Linux 2.5.7-dj3

Linux 2.5.7-dj3

[Dave Jones]

Mostly just cleaning up a lot of stuff that was discovered
whilst syncing with Linus.  More to come.. 1MB down, 6 to go..

As usual,..
Patch against 2.5.7 vanilla is available from:
ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.5/

Check http://www.codemonkey.org.uk/Linux-2.5.html before reporting
known bugs that are also in mainline.

 -- Davej.

2.5.7-dj3
o   Merge 2.5.8-pre1.
o   Fix some bogus changes found whilst syncing.	(Me)
o   Only offer thermal interrupt monitor on P4.		(Me)
o   Only offer bluesmoke background checker on K7.	(Me)


2.5.7-dj2
o   Merge bits of 2.4.19pre3 & pre4.
    | Dropped quite a bit, 2.5 is really starting to diverge
    | in quite a few areas.
o   Numerous compile fixups.				(Me)
o   Various config.help updates.			(Steven Cole, ESR)
o   AFFS compile fix.					(Jarno Paananen)
o   Update bttv to use new v4l API.			(Jarno Paananen)
o   Removal extraneous BUG() from accounting code.	(Bob Miller)
o   Fix up JFS txnmgr race.				(Christoph Hellwig)
o   Update documentation index.				(Rolf Eike Beer)
o   Use seq_file for /proc/partitions.			(Randy Dunlap)
o   Fix initrd kdev_same thinko.			(Al Viro)
o   Minor sg update.					(Doug Gilbert)
o   install .config with 'make install'			(Kelly French) 
o   Add yet another VAIO to the DMI APM blacklist.	(Michael Piotrowski)
o   Detect get_block errors in block_read_full_page.	(Anton Altaparmakov)
o   Kill off noquot.c					(Al Viro)
o   Special case P4/Xeon bluesmoke init.		(Jon Hourd)
o   Fix up ide-scsi documentation typo.			(Bill Jonas)
o   Increment tcpPassiveOpens counter.			(Dave Miller)
o   Fix up some of Linus' criticisms of SEM_UNDO	(Dave Olien)
o   ibmphp compile fix.					(Greg KH)
o   Fix up two possible panics in microcode driver.	(Tigran Aivazian)
o   fcntl return code POSIX compliance.			(Christopher Yeoh)
o   Lessen debug info from ide-tape.			(Alfredo Sanjuán)
o   Various ide fixes.					(Andre Hedrick)
    | pdc4030 bent into shape a little by me, please test carefully.
o   EasyDisk USB storage support.			(Stanislav Karchebny)
o   Flush cache/TLB on MCE.				(Me)
o   Fix double free in efi partition handling.		(Takanori Kawano)
o   Correct pci_set_mwi documentation typo.		(Geert Uytterhoeven)
o   Fix up race in indydog driver.			(Dave Hansen)
o   3ware driver update.				(Adam Radford)
o   Fix mmap bug with drivers that adjust vm_start.	(Benjamin LaHaise)
o   Register mad16 gameport with input subsystem.	(Michael Haardt)
o   Remove stale comment.				(William Lee Irwin III)
o   P4/Xeon Thermal LVT support for MCE.		(Zwane Mwaikambo)
o   Various fbdev compilation fixes.			(James Simmons)
o   Move touchscreen drivers to their own dir.		(James Simmons)
o   Seperate reiserfs_sb_info from struct super_block	(Brian Gerst)


2.5.7-dj1
o   Various x86-64 compile fixes.			(Me, Andi Kleen)
o   Fix up hfs superblock cleanup typo.			(Brian Gerst)
o   Remove 2.1.x/2.3.x checks from rocket driver.	(William Stinson)
o   JFS compile fixes.					(Dave Kleikamp)
o   pci_alloc_consistent ZONE_NORMAL fix.		(Badari Pulavarty)
o   v4l unregister with open file handles fix.		(Gerd Knorr)
o   ACPI throttling fixes.				(Dominik Brodowski)
o   Updated NFS root_server_path patch.			(Martin Schwenke)
o   Backout problematic sym53c8xx_2 changes.		(Douglas Gilbert)
o   Fix up one of my MCE SNAFU's.			(Douglas Gilbert)
    | Also make it AMD only for now, pending further investigation.
o   Support for Belkin Wireless card.			(Brendan W. McAdams)
o   input/gameport makefile & config.in fixes.		(Steven Cole)
o   UDF write compile fix.


2.5.6-dj2
o   machine check background checker SMP improvements.	(Me)
o   Hush INVALIDATE_AGID noise on Toshiba DVD drives.	(Jens Axboe)
o   Remove delayed vfree() in ppp_deflate.		(David Woodhouse)
o   Fix devfs 64-bit portability issues.		(Carsten Otte)
o   UFS blocksize fix.					(Zwane Mwaikambo)
o   Fix proc race on task_struct->sig.			(Chris Mason)
o   Pull NFS server IP from root_server_path.		(Martin Schwenke)
o   Fix broken tristate in ALSA config.			(Steven Cole)
o   Work around mpc_apicid > MAX_APICS problem.		(James Cleverdon)
o   cs4281 OSS compile fix.				(Sebastian Droege)
o   Fix locking in ATM device allocation.		(Maksim Krasnyanskiy)
o   Fix MSDOS filesystem option mistreatment.		(Rene Scharfe)
o   Increase number of JFS transaction locks.		(Steve Best)
o   page_to_phys() fix for >4GB pages on x86.		(Badari Pulavarty)
o   Append kernel config to zImage.			(Randy Dunlap)
    | See also scripts/extract-ikconfig
o   Remove useless check from shmem_link()		(Alexander Viro)
o   Enable Natsemi MMX extensions.			(Zwane Mwaikambo)
o   Superblock cleanups for isofs,udf & bfs.		(Brian Gerst)
o   request_region() checking for hisax avm_a1		(Dan D Carpenter)
o   uninitialise lots of static vars in net/		(William Stinson)
o   D-Link DE620 janitor work.				(Karol Kasprzak)
o   Netfilter conntrack compile fix.			(Christopher Barton)


-- 
Dave Jones.                    http://www.codemonkey.org.uk
SuSE Labs.

Re: Linux 2.5.7-dj3

[Adrian Bunk]

Hi Dave,

I got the following compile error:

<--  snip  -->

...
gcc -D__KERNEL__ -I/home/bunk/linux/kernel-2.5/linux-2.5.7/include -Wall
-Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer
-fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2
-march=k6   -DKBUILD_BASENAME=pdc4030  -c -o pdc4030.o pdc4030.c
pdc4030.c: In function `promise_multwrite':
pdc4030.c:447: warning: passing arg 2 of `bio_kmap_irq' makes pointer from
integer without a cast
pdc4030.c: In function `promise_rw_disk':
pdc4030.c:664: structure has no member named `channel'
make[3]: *** [pdc4030.o] Error 1
make[3]: Leaving directory
`/home/bunk/linux/kernel-2.5/linux-2.5.7/drivers/ide'
make[2]: *** [first_rule] Error 2
make[2]: Leaving directory
`/home/bunk/linux/kernel-2.5/linux-2.5.7/drivers/ide'
make[1]: *** [_subdir_ide] Error 2
make[1]: Leaving directory
`/home/bunk/linux/kernel-2.5/linux-2.5.7/drivers'
make: *** [_dir_drivers] Error 2

<--  snip  -->


The code that failed is the following part of -dj3 that s not in
2.5.8-pre1:


--- linux-2.5.7/drivers/ide/pdc4030.c	Mon Mar 18 20:37:08 2002
+++ linux-2.5/drivers/ide/pdc4030.c	Thu Apr  4 04:19:50 2002
@@ -645,6 +656,13 @@

 	memset(&taskfile, 0, sizeof(struct hd_drive_task_hdr));

+	/* The four drives on the two logical (one physical) interfaces
+	   are distinguished by writing the drive number (0-3) to the
+	   Feature register.
+	   FIXME: Is promise_selectproc now redundant??
+	*/
+	taskfile.feature	= (HWIF(drive)->channel << 1) + drive->select.b.unit;
+
 	taskfile.sector_count	= rq->nr_sectors;
 	taskfile.sector_number	= block;
 	taskfile.low_cylinder	= (block>>=8);


cu
Adrian

Re: Linux 2.5.7-dj3

[Dave Jones]

On Thu, Apr 04, 2002 at 09:50:32AM +0200, Adrian Bunk wrote:
 > pdc4030.c: In function `promise_multwrite':
 > pdc4030.c:447: warning: passing arg 2 of `bio_kmap_irq' makes pointer from
 > integer without a cast
 > pdc4030.c: In function `promise_rw_disk':
 > pdc4030.c:664: structure has no member named `channel'

Ok, I'm confused.
This is a compile failure from 2.5.8-pre1.
The line numbers don't even match up to whats in -dj3

Forgot to apply -dj3 patch ?

    Dave.

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Linux 2.5.7-dj3

[Dave Jones]

On Thu, Apr 04, 2002 at 02:16:47PM +0200, Dave Jones wrote:
 > On Thu, Apr 04, 2002 at 09:50:32AM +0200, Adrian Bunk wrote:
 >  > pdc4030.c: In function `promise_multwrite':
 >  > pdc4030.c:447: warning: passing arg 2 of `bio_kmap_irq' makes pointer from
 >  > integer without a cast
 >  > pdc4030.c: In function `promise_rw_disk':
 >  > pdc4030.c:664: structure has no member named `channel'
 > 
 > Ok, I'm confused.
 > This is a compile failure from 2.5.8-pre1.
 > The line numbers don't even match up to whats in -dj3


My bad. _I_ was looking at a wrong tree.
I'll fix these bits up later and push them Martins way, as there
are a few other small IDE bits that have been lingering in my
tree since 2.5.2 days.

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Linux 2.5.7-dj3

[Martin Dalecki]

Dave Jones wrote:
> On Thu, Apr 04, 2002 at 02:16:47PM +0200, Dave Jones wrote:
>  > On Thu, Apr 04, 2002 at 09:50:32AM +0200, Adrian Bunk wrote:
>  >  > pdc4030.c: In function `promise_multwrite':
>  >  > pdc4030.c:447: warning: passing arg 2 of `bio_kmap_irq' makes pointer from
>  >  > integer without a cast
>  >  > pdc4030.c: In function `promise_rw_disk':
>  >  > pdc4030.c:664: structure has no member named `channel'
>  > 
>  > Ok, I'm confused.
>  > This is a compile failure from 2.5.8-pre1.
>  > The line numbers don't even match up to whats in -dj3
> 
> 
> My bad. _I_ was looking at a wrong tree.
> I'll fix these bits up later and push them Martins way, as there
> are a few other small IDE bits that have been lingering in my
> tree since 2.5.2 days.

Well actually I was short before having a look in to the -dj3 series,
but first I have to read through 2.5.8-pre1.
However I would be really gald if you could throw the code in question
in fragments called patches at me ;-).

Re: Linux 2.5.7-dj3 - BUG & PATCH

[Sebastian Droege]

[-- Attachment #1: Type: text/plain, Size: 3150 bytes --]

Hi,
I have a problem in 2.5.7-dj3 which doesn't exist in 2.5.8-pre1...
My USB keyboard and mouse are detected properly but aren't usable
I get the same behaviour when unsetting CONFIG_USB_HIDINPUT in 2.5.8-pre1
grepping for CONFIG_USB_HIDINPUT in 2.5.7-dj3 finds something but the option doesn't show in old/menuconfig
When setting CONFIG_USB_HIDINPUT=y by hand in 2.5.7-dj3 I get a compile error:

make[3]: Entering directory `/usr/src/linux-2.5.7/drivers/usb'
gcc -D__KERNEL__ -I/usr/src/linux-2.5.7/include -Wall -Wstrict-prototypes -Wno-trigraphs -O6 -fomit-frame-pointer -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686   -DKBUILD_BASENAME=hid_input  -c -o hid-input.o hid-input.c
hid-input.c:335: redefinition of `hidinput_hid_event'
hid.h:411: `hidinput_hid_event' previously defined here
hid-input.c:413: redefinition of `hidinput_connect'
hid.h:412: `hidinput_connect' previously defined here
hid-input.c:458: redefinition of `hidinput_disconnect'
hid.h:413: `hidinput_disconnect' previously defined here
make[3]: *** [hid-input.o] Fehler 1

I don't see any differences between 2.5.7-dj3 and 2.5.8-pre1 which can cause such error but the patch at the bottom solves it ;)
Maybe someone can explain me why 2.5.8-pre1 compiles without the #ifdefs  (with CONFIG_USB_HIDINPUT set and unset) and not in 2.5.7-dj3

With this patch my mouse and keyboard work again

Bye

diff -Nur linux-2.5.7/drivers/usb/Config.in linux-2.5.7-2/drivers/usb/Config.in
--- linux-2.5.7/drivers/usb/Config.in   Thu Apr  4 17:17:57 2002
+++ linux-2.5.7-2/drivers/usb/Config.in Thu Apr  4 17:15:38 2002
@@ -53,6 +53,7 @@
       comment '  Input core support is needed for USB HID'
    else
       dep_tristate '  USB Human Interface Device (full HID) support' CONFIG_USB_HID $CONFIG_USB $CONFIG_INPUT
+        dep_mbool '    HID input layer support' CONFIG_USB_HIDINPUT $CONFIG_INPUT $CONFIG_USB_HID
          dep_mbool '    /dev/hiddev raw HID device support (EXPERIMENTAL)' CONFIG_USB_HIDDEV $CONFIG_USB_HID
       if [ "$CONFIG_USB_HID" != "y" ]; then
          dep_tristate '  USB HIDBP Keyboard (basic) support' CONFIG_USB_KBD $CONFIG_USB $CONFIG_INPUT
diff -Nur linux-2.5.7/drivers/usb/hid-input.c linux-2.5.7-2/drivers/usb/hid-input.c
--- linux-2.5.7/drivers/usb/hid-input.c Mon Mar 18 21:37:13 2002
+++ linux-2.5.7-2/drivers/usb/hid-input.c       Thu Apr  4 17:20:55 2002
@@ -331,6 +331,7 @@
        }
 }
 
+#ifdef CONFIG_USB_HIDINPUT
 void hidinput_hid_event(struct hid_device *hid, struct hid_field *field, struct hid_usage *usage, __s32 value)
 {
        struct input_dev *input = &hid->input;
@@ -373,6 +374,7 @@
        if ((field->flags & HID_MAIN_ITEM_RELATIVE) && (usage->type == EV_KEY))
                input_event(input, usage->type, usage->code, 0);
 }
+#endif
 
 static int hidinput_input_event(struct input_dev *dev, unsigned int type, unsigned int code, int value)
 {
@@ -403,6 +405,7 @@
        hid_close(hid);
 }
 
+#ifdef CONFIG_USB_HIDINPUT
 /*
  * Register the input device; print a message.
  * Configure the input layer interface
@@ -458,3 +461,4 @@
 {
        input_unregister_device(&hid->input);
 }
+#endif

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Linux 2.5.7-dj3 - BUG & PATCH

[Greg KH]

On Thu, Apr 04, 2002 at 05:22:38PM +0200, Sebastian Droege wrote:
> 
> I don't see any differences between 2.5.7-dj3 and 2.5.8-pre1 which can
> cause such error but the patch at the bottom solves it ;) Maybe
> someone can explain me why 2.5.8-pre1 compiles without the #ifdefs
> (with CONFIG_USB_HIDINPUT set and unset) and not in 2.5.7-dj3
> 
> With this patch my mouse and keyboard work again

Wrong patch, do not apply.  Well, half of the patch is wrong.  The
Config.in part is correct.  I'm guessing that the changes made in
2.4.8-pre1 to include/hiddev.h didn't make it into -dj3

Try seeing if adding that patch to the -dj tree fixes things for you.
If you still have problems, let me know and I'll try to figure it out.

thanks,

greg k-h

Re: Linux 2.5.7-dj3 - BUG & PATCH

[Dave Jones]

On Thu, Apr 04, 2002 at 09:54:38AM -0800, Greg KH wrote:

 > Wrong patch, do not apply.  Well, half of the patch is wrong.  The
 > Config.in part is correct.  I'm guessing that the changes made in
 > 2.4.8-pre1 to include/hiddev.h didn't make it into -dj3
 > 
 > Try seeing if adding that patch to the -dj tree fixes things for you.
 > If you still have problems, let me know and I'll try to figure it out.

Bear in mind that my tree still has Vojtech's input layer changes which
moves quite a bit of stuff around, so it may have something to do with
that. Or it could just be that I pooched the merge.

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Linux 2.5.7-dj3 - BUG & PATCH

[Greg KH]

On Thu, Apr 04, 2002 at 05:22:38PM +0200, Sebastian Droege wrote:
> Hi,
> I have a problem in 2.5.7-dj3 which doesn't exist in 2.5.8-pre1...
> My USB keyboard and mouse are detected properly but aren't usable
> I get the same behaviour when unsetting CONFIG_USB_HIDINPUT in 2.5.8-pre1
> grepping for CONFIG_USB_HIDINPUT in 2.5.7-dj3 finds something but the option doesn't show in old/menuconfig
> When setting CONFIG_USB_HIDINPUT=y by hand in 2.5.7-dj3 I get a compile error:
> 
> make[3]: Entering directory `/usr/src/linux-2.5.7/drivers/usb'
> gcc -D__KERNEL__ -I/usr/src/linux-2.5.7/include -Wall -Wstrict-prototypes -Wno-trigraphs -O6 -fomit-frame-pointer -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686   -DKBUILD_BASENAME=hid_input  -c -o hid-input.o hid-input.c
> hid-input.c:335: redefinition of `hidinput_hid_event'
> hid.h:411: `hidinput_hid_event' previously defined here
> hid-input.c:413: redefinition of `hidinput_connect'
> hid.h:412: `hidinput_connect' previously defined here
> hid-input.c:458: redefinition of `hidinput_disconnect'
> hid.h:413: `hidinput_disconnect' previously defined here
> make[3]: *** [hid-input.o] Fehler 1

I can't duplicate this, can you send me your .config?

thanks,

greg k-h

Re: Linux 2.5.7-dj3 - BUG & PATCH

[Sebastian Droege]

[-- Attachment #1: Type: text/plain, Size: 1498 bytes --]

On Thu, 4 Apr 2002 22:45:45 -0800
Greg KH <greg@kroah.com> wrote:

> On Thu, Apr 04, 2002 at 05:22:38PM +0200, Sebastian Droege wrote:
> > Hi,
> > I have a problem in 2.5.7-dj3 which doesn't exist in 2.5.8-pre1...
> > My USB keyboard and mouse are detected properly but aren't usable
> > I get the same behaviour when unsetting CONFIG_USB_HIDINPUT in 2.5.8-pre1
> > grepping for CONFIG_USB_HIDINPUT in 2.5.7-dj3 finds something but the option doesn't show in old/menuconfig
> > When setting CONFIG_USB_HIDINPUT=y by hand in 2.5.7-dj3 I get a compile error:
> > 
> > make[3]: Entering directory `/usr/src/linux-2.5.7/drivers/usb'
> > gcc -D__KERNEL__ -I/usr/src/linux-2.5.7/include -Wall -Wstrict-prototypes -Wno-trigraphs -O6 -fomit-frame-pointer -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686   -DKBUILD_BASENAME=hid_input  -c -o hid-input.o hid-input.c
> > hid-input.c:335: redefinition of `hidinput_hid_event'
> > hid.h:411: `hidinput_hid_event' previously defined here
> > hid-input.c:413: redefinition of `hidinput_connect'
> > hid.h:412: `hidinput_connect' previously defined here
> > hid-input.c:458: redefinition of `hidinput_disconnect'
> > hid.h:413: `hidinput_disconnect' previously defined here
> > make[3]: *** [hid-input.o] Fehler 1
> 
> I can't duplicate this, can you send me your .config?
Sure but I can't reproduce it either with a clean tree
Strange... but I have done a make mrproper in the old tree and the compile error was still there...

Bye

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]