opensuse:kernel.git
16 years agoadapt zotrix radio driver to pre 2.4.7 videodev redesign
Gerd Knorr [Mon, 8 Apr 2002 12:02:38 +0000 (14:02 +0200)]
adapt zotrix radio driver to pre 2.4.7 videodev redesign

16 years ago[PATCH] loop deadlock fix
Andrew Morton [Fri, 5 Apr 2002 11:46:26 +0000 (03:46 -0800)]
[PATCH] loop deadlock fix

This chunk is present in 2.4, missing in 2.5.  It
stops the loop thread from waiting on its owm buffers.

16 years agoCset exclude: davej@suse.de|ChangeSet|20020404222348|23820
Linus Torvalds [Fri, 5 Apr 2002 11:31:39 +0000 (03:31 -0800)]
Cset exclude: davej@suse.de|ChangeSet|20020404222348|23820

16 years agoClean up do_truncate due notify_change() locking change
Linus Torvalds [Fri, 5 Apr 2002 00:44:44 +0000 (16:44 -0800)]
Clean up do_truncate due notify_change() locking change

16 years agoForce some semblance of workingness onto qla1280 driver
Linus Torvalds [Fri, 5 Apr 2002 00:41:16 +0000 (16:41 -0800)]
Force some semblance of workingness onto qla1280 driver

16 years agoOops, remove remnants of old attribute lock
Linus Torvalds [Fri, 5 Apr 2002 00:40:11 +0000 (16:40 -0800)]
Oops, remove remnants of old attribute lock

16 years agoMerge bk://linuxusb.bkbits.net/dir_move-2.5
Linus Torvalds [Fri, 5 Apr 2002 08:25:11 +0000 (00:25 -0800)]
Merge bk://linuxusb.bkbits.net/dir_move-2.5
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoMake legacy drivers who use "virt_to_bus()" and friends work on x86.
Linus Torvalds [Fri, 5 Apr 2002 08:23:11 +0000 (00:23 -0800)]
Make legacy drivers who use "virt_to_bus()" and friends work on x86.

It's up to other architectures to worry about portability for now.

16 years ago[PATCH] shift BKL out of notify_change
Dave Hansen [Fri, 5 Apr 2002 07:13:11 +0000 (23:13 -0800)]
[PATCH] shift BKL out of notify_change

Moved i_sem down into notify_change() and out of the UMSDOS
function. Moved BKL down from notify_change into filesystems.

16 years ago[PATCH] PATCH 4a or 4 : knfs : typo...
Neil Brown [Fri, 5 Apr 2002 07:11:45 +0000 (23:11 -0800)]
[PATCH] PATCH 4a or 4 : knfs : typo...

typo in that last patch, sorry.

16 years ago[PATCH] PATCH 4 of 4 : knfsd : Increase the max block size for NFS replies.
Neil Brown [Fri, 5 Apr 2002 07:11:36 +0000 (23:11 -0800)]
[PATCH] PATCH 4 of 4 : knfsd : Increase the max block size for NFS replies.

This increases the max read/write size for nfsd from 8K to 32K.

This requires making NFSv2 return the right number in statfs
requests.  NFSv3 already does that.

16 years ago[PATCH] PATCH 3 of 4 : knfsd : Store the fsid in the returned attributes instead...
Neil Brown [Fri, 5 Apr 2002 07:11:29 +0000 (23:11 -0800)]
[PATCH] PATCH 3 of 4 : knfsd : Store the fsid in the returned attributes instead of the device number

When a filesystem is exported with  fsid=  we should use that
fsid instead of the i_dev number when returning NFS attributes,
so that there is no chance of clients that depend on the filesys
id in the attributes getting confused by device numbers changing.

We only do this if the reference filehandle uses fsid to identify
the filesystem, so that a server can be converted from non-fsid= to
using fsid= without confusing active clients.

16 years ago[PATCH] PATCH 2 of 4 : knfsd : Allow exporting of deviceless filesystems if fsid...
Neil Brown [Fri, 5 Apr 2002 07:11:22 +0000 (23:11 -0800)]
[PATCH] PATCH 2 of 4 : knfsd : Allow exporting of deviceless filesystems if fsid= given

Previously we could only export FS_REQUIRES_DEV filesystems
as we need a devno to put in the filehandle.
Now that we have fsid= (NFSEXP_FSID) we don't need a devno
to put in the filehandle so we can relax this requirement.

16 years ago[PATCH] PATCH 1 of 4 : knfsd : Use symbols for size calculation for response sizes.
Neil Brown [Fri, 5 Apr 2002 07:11:15 +0000 (23:11 -0800)]
[PATCH] PATCH 1 of 4 : knfsd : Use symbols for size calculation for response sizes.

Use symbolic names for some common size components in the response
size calculation for the NFSD.  This makes it easier to get the
numbers right and to review them.
This patch also fixes a few number for nfsv3 that were wrong.

16 years ago[PATCH] elevator 'buglet'
Jens Axboe [Fri, 5 Apr 2002 07:09:23 +0000 (23:09 -0800)]
[PATCH] elevator 'buglet'

Lets just kill this check -- it usually only catches drivers queueing
something in front of a started request on their own (such as shoving a
request sense in front of a failed packet command, for instance). So
it's either working around this detection in some drivers, or killing
it. I vote for the latter, patch attached against 2.5.8-pre1 :-)

16 years agoUSB
Greg Kroah-Hartman [Fri, 5 Apr 2002 07:09:11 +0000 (23:09 -0800)]
USB

moved the USB_STORAGE Config.help items into the drivers/usb/storage directory.

16 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.make
Linus Torvalds [Fri, 5 Apr 2002 06:43:55 +0000 (22:43 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.make
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoUSB
Greg Kroah-Hartman [Fri, 5 Apr 2002 03:33:36 +0000 (19:33 -0800)]
USB

added a README file to explain what the different subdirectories are for.

16 years agoDuh. Use "device_lock", not "device_root" for locking.
Linus Torvalds [Fri, 5 Apr 2002 03:12:25 +0000 (19:12 -0800)]
Duh. Use "device_lock", not "device_root" for locking.

16 years agoUSB
Greg Kroah-Hartman [Fri, 5 Apr 2002 02:53:38 +0000 (18:53 -0800)]
USB

moved class/storage/ back to storage/
created input/
orderd the makefiles and config.in menus better.

16 years agoMerge master.kernel.org:/home/mochel/BK/linux-2.5
Linus Torvalds [Fri, 5 Apr 2002 02:48:20 +0000 (18:48 -0800)]
Merge master.kernel.org:/home/mochel/BK/linux-2.5
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoAdd platform driver object
Patrick Mochel [Fri, 5 Apr 2002 02:12:08 +0000 (18:12 -0800)]
Add platform driver object

16 years agocompile fix for drivers/base/sys.c
Patrick Mochel [Fri, 5 Apr 2002 02:08:15 +0000 (18:08 -0800)]
compile fix for drivers/base/sys.c

16 years agoMerge master.kernel.org:/home/davem/BK/sparc-2.5
Linus Torvalds [Fri, 5 Apr 2002 01:23:07 +0000 (17:23 -0800)]
Merge master.kernel.org:/home/davem/BK/sparc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

16 years agoAdd device_shutdown() calls to reboot and power off transitions (and let the user...
Patrick Mochel [Fri, 5 Apr 2002 01:22:45 +0000 (17:22 -0800)]
Add device_shutdown() calls to reboot and power off transitions (and let the user know)

16 years agoMerge master.kernel.org:/home/davem/BK/misc-2.5
Linus Torvalds [Fri, 5 Apr 2002 01:22:05 +0000 (17:22 -0800)]
Merge master.kernel.org:/home/davem/BK/misc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

16 years agoMerge master.kernel.org:/home/davem/BK/net-2.5
Linus Torvalds [Fri, 5 Apr 2002 00:35:14 +0000 (16:35 -0800)]
Merge master.kernel.org:/home/davem/BK/net-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

16 years agoMerge hera.kernel.org:/home/torvalds/BK/linux-2.5
Patrick Mochel [Fri, 5 Apr 2002 00:08:09 +0000 (16:08 -0800)]
Merge hera.kernel.org:/home/torvalds/BK/linux-2.5
into hera.kernel.org:/home/mochel/BK/linux-2.5

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 16:18:18 +0000 (08:18 -0800)]
USB

more file movement cleanups.  Now handles misc drivers compiled into
the kernel corectly.

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 16:13:11 +0000 (08:13 -0800)]
USB

moved the host drivers help to the host directory

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 15:18:16 +0000 (07:18 -0800)]
USB

fixed lib Makefile problem with usb files moving
moved drivers/usb/scanner/ to drivers/usb/image/

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 14:05:05 +0000 (06:05 -0800)]
USB

moved some files from misc to image
cleaned up makefile some more.

16 years agoFix exit_notify() to actually do what the comment
Linus Torvalds [Thu, 4 Apr 2002 12:58:58 +0000 (04:58 -0800)]
Fix exit_notify() to actually do what the comment
says it should do - lock out preemption.

16 years agoMore fixups for tlbflush.h header split
Linus Torvalds [Thu, 4 Apr 2002 12:51:50 +0000 (04:51 -0800)]
More fixups for tlbflush.h header split

16 years agoMake the assembly-level code match the preempt_sched
Linus Torvalds [Thu, 4 Apr 2002 10:37:02 +0000 (02:37 -0800)]
Make the assembly-level code match the preempt_sched
changes

16 years agoScheduler preempt fixes and cleanups
Linus Torvalds [Thu, 4 Apr 2002 09:56:34 +0000 (01:56 -0800)]
Scheduler preempt fixes and cleanups

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 09:38:58 +0000 (01:38 -0800)]
USB

moved lots of the Config.in info into the subdirectories.
fixed up the makefiles to work nicer.

16 years ago[PATCH] preemptive kernel behavior change: don't be rude
Robert Love [Thu, 4 Apr 2002 09:25:22 +0000 (01:25 -0800)]
[PATCH] preemptive kernel behavior change: don't be rude

- do not manually set task->state
- instead, in preempt_schedule, set a flag in preempt_count that
  denotes that this task is entering schedule off a kernel preemption.
- use this flag in schedule to jump to pick_next_task
- in preempt_schedule, upon return from schedule, unset the flag
- have entry.S just call preempt_schedule and not duplicate this work,
  as Linus suggested.  I agree.  Note this makes debugging easier as
  we keep a single point of entry for kernel preemptions.

The result: we can safely preempt non-TASK_RUNNING tasks.  If one is
preempted, we can safely survive schedule because we won't handle the
special casing of non-TASK_RUNNING at the top of schedule.  Thus other
tasks can run as desired and our non-TASK_RUNNING task will eventually
be rescheduled, in its original state, and complete happily.

This is the behavior we have in the 2.4 patches and 2.5 until
~2.5.6-pre.  This works.  It requires no other changes elsewhere (it
actually removes some special-casing Ingo did in the signal code).

16 years agouhhuh. Fix duplicate merge from -dj tree
Linus Torvalds [Thu, 4 Apr 2002 09:02:28 +0000 (01:02 -0800)]
uhhuh. Fix duplicate merge from -dj tree

16 years agoFix tlbflush header file dependencies
Linus Torvalds [Thu, 4 Apr 2002 08:57:16 +0000 (00:57 -0800)]
Fix tlbflush header file dependencies

16 years agoupdate x86 defconfig
Linus Torvalds [Thu, 4 Apr 2002 08:56:43 +0000 (00:56 -0800)]
update x86 defconfig

16 years agousb subsystem now builds as modules.
Greg Kroah-Hartman [Thu, 4 Apr 2002 07:51:09 +0000 (23:51 -0800)]
usb subsystem now builds as modules.
dependancies still seem broken.

16 years ago[PATCH] IS_DEADDIR checks (2.5)
Alexander Viro [Thu, 4 Apr 2002 06:48:10 +0000 (22:48 -0800)]
[PATCH] IS_DEADDIR checks (2.5)

2.4 variant will go to Marcelo in a couple of minutes.

Patch moves IS_DEADDIR() checks into may_delete().

16 years agoUSB
Greg Kroah-Hartman [Thu, 4 Apr 2002 06:47:14 +0000 (22:47 -0800)]
USB

moved files to different subdirectories to make try to make sense
of the current mess, and to allow usb client drivers to integrate into
the tree easier.

16 years ago[PATCH] EFI GUID partition support update.
Dave Jones [Thu, 4 Apr 2002 06:43:36 +0000 (22:43 -0800)]
[PATCH] EFI GUID partition support update.

More bits from Matt Domsch. Fixes GUID printing, and updates
to what's in the IA64 tree. Other cleanups are mentioned in
the changelog in the patch.

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:36:14 +0000 (22:36 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:35:52 +0000 (22:35 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:35:30 +0000 (22:35 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:33:45 +0000 (22:33 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:26:53 +0000 (22:26 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:26:30 +0000 (22:26 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:26:08 +0000 (22:26 -0800)]
Automerge

16 years agoAutomerge
Linus Torvalds [Thu, 4 Apr 2002 06:25:42 +0000 (22:25 -0800)]
Automerge

16 years ago[PATCH] struct super_block cleanup - reiserfs
Dave Jones [Thu, 4 Apr 2002 06:24:12 +0000 (22:24 -0800)]
[PATCH] struct super_block cleanup - reiserfs

Original from: Brian Gerst <bgerst@didntduck.org>
Has had a once over by Chris Mason and Al.

Seperates reiserfs_sb_info from struct super_block.

Brian Gerst

16 years ago[PATCH] named structure initialisers for fs/
Dave Jones [Thu, 4 Apr 2002 06:24:04 +0000 (22:24 -0800)]
[PATCH] named structure initialisers for fs/

Originally by Grant R.Guenther
Has had a quick once over by Al, who weeded out one chunk that was
unrelated.

16 years ago[PATCH] ext3 inode generation improvements.
Dave Jones [Thu, 4 Apr 2002 06:23:53 +0000 (22:23 -0800)]
[PATCH] ext3 inode generation improvements.

Originally from Andrew Morton

16 years ago[PATCH] better dquot accounting
Dave Jones [Thu, 4 Apr 2002 06:23:48 +0000 (22:23 -0800)]
[PATCH] better dquot accounting

16 years ago[PATCH] QNX4fs sync
Dave Jones [Thu, 4 Apr 2002 06:23:42 +0000 (22:23 -0800)]
[PATCH] QNX4fs sync

Brings QNX4FS back in sync with 2.4

16 years ago[PATCH] bss bits for isofs
Dave Jones [Thu, 4 Apr 2002 06:23:37 +0000 (22:23 -0800)]
[PATCH] bss bits for isofs

Originally from the kernel janitor folks

16 years ago[PATCH] MSDOS fs option parser cleanup
Dave Jones [Thu, 4 Apr 2002 06:23:32 +0000 (22:23 -0800)]
[PATCH] MSDOS fs option parser cleanup

Original from Rene Scharfe
This fixes a problem where MSDOS fs's ignore their 'check' mount option.

16 years ago[PATCH] increase number of transaction locks in JFS txnmgr
Dave Jones [Thu, 4 Apr 2002 06:23:27 +0000 (22:23 -0800)]
[PATCH] increase number of transaction locks in JFS txnmgr

Original fix from Andi Kleen

16 years ago[PATCH] Improved allocator for NTFS
Dave Jones [Thu, 4 Apr 2002 06:23:21 +0000 (22:23 -0800)]
[PATCH] Improved allocator for NTFS

Originally by Anton Altaparmakov.
I think Anton is going to submit his rewritten NTFS soon making this null and void,
but in the interim, it fixes a known problem with NTFS and large allocations.

16 years ago[PATCH] list_for_each is fs/
Dave Jones [Thu, 4 Apr 2002 06:23:13 +0000 (22:23 -0800)]
[PATCH] list_for_each is fs/

From the kernel janitor folks

16 years ago[PATCH] cleanup list usage in dquot
Dave Jones [Thu, 4 Apr 2002 06:23:07 +0000 (22:23 -0800)]
[PATCH] cleanup list usage in dquot

From the kernel janitor folks

16 years agoUpdate kernel version
Linus Torvalds [Thu, 4 Apr 2002 06:22:55 +0000 (22:22 -0800)]
Update kernel version

16 years agoUSB visor driver
Greg Kroah-Hartman [Thu, 4 Apr 2002 06:20:24 +0000 (22:20 -0800)]
USB visor driver

Added support for the Sony OS 4.1 devices.  Thanks to Hiroyuki ARAKI
<hiro@zob.ne.jp> for the information.

16 years agoOn sparc64, flush_thread needs to setup the PGD cache
David S. Miller [Thu, 4 Apr 2002 06:07:36 +0000 (22:07 -0800)]
On sparc64, flush_thread needs to setup the PGD cache
for 64-bit apps too.

16 years agoIn linux/skbuff.h, always use unsigned long for flags.
Andi Kleen [Thu, 4 Apr 2002 05:55:16 +0000 (21:55 -0800)]
In linux/skbuff.h, always use unsigned long for flags.

16 years agoMake JFS licence boilerplate uniform, update copyright dates
Dave Kleikamp [Thu, 4 Apr 2002 04:41:58 +0000 (22:41 -0600)]
Make JFS licence boilerplate uniform, update copyright dates
Submitted by Christoph Hellwig and Dave Kleikamp

16 years agoNo need to handle regular files in jfs_mknod
Dave Kleikamp [Thu, 4 Apr 2002 04:39:33 +0000 (22:39 -0600)]
No need to handle regular files in jfs_mknod
Submitted by Christoph Hellwig

16 years agoRemove register keyword from JFS code
Dave Kleikamp [Thu, 4 Apr 2002 04:37:50 +0000 (22:37 -0600)]
Remove register keyword from JFS code

Submitted by Christoph Hellwig

16 years agoJFS: simplify sync_metapage
Dave Kleikamp [Thu, 4 Apr 2002 04:21:22 +0000 (22:21 -0600)]
JFS: simplify sync_metapage

Submitted by Christoph Hellwig

16 years agoAdd support for external JFS journal
Dave Kleikamp [Thu, 4 Apr 2002 04:16:43 +0000 (22:16 -0600)]
Add support for external JFS journal

Submitted by Christoph Hellwig & Dave Kleikamp

16 years agoJFS: remove dead code
Dave Kleikamp [Thu, 4 Apr 2002 04:07:12 +0000 (22:07 -0600)]
JFS: remove dead code

Submitted by Christoph Hellwig

16 years agoJFS include cleanup
Dave Kleikamp [Thu, 4 Apr 2002 04:02:58 +0000 (22:02 -0600)]
JFS include cleanup

Remove redundant include of slab.h
Submitted by Christoph Hellwig

16 years agoCset exclude: davej@suse.de|ChangeSet|20020403195622
Linus Torvalds [Thu, 4 Apr 2002 04:01:06 +0000 (20:01 -0800)]
Cset exclude: davej@suse.de|ChangeSet|20020403195622

16 years agoDetect bad JFS directory to avoid infinite loop
Dave Kleikamp [Thu, 4 Apr 2002 04:01:01 +0000 (22:01 -0600)]
Detect bad JFS directory to avoid infinite loop

16 years agoMerge bk://bk.arm.linux.org.uk
Linus Torvalds [Thu, 4 Apr 2002 03:38:56 +0000 (19:38 -0800)]
Merge bk://bk.arm.linux.org.uk
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoMerge flint.arm.linux.org.uk:/usr/src/linux-bk-2.5/linux-2.5
Russell King [Thu, 4 Apr 2002 10:45:09 +0000 (11:45 +0100)]
Merge flint.arm.linux.org.uk:/usr/src/linux-bk-2.5/linux-2.5
into flint.arm.linux.org.uk:/usr/src/linux-bk-2.5/linux-2.5-rmk

16 years agoMerge http://gkernel.bkbits.net/net-drivers-2.5
Linus Torvalds [Thu, 4 Apr 2002 03:35:49 +0000 (19:35 -0800)]
Merge http://gkernel.bkbits.net/net-drivers-2.5
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoMerge http://gkernel.bkbits.net/fs-2.5
Linus Torvalds [Thu, 4 Apr 2002 03:35:23 +0000 (19:35 -0800)]
Merge http://gkernel.bkbits.net/fs-2.5
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoMerge http://gkernel.bkbits.net/irda-2.5
Linus Torvalds [Thu, 4 Apr 2002 03:33:59 +0000 (19:33 -0800)]
Merge http://gkernel.bkbits.net/irda-2.5
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

16 years agoFix up bad time compare from the -dj merge
Linus Torvalds [Thu, 4 Apr 2002 03:33:06 +0000 (19:33 -0800)]
Fix up bad time compare from the -dj merge

16 years agoMerge
Linus Torvalds [Thu, 4 Apr 2002 03:32:49 +0000 (19:32 -0800)]
Merge

16 years agoMerge linux-isdn@linux-isdn.bkbits.net:linux-2.5.make
Kai Germaschewski [Thu, 4 Apr 2002 03:31:15 +0000 (21:31 -0600)]
Merge linux-isdn@linux-isdn.bkbits.net:linux-2.5.make
into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

16 years ago[PATCH] kjournald exits with nonzero preempt_count
Robert Love [Thu, 4 Apr 2002 03:24:52 +0000 (19:24 -0800)]
[PATCH] kjournald exits with nonzero preempt_count

The preempt_count debug check that went into 2.5.8-pre1 already caught a
simple case in kjournald.  Specifically, kjournald does not drop the BKL
when it exits as it knows schedule will do so for it.

For the sake of clarity and exiting with a preempt_count of zero, the
attached patch explicitly calls unlock_kernel when kjournald is exiting.

16 years agoDon't allow preemption to change task state.
Linus Torvalds [Thu, 4 Apr 2002 03:17:31 +0000 (19:17 -0800)]
Don't allow preemption to change task state.

16 years agoe100 net driver update 3/3:
Eli Kupermann [Thu, 4 Apr 2002 01:21:00 +0000 (20:21 -0500)]
e100 net driver update 3/3:
Adding proper print level qualifier to the printk calls.

16 years agoe100 net driver update 2/3:
Eli Kupermann [Thu, 4 Apr 2002 01:20:24 +0000 (20:20 -0500)]
e100 net driver update 2/3:
Adding missing pci write flush to the procedure e100_exec_cmd

16 years agoe100 net driver update 1/3:
Eli Kupermann [Thu, 4 Apr 2002 01:19:38 +0000 (20:19 -0500)]
e100 net driver update 1/3:
The patch separates max busy wait constants making in max of 100 usec for
wait scb and max of 50 usec for wait cus idle. These constants found
sufficient using heavy traffic tests.

16 years agoext3 filesystem sync mount speedup:
Andrew Morton [Thu, 4 Apr 2002 01:17:52 +0000 (20:17 -0500)]
ext3 filesystem sync mount speedup:
Again, we don't need to sync indirects as we dirty them because
we run a commit if IS_SYNC(inode) prior to returning to the
caller of write(2).

Writing a 10 meg file in 0.1 meg chunks is sped up by, err,
a factor of fifty.  That's a best case.

16 years agoAndrew Morton's ext2 sync mount speedup. Description:
Jeff Garzik [Thu, 4 Apr 2002 01:16:12 +0000 (20:16 -0500)]
Andrew Morton's ext2 sync mount speedup.  Description:

At present, when mounted synchronously or with `chattr +S' in effect,
ext2 syncs the indirect blocks for every new block when extending a
file.

This is not necessary, because a sync is performed on the way out of
generic_file_write().  This will pick up all necessary data from
inode->i_dirty_buffers and inode->i_dirty_data_buffers, and is
sufficient.

The patch removes all the syncing of indirect blocks.

On a non-write-caching scsi disk, an untar of the util-linux tarball
runs three times faster.  Writing a 100 megabyte file in one megabyte
chunks speeds up ten times.

The patch also removes the intermediate indirect block syncing on the
truncate() path.  Instead, we sync the indirects at a single place, via
inode->i_dirty_buffers.  This not only means that the writes (may)
cluster better.  It means that we perform much, much less actual I/O
during truncate, because most or all of the indirects will no longer be
needed for the file, and will be invalidated.

fsync() and msync() still work correctly.  One side effect of this
patch is that VM-initiated writepage() against a file hole will no
longer block on writeout of indirect blocks.  This is good.

16 years agoFix jiffies-comparison timeout bug in arlan net driver.
Alan Cox [Thu, 4 Apr 2002 00:57:55 +0000 (19:57 -0500)]
Fix jiffies-comparison timeout bug in arlan net driver.

16 years agoVarious minor bug fixes for 3c59x net driver.
Andrew Morton [Thu, 4 Apr 2002 00:54:47 +0000 (19:54 -0500)]
Various minor bug fixes for 3c59x net driver.

16 years agoMerge some new PCI ids from e100 to eepro100 net driver.
Arjan van de Ven [Thu, 4 Apr 2002 00:50:43 +0000 (19:50 -0500)]
Merge some new PCI ids from e100 to eepro100 net driver.

16 years agoThis fixes the "i_blocks went wrong when the disk filled up"
Andrew Morton [Thu, 4 Apr 2002 00:39:51 +0000 (19:39 -0500)]
This fixes the "i_blocks went wrong when the disk filled up"
problem.

In ext3_new_block() we increment i_blocks early, so the
quota operation can be performed outside lock_super().
But if the block allocation ends up failing, we forget to
undo the allocation.

This is not a serious bug, and probably does not warrant
an upgrade for production machines.  Its effects are:

1) errors are generated from e2fsck and

2) users could appear to be over quota when they really aren't.

The patch undoes the accounting operation if the allocation
ends up failing.

16 years agoMerge hydra net drvr conversion to Zorro-specific
Dave Jones [Thu, 4 Apr 2002 00:32:56 +0000 (19:32 -0500)]
Merge hydra net drvr conversion to Zorro-specific
z_{read,write}[bwl] routines from 2.4.x.

16 years agoIrDA: Correct location of dev tx stats update
Jean Tourrilhes [Thu, 4 Apr 2002 00:04:33 +0000 (19:04 -0500)]
IrDA: Correct location of dev tx stats update

16 years agoIrDA: handle new NSC chip variant
Jean Tourrilhes [Thu, 4 Apr 2002 00:03:32 +0000 (19:03 -0500)]
IrDA: handle new NSC chip variant

16 years agoIrDA USB disconnect changes:
Jean Tourrilhes [Thu, 4 Apr 2002 00:02:42 +0000 (19:02 -0500)]
IrDA USB disconnect changes:
o [CRITICA] Fix race condition between disconnect and the rest
o [CRITICA] Force synchronous unlink of URBs in disconnect
o [CRITICA] Cleanup instance if disconnect before close
        <Following patch from Martin Diehl>
o [CRITICA] Call usb_submit_urb() with GPF_ATOMIC