Linux 2.4.23-rc1

Linux 2.4.23-rc1

[Marcelo Tosatti]

Hi, 

Here goes -rc1.

It contains network driver fixes (b44, tg3, 8139cp), several x86-64
bugfixes, amongst others.


Please help testing!


Summary of changes from v2.4.23-pre9 to v2.4.23-rc1
============================================

<marcelo:logos.cnet>:
  o MAINTAINERS update for HP
  o Backport 2.6 Linus fix for minix corruption problem noted by Konstantin Boldyshev
  o Changed EXTRAVERSION to -rc1

<philipc:snapgear.com>:
  o [netdrvr 8139cp] Fix NAPI race

<pp:ee.oulu.fi>:
  o [netdrvr b44] Fix irq enable/disable; fix oops due to lack of SET_NETDEV_DEV() call

<xose:wanadoo.es>:
  o 2.4.23-pre9 fix for kbuild - hotplug_acpi

Adrian Bunk:
  o fix SOUND_CMPCI Configure.help entry

Andi Kleen:
  o x86-64 update
  o K8 AGP driver updates
  o Make new driver i386 only
  o Fix Documentation.help for K8 AGP driver
  o Add missing nforce3s pci-id
  o Fix TSS limit on x86-64

Andrew Morton:
  o Restore /proc/pid/maps formatting

Dave Kleikamp:
  o JFS: Fix race between link() and unlink()
  o JFS: i_nlink should be checked while holding commit_sem

David S. Miller:
  o [TG3]: Fix bugs in ETHTOOL_SSET introduced by ethtool_ops conversion
  o [TG3]: Bump driver version and release date

David Woodhouse:
  o ilookup() for 2.4
  o JFFS2 garbage collect race fix

Douglas Gilbert:
  o Do not accept negative size's in SG_SET_RESERVED_SIZE

Eric Brower:
  o [SPARC]: Fix _IOC_SIZE() macro when direction is _IOC_NONE

Herbert Xu:
  o Fix BUS_ISA name conflict

Jan Kara:
  o Fix quota accounting bug

John Stultz:
  o Fix x440+ACPI problem
  o Fix cyclone timer (x44x)

Marcel Holtmann:
  o Make firmware loading work builtin

Ralf Bächle:
  o [netdrvr pcnet32] add missing pci_dma_sync_single

Scott Feldman:
  o [e100] sync with 2.6 updates
  o [e1000] sync with 2.6 updates

Stelian Pop:
  o meye driver update

Re: Linux 2.4.23-rc1

[J.A. Magallon]

On 11.10, Marcelo Tosatti wrote:
> 
> Hi, 
> 
> Here goes -rc1.
> 
> It contains network driver fixes (b44, tg3, 8139cp), several x86-64
> bugfixes, amongst others.
> 
> 
> Please help testing!
> 

Sorry to send this so late, but it is trivial.
This patch kills a redundant (and diferent from include/linux/kernel.h)
prototype for printk.

--- linux/include/asm-i386/spinlock.h.orig    2002-10-15 10:12:25.000000000 +0100
+++ linux/include/asm-i386/spinlock.h 2002-10-15 10:12:35.000000000 +0100
@@ -6,9 +6,6 @@
 #include <asm/page.h>
 #include <linux/config.h>
 
-extern int printk(const char * fmt, ...)
-	__attribute__ ((format (printf, 1, 2)));
-
 /* It seems that people are forgetting to
  * initialize their spinlocks properly, tsk tsk.
  * Remember to turn this off in 2.4. -ben

(it is asmlinkage instead of extern in kernel.h)

-- 
J.A. Magallon <jamagallon()able!es>     \                 Software is like sex:
werewolf!able!es                         \           It's better when it's free
Mandrake Linux release 10.0 (Cooker) for i586
Linux 2.4.23-pre9-jam1 (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-4mdk))

Re: Linux 2.4.23-rc1

[Willy Tarreau]

Hi Marcelo,

On Mon, Nov 10, 2003 at 05:28:14PM -0200, Marcelo Tosatti wrote:
> Here goes -rc1.
> 
> It contains network driver fixes (b44, tg3, 8139cp), several x86-64
> bugfixes, amongst others.

Interestingly, tg3, which had a slight tendency to hang on 2.4.22 when duplex
mismatched seems rock solid now. I could even force renegociation and duplex
mismatches during heavy loads (pktgen) without even a warning in the logs.

No pb so far on a test server (DL320 P4/CSB6 IDE/TG3). Good work !

Cheers,
Willy

Re: Linux 2.4.23-rc1

[Kaj-Michael Lang]

> Hi,
>
> Here goes -rc1.
>
> It contains network driver fixes (b44, tg3, 8139cp), several x86-64
> bugfixes, amongst others.

Compiling for Alpha fails with:
...
gcc -D__KERNEL__ -I/work/collection/talinux/kernel/kernel24/tmp/alpha/linux-
2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-alia
sing -fno-common -fomit-frame-pointer -pipe -mno-fp-regs -ffixed-8 -mcpu=ev5
 -Wa,-mev6   -nostdinc -iwithprefix
nclude -DKBUILD_BASENAME=agpgart_be  -DEXPORT_SYMTAB -c agpgart_be.c
agpgart_be.c:52: asm/msr.h: No such file or directory
agpgart_be.c:493: warning: `agp_generic_create_gatt_table' defined but not
used
agpgart_be.c:627: warning: `agp_generic_free_gatt_table' defined but not
used
...

Same config works fine for 2.4.22

-- 
Kaj-Michael Lang , milang@tal.org

Re: Linux 2.4.23-rc1

[Kaj-Michael Lang]

> > Hi,
> >
> > Here goes -rc1.
> >
> > It contains network driver fixes (b44, tg3, 8139cp), several x86-64
> > bugfixes, amongst others.
>

And on sparc64 I get:
...
sparc64-linux-gcc -D__KERNEL__ -I/work/collection/talinux/kernel/kernel24/tm
p/sparc64/linux-2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 
-fno-strict-aliasing -fno-common -fomit-frame-pointer -m64 -pipe -mno-fpu -m
cpu=ultrasparc -mcmodel=medlow -ffixed-g4 -fcall-used-g5 -fcall-used-g7 -Wno
-sign-compare -Wa,--undeclared-regs -DMODULE  -nostdinc -iwithprefix
include -DKBUILD_BASENAME=ir_usb  -c -o ir-usb.o ir-usb.c
ir-usb.c: In function `ir_set_termios':
ir-usb.c:571: `B4000000' undeclared (first use in this function)
ir-usb.c:571: (Each undeclared identifier is reported only once
ir-usb.c:571: for each function it appears in.)
make[6]: *** [ir-usb.o] Error 1
make[5]: *** [_modsubdir_serial] Error 2
make[4]: *** [_modsubdir_usb] Error 2
make[3]: *** [_mod_drivers] Error 2
..

-- 
Kaj-Michael Lang , milang@tal.org

Re: Linux 2.4.23-rc1

[David S. Miller]

On Wed, 12 Nov 2003 00:18:10 +0200
"Kaj-Michael Lang" <milang@tal.org> wrote:

> And on sparc64 I get:
> ...
> sparc64-linux-gcc -D__KERNEL__ -I/work/collection/talinux/kernel/kernel24/tm
> p/sparc64/linux-2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 
> -fno-strict-aliasing -fno-common -fomit-frame-pointer -m64 -pipe -mno-fpu -m
> cpu=ultrasparc -mcmodel=medlow -ffixed-g4 -fcall-used-g5 -fcall-used-g7 -Wno
> -sign-compare -Wa,--undeclared-regs -DMODULE  -nostdinc -iwithprefix
> include -DKBUILD_BASENAME=ir_usb  -c -o ir-usb.o ir-usb.c
> ir-usb.c: In function `ir_set_termios':
> ir-usb.c:571: `B4000000' undeclared (first use in this function)
> ir-usb.c:571: (Each undeclared identifier is reported only once
> ir-usb.c:571: for each function it appears in.)

ir-usb.c should not assume all baud speed rate macros are
available on all platforms.

Re: Linux 2.4.23-rc1

[David S. Miller]

On Tue, 11 Nov 2003 23:10:27 +0000
viro@parcelfarce.linux.theplanet.co.uk wrote:

> > ir-usb.c should not assume all baud speed rate macros are
> > available on all platforms.
> 
> Umm...   Why are they defined in asm-*/* and not in linux/*?

Because each platform may wish to use different values
for these things as they are part of the user visible ABI.

For example, on Sparc we choose values which matched up closely
to what SunOS uses in order to make the compatability layer
for SunOS binaries easier to implement.

It just so happens that the bits are layed out on Sparc such
that we could only fit so many baud rate encodings, we didn't
have enough left to define one for B4000000.

You will notice that in the entire 2.6.x tree, ir-usb.c is the
only piece of code which makes reference to this value.

Re: Linux 2.4.23-rc1

[viro]

On Tue, Nov 11, 2003 at 02:57:34PM -0800, David S. Miller wrote:
> On Wed, 12 Nov 2003 00:18:10 +0200
> "Kaj-Michael Lang" <milang@tal.org> wrote:
> 
> > And on sparc64 I get:
> > ...
> > sparc64-linux-gcc -D__KERNEL__ -I/work/collection/talinux/kernel/kernel24/tm
> > p/sparc64/linux-2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 
> > -fno-strict-aliasing -fno-common -fomit-frame-pointer -m64 -pipe -mno-fpu -m
> > cpu=ultrasparc -mcmodel=medlow -ffixed-g4 -fcall-used-g5 -fcall-used-g7 -Wno
> > -sign-compare -Wa,--undeclared-regs -DMODULE  -nostdinc -iwithprefix
> > include -DKBUILD_BASENAME=ir_usb  -c -o ir-usb.o ir-usb.c
> > ir-usb.c: In function `ir_set_termios':
> > ir-usb.c:571: `B4000000' undeclared (first use in this function)
> > ir-usb.c:571: (Each undeclared identifier is reported only once
> > ir-usb.c:571: for each function it appears in.)
> 
> ir-usb.c should not assume all baud speed rate macros are
> available on all platforms.

Umm...   Why are they defined in asm-*/* and not in linux/*?

Re: Linux 2.4.23-rc1

[Greg KH]

On Tue, Nov 11, 2003 at 03:08:15PM -0800, David S. Miller wrote:
> 
> It just so happens that the bits are layed out on Sparc such
> that we could only fit so many baud rate encodings, we didn't
> have enough left to define one for B4000000.
> 
> You will notice that in the entire 2.6.x tree, ir-usb.c is the
> only piece of code which makes reference to this value.

Then how is the ir-usb driver supposed to be able to set a baud rate of
4000000 in a portable manner?

thanks,

greg k-h

Re: Linux 2.4.23-rc1

[Willy Tarreau]

Hi,

for me, -rc1 compiles correctly on Alpha, but I don't use agpgart. So I
guess it's about your only problem here.

Regards,
Willy

On Tue, Nov 11, 2003 at 11:27:59PM +0200, Kaj-Michael Lang wrote:
> > Hi,
> >
> > Here goes -rc1.
> >
> > It contains network driver fixes (b44, tg3, 8139cp), several x86-64
> > bugfixes, amongst others.
> 
> Compiling for Alpha fails with:
> ...
> gcc -D__KERNEL__ -I/work/collection/talinux/kernel/kernel24/tmp/alpha/linux-
> 2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-alia
> sing -fno-common -fomit-frame-pointer -pipe -mno-fp-regs -ffixed-8 -mcpu=ev5
>  -Wa,-mev6   -nostdinc -iwithprefix
> nclude -DKBUILD_BASENAME=agpgart_be  -DEXPORT_SYMTAB -c agpgart_be.c
> agpgart_be.c:52: asm/msr.h: No such file or directory
> agpgart_be.c:493: warning: `agp_generic_create_gatt_table' defined but not
> used
> agpgart_be.c:627: warning: `agp_generic_free_gatt_table' defined but not
> used
> ...
> 
> Same config works fine for 2.4.22
> 
> -- 
> Kaj-Michael Lang , milang@tal.org
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Linux 2.4.23-rc1

[David S. Miller]

On Tue, 11 Nov 2003 17:54:33 -0800
Greg KH <greg@kroah.com> wrote:

> Then how is the ir-usb driver supposed to be able to set a baud rate of
> 4000000 in a portable manner?

You just won't get support for it on Sparc, nor any other
platform that does not have the B4000000 macro defined.
Ie.

#ifdef B4000000
		case B4000000:
			....
#endif

If, some day, I figure out a clever way to encode these higher
speeds on Sparc and other platforms figure out how to do so as
well, we can remove the ifdef.

For now, the ifdef allows the thing to build properly on all
platforms.

Re: Linux 2.4.23-rc1

[glee]

[-- Attachment #1: Type: text/plain, Size: 275 bytes --]

On Wed, Nov 12, 2003 at 07:19:09AM +0100, Willy Tarreau wrote:
> Hi,
> 
> for me, -rc1 compiles correctly on Alpha, but I don't use agpgart. So I
> guess it's about your only problem here.
> 


I think that we should wrap the msr.h include around a CONFIG_X86_MSR.

	- g.
	


[-- Attachment #2: agpgart.diff --]
[-- Type: text/plain, Size: 367 bytes --]

--- linux-2.4.22/drivers/char/agp/agpgart_be.c.orig	2003-11-12 14:16:40.000000000 +0800
+++ linux-2.4.22/drivers/char/agp/agpgart_be.c	2003-11-12 14:17:54.000000000 +0800
@@ -49,7 +49,10 @@
 #include <asm/uaccess.h>
 #include <asm/io.h>
 #include <asm/page.h>
+
+#ifdef CONFIG_X86_MSR
 #include <asm/msr.h>
+#endif
 
 #include <linux/agp_backend.h>
 #include "agp.h"

Re: Linux 2.4.23-rc1

[Dave Jones]

On Wed, Nov 12, 2003 at 02:49:42PM +0800, glee@gnupilgrims.org wrote:
 > 
 > I think that we should wrap the msr.h include around a CONFIG_X86_MSR.

That config option is for the x86 MSR driver. You probably want to be
ifdefing CONFIG_X86

		Dave

Re: Linux 2.4.23-rc1

[Willy Tarreau]

On Wed, Nov 12, 2003 at 02:49:42PM +0800, glee@gnupilgrims.org wrote:
> On Wed, Nov 12, 2003 at 07:19:09AM +0100, Willy Tarreau wrote:
> > Hi,
> > 
> > for me, -rc1 compiles correctly on Alpha, but I don't use agpgart. So I
> > guess it's about your only problem here.
> > 
> 
> 
> I think that we should wrap the msr.h include around a CONFIG_X86_MSR.

Or simply remove it ? it doesn't seem to me that it's used anywhere in this
file. Could anybody try this patch ?

Regards,
Willy

diff -urN linux-2.4.23-rc1/drivers/char/agp/agpgart_be.c linux-2.4.23-rc1-nomsr/drivers/char/agp/agpgart_be.c
--- linux-2.4.23-rc1/drivers/char/agp/agpgart_be.c	Mon Nov 10 22:24:58 2003
+++ linux-2.4.23-rc1-nomsr/drivers/char/agp/agpgart_be.c	Wed Nov 12 10:08:34 2003
@@ -49,7 +49,6 @@
 #include <asm/uaccess.h>
 #include <asm/io.h>
 #include <asm/page.h>
-#include <asm/msr.h>
 
 #include <linux/agp_backend.h>
 #include "agp.h"

Re: Linux 2.4.23-rc1

[glee]

On Wed, Nov 12, 2003 at 10:10:01AM +0100, Willy Tarreau wrote:
> On Wed, Nov 12, 2003 at 02:49:42PM +0800, glee@gnupilgrims.org wrote:
> > On Wed, Nov 12, 2003 at 07:19:09AM +0100, Willy Tarreau wrote:
> > > Hi,
> > > 
> > > for me, -rc1 compiles correctly on Alpha, but I don't use agpgart. So I
> > > guess it's about your only problem here.
> > > 
> > 
> > 
> > I think that we should wrap the msr.h include around a CONFIG_X86_MSR.
> 
> Or simply remove it ? it doesn't seem to me that it's used anywhere in this
> file. Could anybody try this patch ?
> 


Hi Willy,

I think your analysis is correct.  The stuff in msr.h was used back in
2.4.22 for the nvidia stuff, now it's been removed and should 
no longer be required.

	- g.

Re: Linux 2.4.23-rc1

[Greg KH]

On Tue, Nov 11, 2003 at 10:36:30PM -0800, David S. Miller wrote:
> On Tue, 11 Nov 2003 17:54:33 -0800
> Greg KH <greg@kroah.com> wrote:
> 
> > Then how is the ir-usb driver supposed to be able to set a baud rate of
> > 4000000 in a portable manner?
> 
> You just won't get support for it on Sparc, nor any other
> platform that does not have the B4000000 macro defined.
> Ie.
> 
> #ifdef B4000000
> 		case B4000000:
> 			....
> #endif

I've finally made this change, and will send it onward in a few days.

thanks,

greg k-h