 2.4.1 final?
>
> Does it fix stuff? Andrew?

I am the loser :-(
2.4.1-pre10 (with Andrew's ACPI fixes included) and
2.4.1-pre11 + 1-25-2001 patch bring back the pppd slowdown on my system.

2.4.1-pre9 was fine...

AMD K7
MSI MS-6167 Rev. 1.0B (Irongate C4)

-Dieter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Linus Torvalds]

On Sun, 28 Jan 2001, David D.W. Downey wrote:
> 
> I was on ftp.kernel.org and ftp.us.kernel.org and could not find it in the
> /pub/linux/kernel/v2.4 or /pub/linux/kernel/v2.4/test-kernels/
> directories. Is it somewhere different?

All my "current" test-patches are always under /pub/linux/kernel/testing.

The "v2.4/test-kernels/ directory is for historic files - the
test-kernels that led up to 2.4.0

		Linus

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Jeff Garzik]

Drew Bertola wrote:
> 
> Drew Bertola writes:
> > Andrew's latest ACPI fixes (acpica-linux-20000125 patched against
> > 2.4.0) compile fine here and don't hang on my Vaio after loading
> > tables.
> >
> > That's a start.  I'll play around some more.
> 
> Unfortunately, pcmcia modules fail to load.  I can't understand the
> interaction.
> 
> The message displayed on boot when starting the service says:
> 
> ds: no socket drivers loaded

Personally I advocate building all pcmcia stuff into the kernel.  It has
never failed before, and its core hardware on your laptop, so it will
always be there.  Why bother with modules at all for core
functionality...

	Jeff


-- 
Jeff Garzik       | "You see, in this world there's two kinds of
Building 1024     |  people, my friend: Those with loaded guns
MandrakeSoft      |  and those who dig. You dig."  --Blondie
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Ed Tomlinson]

Feedback is nice...

The way you would like refill_inactive defined raises a question or two.  The
patches now have it runing in two modes, which you object to.  The way I read
what you say the real way to fix this would be to refactor refill_inactive.

result = refill_inactive(count) 

which has one goal of getting count pages inactivated.  There should also be 
a function to do the bg_aging, something like.

result = balance_activates(background)

The goal of which is to ensure that for every page we activate we also age a 
page down.  This might move some pages to the inactive list and might trigger 
some swapouts.  

This could also be hidden in refill_inactive triggered by a count of zero but 
this implies that refill_inactive must have a side effect of decrementing the 
background counter which is not the nicest thing to do.

Comments?
Ed Tomlinson


On January 28, 2001 02:32 pm, Linus Torvalds wrote:
> [ Cc'd to linux-mm, because I'm trying to also explain what kind of code I
>   like, and because the more people who understand how I judge the
>   "goodness" of code, the happier I hope I will be. ]
>
> On Sun, 28 Jan 2001, Marcelo Tosatti wrote:
> > Why you have not applied the conditional background page aging patch?
>
> Because it's only performance, not stability, and every time I look at the
> patch it looks ugly to me. Let me explain.
>
> Quite frankly, I suspect I'd like the thing much more if it didn't have a
> "background" flag, but a more specific thing. I hate overloading functions
> with two things and then having flags that change the behaviour of them. I
> like arguments that say "do THIS". I do not like arguments that say "I'm
> THIS, do whatever you want that you think will satisfy me".
>
> For example, I don't like how the current one has the "oneshot" parameter.
> I'd much rather just have
>
> 	deactivated = refill_inactive_scan(count);
>
> where the "count" would be the thing we ask for, and "deactivated" would
> obviously be how many we got. No "background" or "oneshot". The "oneshot"
> stuff in particular makes the
>
> 	while (refill_inactive_scan(DEF_PRIORITY, one)) ...
>
> logic look completely buggered: first we ask for it to exit after having
> found _one_, and then we have a loop that does this "count" times. Where's
> the logic?
>
> But at the same time, the "oneshot" parameter _does_ fulfill my need for
> well-defined behaviour. It says "DO THIS!". Or at least it tries to. It
> doesn't say "I like the color blue, so try to take that into account when
> you do whatever you like to do".
>
> Your "background" paramater makes the whole parameter pretty fluffy, in my
> opinion.. Why do we exit early when '!background'? Where's the
> _philosophy_ of the function? What do the arguments really mean? You can't
> read the callers and understand what the callers really are trying to do..
>
> Now, to me the "background" check should be in the caller. Th ecaller
> knows that it's doing background activity, so the _caller_ should be the
> one that say "when in the background, DO THIS!". Instead of letting the
> function try to decide on its own what it is we want when we're in the
> background.
>
> And conversely the "how many pages should be try to de-activate" logic
> should be in "refill_inactive_scan". So to me, the following calling
> convetion would make some amount of sense:
>
>  - refill_inactive_scan() calling convention change: it should be
>
> 	int refill_inactive_scan(int nr_to_deactivate);
>
>    and basically return the number of pages it de-activated, with the
>    parameter being how many pages we _want_ to de-activate. We've already
>    stopped using the priority (it's always DEF_PROPROTY), so we can drop
>    that. And I'd like the other parameter to _mean_ something, if you see
>    what I'm saying.
>
>    "Try to scan for THIS many" is a meaning. "Try to scan in the
>    background" doesn't really mean anything. What does "background" mean
>    to the scanning logic? It means something to the caller, but not to the
>    scanner.
>
>  - kswapd, before it calls refill_inactive_scan(), would basically do
>    something like
>
> 	nr = free_shortage() + inactive_shortage();
> 	if (nr > MAXSCAN)
> 		nr = MAXSCAN;
> 	if (nr || bg_page_aging)
> 		refill_inactive_scan(nr);
>
>    which again has some _meaning_. You can point to it and say: we want to
>    de-activate "nr" pages because we're short on memory. But even if the
>    answer is "zero", we may want to do some aging in the background, so
>    "refill_inactive_scan()" can know that a zero argument means that we
>    don't _really_ want to deactivate anything.
>
>    See how you can _read_ the calling code directly? Show the above five
>    lines to a programmer who hasn't even seen what it is that
>    "refill_inactive_scan()" actually does, and I bet he can guess what
>    we're trying to do.
>
>    That's what I mean with _meaning_. Which the current code lacks, and
>    which your change makes even less of.
>
>    And note how "refill_inactive_scan(0)" is automatically a logical
>    special case. It tells refill_inactive_scan() that we don't actually
>    want to _really_ deactivate anything, so we're obviously just doing the
>    aging.
>
>  - refill_inactive() can do
>
> 	count -= refill_inactive_scan(count);
>
>    instead of the current while-loop. Again, you can pretty much see from
>    that one line what it tries to do.
>
> Now, I'm not saying the above is how it must be done. The above is meant
> more as an example of an interface and a logic that I can follow, and that
> I think is more appropriate for programming. Programming is never about
> asking the computer to do something and telling it what the constraints
> are (a constraint would be "do this, but remember that we're a background
> process"). Programming is about telling the computer what to do. You
> should NOT say:
>
>  "please scan some pages in the ackground"
>
> but instead say
>
>  "scan and age the active list, and deactive up to 5 pages"
>
> You're the captain. Don't say "Please". Say "Make it so".
>
> 		Linus
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org.  For more info on Linux MM,
> see: http://www.linux.eu.org/Linux-MM/
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Linux-2.4.1-pre11

[Meelis Roos]

LT> I just uploaded it to kernel.org, and I expect that I'll do the final
LT> 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
LT> pre-kernel works for you..
[...]
LT> pre10:
[...]
LT>  - Andy Grover: APCI update

I tried test11 yesterday. Works fine except when I enable ACPI instead of
APM. Never tried this before but now I decided to give it a try.
After initialising ACPI, the machine became slow as a 386 or a 286.

AMD Duron 600, Soltek 75KV mobo w/VIA KT133 chipset, UP-APIC kernel, IDE-only
system.

-- 
Meelis Roos (mroos@linux.ee)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Harold Oga]

On Mon, Jan 29, 2001 at 12:02:33PM +0200, Meelis Roos wrote:
>LT> I just uploaded it to kernel.org, and I expect that I'll do the final
>LT> 2.4.1 tomorrow, before leaving for NY and LinuxWorld. Please test that the
>LT> pre-kernel works for you..
>[...]
>LT> pre10:
>[...]
>LT>  - Andy Grover: APCI update
>
>I tried test11 yesterday. Works fine except when I enable ACPI instead of
>APM. Never tried this before but now I decided to give it a try.
>After initialising ACPI, the machine became slow as a 386 or a 286.
>
>AMD Duron 600, Soltek 75KV mobo w/VIA KT133 chipset, UP-APIC kernel, IDE-only
>system.
Hi,
   I'm seeing similar problems with my system on 2.4.1-pre10.  This is an
AMD Thunderbird 900, MSI K7T Pro2-A mobo w/VIA KT133 chipset, UP, ide/scsi
mix.  2.4.1-pre10 works fine if I don't configure ACPI.  I'll try to
narrow down when this problem started showing up later today, as I
initially moved from 2.4.1-pre3 straight to 2.4.1-pre10.

-Harold
-- 
"Life sucks, deal with it!"
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Ondrej Sury]

[-- Attachment #1: Type: text/plain, Size: 801 bytes --]

Harold Oga <ogah@home.com> writes:

> Hi,
>    I'm seeing similar problems with my system on 2.4.1-pre10.  This is an
> AMD Thunderbird 900, MSI K7T Pro2-A mobo w/VIA KT133 chipset, UP, ide/scsi
> mix.  2.4.1-pre10 works fine if I don't configure ACPI.  I'll try to
> narrow down when this problem started showing up later today, as I
> initially moved from 2.4.1-pre3 straight to 2.4.1-pre10.

It's something between pre9 and pre10, and probably it's VIA chipset
problem.

-- 
Ondřej Surý <ondrej@globe.cz>         Globe Internet s.r.o. http://globe.cz/
Tel: +420235365000   Fax: +420235365009         Pláničkova 1, 162 00 Praha 6
Mob: +420605204544   ICQ: 24944126             Mapa: http://globe.namape.cz/
GPG fingerprint:          CC91 8F02 8CDE 911A 933F  AE52 F4E6 6A7C C20D F273

[-- Attachment #2: Type: application/pgp-signature, Size: 276 bytes --]

RE: Linux-2.4.1-pre11

[Grover, Andrew]

> On Sun, 28 Jan 2001, Dieter Nützel wrote:
> 
> > > I just uploaded it to kernel.org, and I expect that I'll 
> do the final
> > > 2.4.1 tomorrow, before leaving for NY and LinuxWorld. 
> Please test that the
> > > pre-kernel works for you..
> > 
> > Hello Linus,
> > 
> > can we please see Andrew's latest ACPI fixes ([Acpi] ACPI 
> source release 
> > updated: 1-25-2001)  in 2.4.1 final?
> 
> Does it fix stuff? Andrew?

This fixes VAIO and power down. I've sent the patch separately, so please
apply it when you get the chance.

Thanks -- Regards -- Andy

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Linux-2.4.1-pre11

[Marcelo Tosatti]

On Mon, 29 Jan 2001, Rik van Riel wrote:

> Wouldn't that be:
> 	if (nr < MINSCAN)
> 		nr = MINSCAN;
> 
> Btw, if we fix the inactive_shortage() function to take
> per-zone inactive shortage into account, we can just skip
> scanning if we only have a free shortage but no inactive
> shortage.

inactive_shortage() already takes this into account. 

> Page_launder() then needs to be changed a bit too. If it's
> called from a user process, it can have the same behaviour
> it has now.

Btw, look at this part of code from kswapd: 

                /* 
                 * We go to sleep if either the free page shortage
                 * or the inactive page shortage is gone. We do this
                 * because:
                 * 1) we need no more free pages   or
                 * 2) the inactive pages need to be flushed to disk,
                 *    it wouldn't help to eat CPU time now ...
                 *
                 * We go to sleep for one second, but if it's needed
                 * we'll be woken up earlier...
                 */
                if (!free_shortage() || !inactive_shortage()) {
                        interruptible_sleep_on_timeout(&kswapd_wait, HZ);


kswapd goes to sleep if there is no free shortage, even if the inactive
list is under shortage.

Why not refill the inactive list when the inactive list is under
shortage? :)


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Linux-2.4.1-pre11

[Rik van Riel]

On Sun, 28 Jan 2001, Linus Torvalds wrote:

> Quite frankly, I suspect I'd like the thing much more if it
> didn't have a "background" flag, but a more specific thing.

> For example, I don't like how the current one has the "oneshot" parameter.
> I'd much rather just have
> 
> 	deactivated = refill_inactive_scan(count);

I have this buried somewhere in the 3 weeks of code I wrote
while at linux.conf.au. I'll dig up some parts of my code and
I'll give it to you ASAP.

>  - refill_inactive_scan() calling convention change: it should be
> 
> 	int refill_inactive_scan(int nr_to_deactivate);
> 
>    and basically return the number of pages it de-activated, with the
>    parameter being how many pages we _want_ to de-activate.

One moment please, I'll make you a patch.  ;)

>  - kswapd, before it calls refill_inactive_scan(), would basically do
>    something like
> 
> 	nr = free_shortage() + inactive_shortage();
> 	if (nr > MAXSCAN)
> 		nr = MAXSCAN;
> 	if (nr || bg_page_aging)
> 		refill_inactive_scan(nr);
> 
>    which again has some _meaning_. You can point to it and say: we want to
>    de-activate "nr" pages because we're short on memory. But even if the
>    answer is "zero", we may want to do some aging in the background, so
>    "refill_inactive_scan()" can know that a zero argument means that we
>    don't _really_ want to deactivate anything.

Wouldn't that be:
	if (nr < MINSCAN)
		nr = MINSCAN;

Btw, if we fix the inactive_shortage() function to take
per-zone inactive shortage into account, we can just skip
scanning if we only have a free shortage but no inactive
shortage.

Page_launder() then needs to be changed a bit too. If it's
called from a user process, it can have the same behaviour
it has now.

OTOH, when it's called from kswapd, kswapd should just flush
as many pages as needed to make the free shortage go away.

Suppose you have a system with 500 kB free shortage. In the
current setup, kswapd will flush MAX_LAUNDER pages and go to
sleep. An alternative would be to let kswapd not go to sleep
when we have a free shortage, but the result of this would
be that kswapd will keep submitting IO until the first 500 kB
is flushed (also wrong).

Only when kswapd submits 500 kB of dirty pages for IO in the
launder loop, everything will work out (didn't submit too much
IO, get the amount of free pages we need).

>  - refill_inactive() can do
> 
> 	count -= refill_inactive_scan(count);

Have you been reading the (partly messy) big snapshot patch
I posted on january 15th? ;)


cheers,

Rik
--
Virtual memory is like a game you can't win;
However, without VM there's truly nothing to lose...

		http://www.surriel.com/
http://www.conectiva.com/	http://distro.conectiva.com.br/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Linux-2.4.1-pre11

[Rik van Riel]

On Mon, 29 Jan 2001, Marcelo Tosatti wrote:

> Btw, look at this part of code from kswapd: 
> 
>                  * 1) we need no more free pages   or
>                  * 2) the inactive pages need to be flushed to disk,
>                  *    it wouldn't help to eat CPU time now ...

>                 if (!free_shortage() || !inactive_shortage()) {
>                         interruptible_sleep_on_timeout(&kswapd_wait, HZ);
> 
> kswapd goes to sleep if there is no free shortage, even if the
> inactive list is under shortage.
> 
> Why not refill the inactive list when the inactive list is under
> shortage? :)

At this point, we already scanned the active list and we
know we're not getting any more aging information at this
point.

In this case, it might be better to leave the active pages
alone for a while and give userland a chance to use the
pages so we can get some aging information the next time we
scan the list.

regards,

Rik
--
Virtual memory is like a game you can't win;
However, without VM there's truly nothing to lose...

		http://www.surriel.com/
http://www.conectiva.com/	http://distro.conectiva.com.br/



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Linux-2.4.1-pre11

[Harold Oga]

On Mon, Jan 29, 2001 at 11:55:44AM +0100, Ondrej Sury wrote:
>Harold Oga <ogah@home.com> writes:
>
>> Hi,
>>    I'm seeing similar problems with my system on 2.4.1-pre10.  This is an
>> AMD Thunderbird 900, MSI K7T Pro2-A mobo w/VIA KT133 chipset, UP, ide/scsi
>> mix.  2.4.1-pre10 works fine if I don't configure ACPI.  I'll try to
>> narrow down when this problem started showing up later today, as I
>> initially moved from 2.4.1-pre3 straight to 2.4.1-pre10.
>
>It's something between pre9 and pre10, and probably it's VIA chipset
>problem.
Hi,
   Ok, it appears that I have 2.4.1-pre10 working properly again.  Looks
like the changes Andy made to the acpi_idle stuff was the problem.  I made
the change Andy suggested on the acpi list, namely commenting out the line
"pm_idle = acpi_idle;" at the bottom of /usr/src/linux/drivers/acpi/cpu.c,
which seems to fix the problem.

This patch makes it clear what I did:
--- linux/drivers/acpi/cpu.c.orig       Mon Jan 29 15:19:21 2001
+++ linux/drivers/acpi/cpu.c    Mon Jan 29 15:22:14 2001
@@ -329,12 +329,5 @@
        acpi_pm_timer_init();


-#ifdef CONFIG_SMP
-       if (smp_num_cpus == 1)
-               pm_idle = acpi_idle;
-#else
-       pm_idle = acpi_idle;
-#endif
-
        return 0;
 }

-Harold  
-- 
"Life sucks, deal with it!"
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/