linux-2.6
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/tglx/mtd-2.6
Linus Torvalds [Wed, 13 Jul 2005 19:19:30 +0000 (12:19 -0700)] 
Merge /pub/scm/linux/kernel/git/tglx/mtd-2.6

19 years ago[PATCH] Amiga joystick: Fix typo introduced by the open/close race fixes
Geert Uytterhoeven [Wed, 13 Jul 2005 10:56:42 +0000 (12:56 +0200)] 
[PATCH] Amiga joystick: Fix typo introduced by the open/close race fixes

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: add 970MP PVR
Olof Johansson [Wed, 13 Jul 2005 08:11:44 +0000 (01:11 -0700)] 
[PATCH] ppc64: add 970MP PVR

Add PVR value and tests for 970MP.  Also switch to a simpler (but slightly
longer) check at init time for simplicity.

Signed-off-by: Olof Johansson <olof@austin.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: kill bitfields in ppc64 hash code
David Gibson [Wed, 13 Jul 2005 08:11:42 +0000 (01:11 -0700)] 
[PATCH] ppc64: kill bitfields in ppc64 hash code

This patch removes the use of bitfield types from the ppc64 hash table
manipulation code.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: make -j12 all fails in uImage target
Olaf Hering [Wed, 13 Jul 2005 08:11:41 +0000 (01:11 -0700)] 
[PATCH] ppc32: make -j12 all fails in uImage target

make -j zImage may call if_changed twice at the same time, the result is a
corrupted vmlinux.gz

Write to a temporary file for the time being until someone with make skills
fix the serialization properly.

Signed-off-by: Olaf Hering <olh@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Tom Rini <trini@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] NFS: procfs/sysctl interfaces for lockd do not work on x86_64
Steve Dickson [Wed, 13 Jul 2005 08:10:47 +0000 (01:10 -0700)] 
[PATCH] NFS: procfs/sysctl interfaces for lockd do not work on x86_64

Allow the setting of NLM timeouts and grace periods through the proc and
sysclt interfaces on x86_64 architectures

Signed-off-by: Steve Dickson <steved@redhat.com>
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] s390: fadvise hint values.
Martin Schwidefsky [Wed, 13 Jul 2005 08:10:46 +0000 (01:10 -0700)] 
[PATCH] s390: fadvise hint values.

Add special case for the POSIX_FADV_DONTNEED and POSIX_FADV_NOREUSE hint
values for s390-64.  The user space values in the s390-64 glibc headers for
these two defines have always been 6 and 7 instead of 4 and 5.  All 64 bit
applications therefore use the "wrong" values.  To get these applications
working without recompiling the kernel needs to accept the "wrong" values.
Since the values for s390-31 are 4 and 5 the compat wrapper for fadvise64
and fadvise64_64 need to rewrite the values for 31 bit system calls.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: fix destroy_context() race condition
Guillaume Autran [Wed, 13 Jul 2005 08:10:45 +0000 (01:10 -0700)] 
[PATCH] ppc32: fix destroy_context() race condition

Fix for a race condition when a task gets preempted by another task while
executing the destroy_context(...) in a FEW_CONTEXTS environment.
mm->context == NO_CONTEXT but the context_map may indicate all contexts are
in use.

The solution to this problem is to disable kernel preemption while
destroying a MMU context.

Signed-off-by: Guillaume Autran <gautran@mrv.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix soft lockup due to NTFS: VFS part and explanation
Anton Altaparmakov [Wed, 13 Jul 2005 08:10:44 +0000 (01:10 -0700)] 
[PATCH] Fix soft lockup due to NTFS: VFS part and explanation

Something has changed in the core kernel such that we now get concurrent
inode write outs, one e.g via pdflush and one via sys_sync or whatever.
This causes a nasty deadlock in ntfs.  The only clean solution
unfortunately requires a minor vfs api extension.

First the deadlock analysis:

Prerequisive knowledge: NTFS has a file $MFT (inode 0) loaded at mount
time.  The NTFS driver uses the page cache for storing the file contents as
usual.  More interestingly this file contains the table of on-disk inodes
as a sequence of MFT_RECORDs.  Thus NTFS driver accesses the on-disk inodes
by accessing the MFT_RECORDs in the page cache pages of the loaded inode
$MFT.

The situation: VFS inode X on a mounted ntfs volume is dirty.  For same
inode X, the ntfs_inode is dirty and thus corresponding on-disk inode,
which is as explained above in a dirty PAGE_CACHE_PAGE belonging to the
table of inodes ($MFT, inode 0).

What happens:

Process 1: sys_sync()/umount()/whatever...  calls __sync_single_inode() for
$MFT -> do_writepages() -> write_page for the dirty page containing the
on-disk inode X, the page is now locked -> ntfs_write_mst_block() which
clears PageUptodate() on the page to prevent anyone else getting hold of it
whilst it does the write out (this is necessary as the on-disk inode needs
"fixups" applied before the write to disk which are removed again after the
write and PageUptodate is then set again).  It then analyses the page
looking for dirty on-disk inodes and when it finds one it calls
ntfs_may_write_mft_record() to see if it is safe to write this on-disk
inode.  This then calls ilookup5() to check if the corresponding VFS inode
is in icache().  This in turn calls ifind() which waits on the inode lock
via wait_on_inode whilst holding the global inode_lock.

Process 2: pdflush results in a call to __sync_single_inode for the same
VFS inode X on the ntfs volume.  This locks the inode (I_LOCK) then calls
write-inode -> ntfs_write_inode -> map_mft_record() -> read_cache_page() of
the page (in page cache of table of inodes $MFT, inode 0) containing the
on-disk inode.  This page has PageUptodate() clear because of Process 1
(see above) so read_cache_page() blocks when tries to take the page lock
for the page so it can call ntfs_read_page().

Thus Process 1 is holding the page lock on the page containing the on-disk
inode X and it is waiting on the inode X to be unlocked in ifind() so it
can write the page out and then unlock the page.

And Process 2 is holding the inode lock on inode X and is waiting for the
page to be unlocked so it can call ntfs_readpage() or discover that
Process 1 set PageUptodate() again and use the page.

Thus we have a deadlock due to ifind() waiting on the inode lock.

The only sensible solution: NTFS does not care whether the VFS inode is
locked or not when it calls ilookup5() (it doesn't use the VFS inode at
all, it just uses it to find the corresponding ntfs_inode which is of
course attached to the VFS inode (both are one single struct); and it uses
the ntfs_inode which is subject to its own locking so I_LOCK is irrelevant)
hence we want a modified ilookup5_nowait() which is the same as ilookup5()
but it does not wait on the inode lock.

Without such functionality I would have to keep my own ntfs_inode cache in
the NTFS driver just so I can find ntfs_inodes independent of their VFS
inodes which would be slow, memory and cpu cycle wasting, and incredibly
stupid given the icache already exists in the VFS.

Below is a patch that does the ilookup5_nowait() implementation in
fs/inode.c and exports it.

ilookup5_nowait.diff:

Introduce ilookup5_nowait() which is basically the same as ilookup5() but
it does not wait on the inode's lock (i.e. it omits the wait_on_inode()
done in ifind()).

This is needed to avoid a nasty deadlock in NTFS.

Signed-off-by: Anton Altaparmakov <aia21@cantab.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] inotify: misc cleanup
Robert Love [Wed, 13 Jul 2005 17:49:23 +0000 (13:49 -0400)] 
[PATCH] inotify: misc cleanup

Really simple, basic cleanup.

Signed-off-by: Robert Love <rml@novell.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] inotify: event ordering
Robert Love [Wed, 13 Jul 2005 16:38:39 +0000 (12:38 -0400)] 
[PATCH] inotify: event ordering

This rearranges the event ordering for "open" to be consistent with the
ordering of the other events.

Signed-off-by: Robert Love <rml@novell.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] inotify: move sysctl
Robert Love [Wed, 13 Jul 2005 16:38:18 +0000 (12:38 -0400)] 
[PATCH] inotify: move sysctl

This moves the inotify sysctl knobs to "/proc/sys/fs/inotify" from
"/proc/sys/fs".  Also some related cleanup.

Signed-off-by: Robert Love <rml@novell.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] fix voyager subarchitecture EXPORT_SYMBOL breakage caused by i386_ksym reduction
James Bottomley [Wed, 13 Jul 2005 13:38:05 +0000 (09:38 -0400)] 
[PATCH] fix voyager subarchitecture EXPORT_SYMBOL breakage caused by i386_ksym reduction

This patch:

[PATCH] Remove i386_ksyms.c, almost

made files like smp.c do their own EXPORT_SYMBOLS.  This means that all
subarchitectures that override these symbols now have to do the exports
themselves.  This patch adds the exports for voyager (which is the most
affected since it has a separate smp harness).  However, someone should
audit all the other subarchitectures to see if any others got broken.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge head 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/shaggy/jfs-2.6
Linus Torvalds [Wed, 13 Jul 2005 17:37:36 +0000 (10:37 -0700)] 
Merge head 'for-linus' of /linux/kernel/git/shaggy/jfs-2.6

19 years ago[PATCH] Lindent: ignore .indent.pro
Jeff Mahoney [Wed, 13 Jul 2005 15:55:42 +0000 (11:55 -0400)] 
[PATCH] Lindent: ignore .indent.pro

 When I recently submitted a Lindent patch, it turned out that my .indent.pro
 options were also applied to the tree. This patch directs indent(1) to ignore
 the .indent.pro directives and only use options specified on the command
 line.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge /home/torvalds/linux-2.6-arm
Linus Torvalds [Wed, 13 Jul 2005 17:12:50 +0000 (10:12 -0700)] 
Merge /home/torvalds/linux-2.6-arm

19 years ago[MTD] cfi_cmdset_0002: Plugged a mem leak.
Joern Engel [Wed, 13 Jul 2005 15:45:43 +0000 (16:45 +0100)] 
[MTD] cfi_cmdset_0002: Plugged a mem leak.

Signed-off-by: Joern Engel <joern@wohnheim.fh-wedel.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years agoJFS: Need to be root to create files with security context
Ian Dall [Wed, 13 Jul 2005 14:15:18 +0000 (09:15 -0500)] 
JFS: Need to be root to create files with security context

It turns out this is due to some inverted logic in xattr.c

Signed-off-by: Dave Kleikamp <shaggy@austin.ibm.com>
19 years agoJFS: Allow security.* xattrs to be set on symlinks
Dave Kleikamp [Wed, 13 Jul 2005 14:07:53 +0000 (09:07 -0500)] 
JFS: Allow security.* xattrs to be set on symlinks

All of the different xattr namespaces have different rules.
user.* and ACL's are not allowed on symlinks, and since these were the
first xattrs implemented, I assumed there was no need to support xattrs
on symlinks.  This one-line patch should fix it.

Signed-off-by: Dave Kleikamp <shaggy@austin.ibm.com>
19 years agoMerge with /home/shaggy/git/linus-clean/
Dave Kleikamp [Wed, 13 Jul 2005 13:57:38 +0000 (08:57 -0500)] 
Merge with /home/shaggy/git/linus-clean/

Signed-off-by: Dave Kleikamp <shaggy@austin.ibm.com>
19 years agoMerge with rsync://fileserver/linux
Thomas Gleixner [Wed, 13 Jul 2005 08:45:00 +0000 (10:45 +0200)] 
Merge with rsync://fileserver/linux

19 years agoLinux 2.6.13-rc3 v2.6.13-rc3
Linus Torvalds [Wed, 13 Jul 2005 04:46:46 +0000 (21:46 -0700)] 
Linux 2.6.13-rc3

Yeah, this time hopefully I'm not confusing the version
numbers. The last release was -rc2, _this_ is -rc3.

19 years ago[PATCH] inotify
Robert Love [Tue, 12 Jul 2005 21:06:03 +0000 (17:06 -0400)] 
[PATCH] inotify

inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:

        * dnotify requires the opening of one fd per each directory
          that you intend to watch. This quickly results in too many
          open files and pins removable media, preventing unmount.
        * dnotify is directory-based. You only learn about changes to
          directories. Sure, a change to a file in a directory affects
          the directory, but you are then forced to keep a cache of
          stat structures.
        * dnotify's interface to user-space is awful.  Signals?

inotify provides a more usable, simple, powerful solution to file change
notification:

        * inotify's interface is a system call that returns a fd, not SIGIO.
  You get a single fd, which is select()-able.
        * inotify has an event that says "the filesystem that the item
          you were watching is on was unmounted."
        * inotify can watch directories or files.

Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.

See Documentation/filesystems/inotify.txt.

Signed-off-by: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoreiserfs: run scripts/Lindent on reiserfs code
Linus Torvalds [Wed, 13 Jul 2005 03:21:28 +0000 (20:21 -0700)] 
reiserfs: run scripts/Lindent on reiserfs code

This was a pure indentation change, using:

scripts/Lindent fs/reiserfs/*.c include/linux/reiserfs_*.h

to make reiserfs match the regular Linux indentation style.  As Jeff
Mahoney <jeffm@suse.com> writes:

 The ReiserFS code is a mix of a number of different coding styles, sometimes
 different even from line-to-line. Since the code has been relatively stable
 for quite some time and there are few outstanding patches to be applied, it
 is time to reformat the code to conform to the Linux style standard outlined
 in Documentation/CodingStyle.

 This patch contains the result of running scripts/Lindent against
 fs/reiserfs/*.c and include/linux/reiserfs_*.h. There are places where the
 code can be made to look better, but I'd rather keep those patches separate
 so that there isn't a subtle by-hand hand accident in the middle of a huge
 patch. To be clear: This patch is reformatting *only*.

 A number of patches may follow that continue to make the code more consistent
 with the Linux coding style.

 Hans wasn't particularly enthusiastic about these patches, but said he
 wouldn't really oppose them either.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] reiserfs: fix up case where indent misreads the code
Jeff Mahoney [Tue, 12 Jul 2005 23:19:30 +0000 (19:19 -0400)] 
[PATCH] reiserfs: fix up case where indent misreads the code

 indent(1) doesn't know how to handle the "do not compile" error. It results
 in the item_ops array declaration being indented a tab stop in when it should
 not be. This patch replaces it with a #error that describes why it's failing.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper snapshots: Handle origin extension
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:05 +0000 (15:53 -0700)] 
[PATCH] device-mapper snapshots: Handle origin extension

Handle writes to a snapshot-origin device that has been extended since the
snapshot was taken.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper: Fix dm_swap_table error cases
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:05 +0000 (15:53 -0700)] 
[PATCH] device-mapper: Fix dm_swap_table error cases

Fix dm_swap_table() __bind error cases: a missing unlock, and EINVAL
preferable to EPERM.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper multipath: Fix pg initialisation races
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:04 +0000 (15:53 -0700)] 
[PATCH] device-mapper multipath: Fix pg initialisation races

Prevent more than one priority group initialisation function from being
outstanding at once.  Otherwise the completion functions interfere with each
other.  Also, reloading the table could reference a freed pointer.

Only reset queue_io in pg_init_complete if another pg_init isn't required.
Skip process_queued_ios if the queue is empty so that we only trigger a
pg_init if there's I/O.

Signed-off-by: Lars Marowsky-Bree <lmb@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper multipath: Avoid possible suspension deadlock
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:03 +0000 (15:53 -0700)] 
[PATCH] device-mapper multipath: Avoid possible suspension deadlock

To avoid deadlock when suspending a multipath device after all its paths have
failed, stop queueing any I/O that is about to fail *before* calling
freeze_bdev instead of after.

Instead of setting a multipath 'suspended' flag which would have to be reset
if an error occurs during the process, save the previous queueing state and
leave userspace to restore if it wishes.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper multipath: Flush workqueue when destroying
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:02 +0000 (15:53 -0700)] 
[PATCH] device-mapper multipath: Flush workqueue when destroying

The multipath destructor must flush its workqueue.  Otherwise items that
reference the destroyed object could remain.

From: "goggin, edward" <egoggin@emc.com>
Signed-off-by: Lars Marowsky-Bree <lmb@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] device-mapper multipath: Barriers not supported
Alasdair G Kergon [Tue, 12 Jul 2005 22:53:01 +0000 (15:53 -0700)] 
[PATCH] device-mapper multipath: Barriers not supported

dm multipath will report barriers as not supported with this patch.

Signed-off-by: Lars Marowsky-Bree <lmb@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/lenb/linux-2.6
Linus Torvalds [Tue, 12 Jul 2005 23:04:50 +0000 (16:04 -0700)] 
Merge /pub/scm/linux/kernel/git/lenb/linux-2.6

19 years ago[PATCH] v4l: broken hybrid dvb inclusion
Michael Krufky [Tue, 12 Jul 2005 20:59:08 +0000 (13:59 -0700)] 
[PATCH] v4l: broken hybrid dvb inclusion

Always include dvb frontend code for hybrid cx88 and saa7134 boards.

Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: TV EEPROM
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:59:07 +0000 (13:59 -0700)] 
[PATCH] v4l: TV EEPROM

- Eliminated unused code.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: tuner-3026 - replace obsolete ioctl value
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:59:06 +0000 (13:59 -0700)] 
[PATCH] v4l: tuner-3026 - replace obsolete ioctl value

- obsolete TUNER_SET_TVFREQ changed to VIDIOCSFREQ.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: SAA7134 Update
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:59:01 +0000 (13:59 -0700)] 
[PATCH] v4l: SAA7134 Update

- Corrected all cards marked as 7135 cards to 7133.
- Add new card support for Compro VideoMate TV Gold+II.
- Add new card support for Kworld Xpert TV PVR7134
- Add new card support for Typhoon DVB-T Cardbus.
- Changes to comply with CodingStyle: // comments converted to /* */
- Remove irq2_mask field from saa7134_dev structure.
- Collect all the bits needed in saa7134_hwinit2() instead.
- Distinguish the different variants of the Medion MD7134 modules via eeprom
- moved Philips FMD1216 radio specific setup to saa7134-core.c
- Fix kernel compile error with CONFIG_MODULES=n
- Cleanup tuner private calls.
- Some Indent fixes.

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Hannibal <hannibal@megapolis.pl>
Signed-off-by: Elshin Roman <roxmail@list.ru>
Signed-off-by: Hermann Pitton <hermann.pitton@onlinehome.de>
Signed-off-by: Juergen Orschiedt <jorschiedt@web.de>
Signed-off-by: Hartmut Hackmann <hartmut.hackmann@t-online.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: MXB fix to correct tuner ioctl
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:59 +0000 (13:58 -0700)] 
[PATCH] v4l: MXB fix to correct tuner ioctl

- driver command adapted to use new control (TUNER_SET_TYPE_ADDR,
  instead of TUNER_SET_TYPE)

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: drivers/media/video/Kconfig
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:57 +0000 (13:58 -0700)] 
[PATCH] v4l: drivers/media/video/Kconfig

- Removed obsolete option. Current code needs multi tuner.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: I2C Tuner
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:55 +0000 (13:58 -0700)] 
[PATCH] v4l: I2C Tuner

- Fixed a trouble on tuner-core that generates erros on computers with more
  than one TV card.
- Rename tuner structures fields.
- Tail spaces removed.
- I2C cleanups and converged to a basic reference structure.
- Removed unused structures.
- Fix setting frequency on tda8290.
- Added code for TEA5767 autodetection.
- Standby mode support implemented. It is used to disable
  a non used tuner. Currenlty implemented on tea5767.
- New macro: set_type disables other tuner when changing mode.
- Some cleanups.
- Use 50 kHz step when tunning radio for most tuners to improve precision.

Signed-off-by: Fabien Perrot <perrot1983@yahoo.fr>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-By: Nickolay V. Shmyrev <nshmyrev@yandex.ru>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: I2C Miscelaneous
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:53 +0000 (13:58 -0700)] 
[PATCH] v4l: I2C Miscelaneous

- Removed unused structures.
- CodingStyle rules applied to comments.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: I2C Infrared Remote Control
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:52 +0000 (13:58 -0700)] 
[PATCH] v4l: I2C Infrared Remote Control

- Removed unused structures.
- CodingStyle rules applied to comments.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: I2C BT832
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:51 +0000 (13:58 -0700)] 
[PATCH] v4l: I2C BT832

- Removed unused structures.

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: Documentation
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:49 +0000 (13:58 -0700)] 
[PATCH] v4l: Documentation

- Card definitions updated.
- Tail spaces removed.
- Mark all 7135 cards as 7133.
- Correct info about sync byte for MPEG-2 transport stream packets.

Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: hermann pitton <hermann.pitton@onlinehome.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: SAA7134 hybrid DVB
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:47 +0000 (13:58 -0700)] 
[PATCH] v4l: SAA7134 hybrid DVB

- Add new Typhoon DVB-T Cardbus.
- DVB-T support for MD7134 cardbus and the PCI variants
- initial DVB-T support for Lifeview Flydvb-t duo
- DVB-T support for Philips TOUGH reference design
- Don't turn off the xtal output of tda8274/75 in sleep mode
- Let Kconfig decide whether to include frontend-specific code in saa7134-dvb.
- Removed unused structures.

Signed-off-by: Juergen Orschiedt <jorschiedt@web.de>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Hartmut Hackmann <hartmut.hackmann@t-online.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: CX88 Update
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:44 +0000 (13:58 -0700)] 
[PATCH] v4l: CX88 Update

- Removed unused structures.
- Removed BTTV version check.
- Some debug structs moved to their own .c file and converted to static
- Comment changed to express better when attach_inform is running
- set_freq removed from set_mode at tuner-core.c.
- I2C cleanups and converged to a basic reference structure.
- Rename tuner structures fields.
- It calls VIDIOC_G_FREQUENCY to get tuner freq from tuner.
- added missing contrast offset value, set to 0.
- Let Kconfig decide whether to include frontend-specific code.

Signed-Off-By: Nickolay V. Shmyrev <nshmyrev@yandex.ru>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: BTTV update
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:42 +0000 (13:58 -0700)] 
[PATCH] v4l: BTTV update

- use DMA_32BIT_MASK.
- Rename tuner structures fields.
- Tail spaces removed.
- I2C cleanups and converged to a basic reference structure.
- Removed unused structures.
- Removed BTTV version check.

Signed-off-by: <domen@coderock.org>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-Off-By: Nickolay V. Shmyrev <nshmyrev@yandex.ru>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v4l: BTTV input
Mauro Carvalho Chehab [Tue, 12 Jul 2005 20:58:41 +0000 (13:58 -0700)] 
[PATCH] v4l: BTTV input

Changes to comply with CodingStyle: // comments converted to /* */

Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] dvb: dst: printk -> dprintk
Johannes Stezenbach [Tue, 12 Jul 2005 20:58:40 +0000 (13:58 -0700)] 
[PATCH] dvb: dst: printk -> dprintk

- stop log spamming when running femon (printk -> dprintk)

Signed-off-by: Johannes Stezenbach <js@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] dvb: fix kobject names (no slashes)
Julian Scheel [Tue, 12 Jul 2005 20:58:39 +0000 (13:58 -0700)] 
[PATCH] dvb: fix kobject names (no slashes)

The / in the driver name (budget dvb /w video in) is not a valid character for
device names - removed it, now it works!

Same for ttusb-budget.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Johannes Stezenbach <js@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] dvb: usb: fix some typos
Patrick Boettcher [Tue, 12 Jul 2005 20:58:38 +0000 (13:58 -0700)] 
[PATCH] dvb: usb: fix some typos

corrected some typos.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Johannes Stezenbach <js@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] dvb: LGDT3302 QAM256 initialization fix
Michael Krufky [Tue, 12 Jul 2005 20:58:37 +0000 (13:58 -0700)] 
[PATCH] dvb: LGDT3302 QAM256 initialization fix

- Initialize all non mutually exclusive variables
  without regard to the mode selected.
- Do a software reset each time the parameters are
  set, regardless of whether anything changes.
  This may allow an application to recover from a
  hung condition.
- Improved error reporting.
- Removed $Id:$

Signed-off-by: Mac Michaels <wmichaels1@earthlink.net>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] hardirq uses preempt
Randy Dunlap [Tue, 12 Jul 2005 20:58:36 +0000 (13:58 -0700)] 
[PATCH] hardirq uses preempt

hardirq.h uses preempt_count() from preempt.h

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] MAINTAINERS: irda-users@lists.sourceforge.net is subscribers only
Olaf Hering [Tue, 12 Jul 2005 20:58:35 +0000 (13:58 -0700)] 
[PATCH] MAINTAINERS: irda-users@lists.sourceforge.net is subscribers only

Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kexec-ppc: fix for ksysfs crash_notes
Albert Herranz [Tue, 12 Jul 2005 20:58:34 +0000 (13:58 -0700)] 
[PATCH] kexec-ppc: fix for ksysfs crash_notes

The following patch prevents the crash dump helper code found within kexec
from breaking ppc which still lacks crash dump functionality.

ksysfs crash_notes attribute handling was left under CONFIG_KEXEC for
simplicity although it is not strictly kexec related.

We provide here a dummy definition for crash_notes on ppc.

Signed-off-by: Albert Herranz <albert_herranz@yahoo.es>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Documentation/kernel-parameters.txt: fix a typo
Adrian Bunk [Tue, 12 Jul 2005 20:58:33 +0000 (13:58 -0700)] 
[PATCH] Documentation/kernel-parameters.txt: fix a typo

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] TB0219: add PCI IRQ initialization
Yoichi Yuasa [Tue, 12 Jul 2005 20:58:32 +0000 (13:58 -0700)] 
[PATCH] TB0219: add PCI IRQ initialization

This patch adds PCI IRQ initialization to TB0219 driver.

Signed-off-by: Yoichi Yuasa <yuasa@hh.iij4u.or.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IBM_ASM Kconfig corrections
Adrian Bunk [Tue, 12 Jul 2005 20:58:32 +0000 (13:58 -0700)] 
[PATCH] IBM_ASM Kconfig corrections

This patch contains the following fixes:
- IBM_ASM must depend on PCI
- remove useless "default n"
- correct the URL to further information

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] tlb.h warning fix
Andrew Morton [Tue, 12 Jul 2005 20:58:31 +0000 (13:58 -0700)] 
[PATCH] tlb.h warning fix

free_pages_and_swap_cache() and free_page_and_swap_cache() use release_pages()
and page_cache_release() respectively, so make sure that we have the
declarations in scope.

Cc: Olaf Hering <olh@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cdev: cdev_put oops
Brian King [Tue, 12 Jul 2005 20:58:30 +0000 (13:58 -0700)] 
[PATCH] cdev: cdev_put oops

While fixing an oops in the st driver in a dirty release path, I
encountered an oops in cdev_put for cdevs allocated using cdev_alloc.  If
cdev_del is called when the cdev kobject still has an open user, when the
last cdev_put is called, the cdev_put will call kobject_put, which will end
up ultimately releasing the cdev in cdev_dynamic_release.  Patch fixes the
oops by preventing cdev_put from accessing freed memory.

Signed-off-by: Brian King <brking@us.ibm.com>
Cc: <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ext2: fix mount options parting
Jan Kara [Tue, 12 Jul 2005 20:58:29 +0000 (13:58 -0700)] 
[PATCH] ext2: fix mount options parting

Restore old set of ext2 mount options when remounting of a filesystem
fails.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ext3: fix options parsing
Jan Kara [Tue, 12 Jul 2005 20:58:28 +0000 (13:58 -0700)] 
[PATCH] ext3: fix options parsing

Fix a problem with ext3 mount option parsing.  When remount of a filesystem
fails, old options are now restored.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] reset real_timer target on exec leader change
Roland McGrath [Tue, 12 Jul 2005 20:58:27 +0000 (13:58 -0700)] 
[PATCH] reset real_timer target on exec leader change

When a noninitial thread does exec, it becomes the new group leader.  If
there is a ITIMER_REAL timer running, it points at the old group leader and
when it fires it can follow a stale pointer.  The timer data needs to be
reset to point at the exec'ing thread that is becoming the group leader.
This has to synchronize with any concurrent firing of the timer to make
sure that it_real_fn can never run when the data points to a thread that
might have been reaped already.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] xtensa: use ssleep() instead of schedule_timeout()
Nishanth Aravamudan [Tue, 12 Jul 2005 20:58:26 +0000 (13:58 -0700)] 
[PATCH] xtensa: use ssleep() instead of schedule_timeout()

Replace schedule_timeout() with ssleep() to guarantee the task delays as
expected.

Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] xtensa: remove old syscalls
Chris Zankel [Tue, 12 Jul 2005 20:58:25 +0000 (13:58 -0700)] 
[PATCH] xtensa: remove old syscalls

This patch fixes some minor bugs introduced by the previous patch (remove
old syscalls).  Both patches remove the obsolete syscalls.  The changes in
this patch were suggested by Arnd Bergmann.  The vmlinux.lds.S changes are
required for the latest gcc/binutils.

Signed-off-by: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] uml: tlb flushing fix
Benjamin LaHaise [Tue, 12 Jul 2005 20:58:22 +0000 (13:58 -0700)] 
[PATCH] uml: tlb flushing fix

This patch fixes a fairly serious tlb flushing bug that makes aio use under
uml very unreliable -- SEGVs, Oops and panic()s occur as a result of stale
tlb entires being used by uml when aio switches mms due to the fact that
uml does not implement the activate_mm() hook.  This patch introduces a
simple but correct approach (read: hammer) for implementing activate_mm()
in uml by doing a force_flush_all() if the new mm is different from old.
With this patch in place, uml is able to succeed at the aio test case that
was randomly faulting for me before.

Cc: Jeff Dike <jdike@addtoit.com>
Cc: <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] uml:remove user_constants.h on clean
Paolo 'Blaisorblade' Giarrusso [Tue, 12 Jul 2005 20:58:20 +0000 (13:58 -0700)] 
[PATCH] uml:remove user_constants.h on clean

make clean ARCH=um does not remove the generated file
arch/um/include/user_constants.h, fix this.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] alpha: pgprot_uncached() comment
Andrew Morton [Tue, 12 Jul 2005 20:58:19 +0000 (13:58 -0700)] 
[PATCH] alpha: pgprot_uncached() comment

Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mm/filemap_xip.c compilation fix
Geert Uytterhoeven [Tue, 12 Jul 2005 20:58:18 +0000 (13:58 -0700)] 
[PATCH] mm/filemap_xip.c compilation fix

mm/filemap_xip.c: In function `__xip_unmap':
mm/filemap_xip.c:194: request for member `pte' in something not a structure or union

Apparently pte_pfn() takes a pte_t, not a pointer to a pte_t.  From looking
at asm/page.h, it seems to be the same on ia32 or ppc (iff
STRICT_MM_TYPECHECKS is enabled, which is disabled by default on ppc).

Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] yenta: allocate resource fixes
Dominik Brodowski [Tue, 12 Jul 2005 20:58:17 +0000 (13:58 -0700)] 
[PATCH] yenta: allocate resource fixes

The current CardBus window allocation code in yenta_socket is unable to handle
the transparent PCI-bridge handling update in 2.6.13.  We need to check _all_
resources of a given type to find the best one suitable for CardBus windows,
not just the first one.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] yenta: same resources in same structs
Dominik Brodowski [Tue, 12 Jul 2005 20:58:16 +0000 (13:58 -0700)] 
[PATCH] yenta: same resources in same structs

drivers/pci/setup-bus.c enumerates the CardBus windows (bus->resources[])

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: Documentation update
Dominik Brodowski [Tue, 12 Jul 2005 20:58:15 +0000 (13:58 -0700)] 
[PATCH] pcmcia: Documentation update

Update PCMCIA driver changes for patches merged in 2.6.13

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] yenta: fix parent resource determination
Dominik Brodowski [Tue, 12 Jul 2005 20:58:15 +0000 (13:58 -0700)] 
[PATCH] yenta: fix parent resource determination

If the CardBus windows were pre-configured and the CardBus bridge is behind a
transparent PCI-PCI bridge, pci_find_parent_resource() might return a
different resource than the real parent if it is called before the window is
determined.  Therefore, move that call around.

Also fix return of value in void function.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: fix pcmcia-cs compilation
Dominik Brodowski [Tue, 12 Jul 2005 20:58:14 +0000 (13:58 -0700)] 
[PATCH] pcmcia: fix pcmcia-cs compilation

Fix pcmcia-cs compilation with recent pcmcia kernel changes.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] x86_64: section alignment fix
Andrew Morton [Tue, 12 Jul 2005 20:58:13 +0000 (13:58 -0700)] 
[PATCH] x86_64: section alignment fix

This is the second time this has happened: inserting a new section requires
that we adjust the arithmetic which is used to calculate the vsyscall page's
offset.

Cc: Christoph Lameter <christoph@lameter.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] bugfix: two read_inode() calls without clear_inode() call between
Artem B. Bityuckiy [Tue, 12 Jul 2005 20:58:12 +0000 (13:58 -0700)] 
[PATCH] bugfix: two read_inode() calls without clear_inode() call between

Bug symptoms
~~~~~~~~~~~~
For the same inode VFS calls read_inode() twice and doesn't call
clear_inode() between the two read_inode() invocations.

Bug description
~~~~~~~~~~~~~~~
Suppose we have an inode which has zero reference count but is still in
the inode cache. Suppose kswapd invokes shrink_icache_memory() to free
some RAM. In prune_icache() inodes are removed from i_hash. prune_icache
() is then going to call clear_inode(), but drops the inode_lock
spinlock before this. If in this moment another task calls iget() for an
inode which was just removed from i_hash by prune_icache(), then iget()
invokes read_inode() for this inode, because it is *already removed*
from i_hash.

The end result is: we call iget(#N) then iput(#N); inode #N has zero
i_count now and is in the inode cache; kswapd starts. kswapd removes the
inode #N from i_hash ans is preempted; we call iget(#N) again;
read_inode() is invoked as the result; but we expect clear_inode()
before.

Fix
~~~~~~~
To fix the bug I remove inodes from i_hash later, when clear_inode() is
actually called. I remove them from i_hash under spinlock protection.
Since the i_state is set to I_FREEING, it is safe to do this. The others
will sleep waiting for the inode state change.

I also postpone removing inodes from i_sb_list. It is not compulsory to
do so but I do it for readability reasons. Inodes are added/removed to
the lists together everywhere in the code and there is no point to
change this rule. This is harmless because the only user of i_sb_list
which somehow may interfere with me (invalidate_list()) is excluded by
the iprune_sem mutex.

The same race is possible in invalidate_list() so I do the same for it.

Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] __wait_on_freeing_inode fix
Miklos Szeredi [Tue, 12 Jul 2005 20:58:10 +0000 (13:58 -0700)] 
[PATCH] __wait_on_freeing_inode fix

This patch fixes queer behavior in __wait_on_freeing_inode().

If I_LOCK was not set it called yield(), effectively busy waiting for the
removal of the inode from the hash.  This change was introduced within
"[PATCH] eliminate inode waitqueue hashtable" Changeset 1.1938.166.16 last
october by wli.

The solution is to restore the old behavior, of unconditionally waiting on
the waitqueue.  It doesn't matter if I_LOCK is not set initally, the task
will go to sleep, and wake up when wake_up_inode() is called from
generic_delete_inode() after removing the inode from the hash chain.

Comment is also updated to better reflect current behavior.

This condition is very hard to trigger normally (simultaneous clear_inode()
with iget()) so probably only heavy stress testing can reveal any change of
behavior.

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] lower VM_DONTCOPY total_vm
Hugh Dickins [Tue, 12 Jul 2005 20:58:09 +0000 (13:58 -0700)] 
[PATCH] lower VM_DONTCOPY total_vm

dup_mmap of a VM_DONTCOPY vma forgot to lower the child's total_vm.  (But
no way does this account for the recent report of total_vm seen too low.)

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] quiet ide-cd warning
Matt Mackall [Tue, 12 Jul 2005 20:58:09 +0000 (13:58 -0700)] 
[PATCH] quiet ide-cd warning

This shuts up a potential uninitialized variable warning.

Signed-off-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] aacraid: swapped kmalloc args.
Dave Jones [Tue, 12 Jul 2005 20:58:08 +0000 (13:58 -0700)] 
[PATCH] aacraid: swapped kmalloc args.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] name_to_dev_t warning fix
Andrew Morton [Tue, 12 Jul 2005 20:58:07 +0000 (13:58 -0700)] 
[PATCH] name_to_dev_t warning fix

kernel/power/disk.c needs a declaration of name_to_dev_t() in scope.  mount.h
seems like an appropriate choice.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge with rsync://fileserver/linux
Todd Poynor [Tue, 12 Jul 2005 02:34:39 +0000 (03:34 +0100)] 
Merge with rsync://fileserver/linux

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/i2c-2.6
Linus Torvalds [Tue, 12 Jul 2005 22:54:36 +0000 (15:54 -0700)] 
Merge /pub/scm/linux/kernel/git/gregkh/i2c-2.6

19 years ago[JFFS2] Avoid compiler warnings when JFFS2_FS_WRITEBUFFER=n
Todd Poynor [Tue, 12 Jul 2005 02:34:39 +0000 (03:34 +0100)] 
[JFFS2] Avoid compiler warnings when JFFS2_FS_WRITEBUFFER=n

Signed-off-by: Todd Poynor <tpoynor@mvista.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years ago[JFFS2] Init locks early during mount
Artem B. Bityuckiy [Tue, 12 Jul 2005 16:37:12 +0000 (17:37 +0100)] 
[JFFS2] Init locks early during mount

In case of a mount error locks might be uninitialized but
accessed by the resulting call to jffs2_kill_sb().

Signed-off-by: Artem B. Bityuckiy <dedekind@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years ago[JFFS2] Rename function and update comments
Artem B. Bityuckiy [Sun, 10 Jul 2005 15:15:36 +0000 (16:15 +0100)] 
[JFFS2] Rename function and update comments

We recently changed the method of collecting and sorting of
tmp_dnode objects to use a temporary RB-tree instead of a
temporary list. Rename function and update comments.

Signed-off-by: Artem B. Bityuckiy <dedekind@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years ago[JFFS2] Remove needless variable initialization
Artem B. Bityuckiy [Sun, 10 Jul 2005 13:13:58 +0000 (14:13 +0100)] 
[JFFS2] Remove needless variable initialization

Signed-off-by: Artem B. Bityuckiy <dedekind@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years ago[JFFS2] Avoid alloc/dealloc for zero sized nodes
Artem B. Bityuckiy [Thu, 7 Jul 2005 15:45:32 +0000 (16:45 +0100)] 
[JFFS2] Avoid alloc/dealloc for zero sized nodes

Signed-off-by: Artem B. Bityuckiy <dedekind@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
19 years ago[ACPI] merge acpi-2.6.12 branch into latest Linux 2.6.13-rc...
Len Brown [Tue, 12 Jul 2005 21:21:56 +0000 (17:21 -0400)] 
[ACPI] merge acpi-2.6.12 branch into latest Linux 2.6.13-rc...

Signed-off-by: Len Brown <len.brown@intel.com>
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Tue, 12 Jul 2005 20:17:42 +0000 (13:17 -0700)] 
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 12 Jul 2005 20:16:40 +0000 (13:16 -0700)] 
Merge /pub/scm/linux/kernel/git/davem/net-2.6

19 years ago[VLAN]: Fix early vlan adding leads to not functional device
Tommy Christensen [Tue, 12 Jul 2005 19:13:49 +0000 (12:13 -0700)] 
[VLAN]: Fix early vlan adding leads to not functional device

OK, I can see what's happening here. eth0 doesn't detect link-up until
after a few seconds, so when the vlan interface is opened immediately
after eth0 has been opened, it inherits the link-down state. Subsequently
the vlan interface is never properly activated and are thus unable to
transmit any packets.

dev->state bits are not supposed to be manipulated directly. Something
similar is probably needed for the netif_device_present() bit, although
I don't know how this is meant to work for a virtual device.

Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SPARC64]: Fix SMP build failure.
Andrew Morton [Tue, 12 Jul 2005 19:09:43 +0000 (12:09 -0700)] 
[SPARC64]: Fix SMP build failure.

arch/sparc64/kernel/smp.c:48: error: parse error before "__attribute__"
arch/sparc64/kernel/smp.c:49: error: parse error before "__attribute__"

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[NET]: __be'ify *_type_trans()
Alexey Dobriyan [Tue, 12 Jul 2005 19:08:43 +0000 (12:08 -0700)] 
[NET]: __be'ify *_type_trans()

tr_type_trans(), hippi_type_trans() left as-is.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[NETFILTER]: Revert nf_reset change
Phil Oester [Tue, 12 Jul 2005 18:57:52 +0000 (11:57 -0700)] 
[NETFILTER]: Revert nf_reset change

Revert the nf_reset change that caused so much trouble, drop conntrack
references manually before packets are queued to packet sockets.

Signed-off-by: Phil Oester <kernel@linuxace.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] USB: add ldusb driver
Michael Hund [Mon, 27 Jun 2005 20:44:22 +0000 (22:44 +0200)] 
[PATCH] USB: add ldusb driver

The following driver provides complete interrupt-in and interrupt-out
reports (raw data) to a user program. Until now it uses the
HIDIOCGDEVINFO ioctl call, because I don't know better :-(. Perhaps, it
will be ok for you - and I will be happy, if you assign 8 minor numbers.

I have tested it in several environments and it works very well for me.
However, it has a problem with two or more devices at the same hub, if
the two or more devices need 1 ms interrupt-in transfers. Unfortunately
more than one interrupt-in transfer every ms isn't possible (ehci
driver?). This is why the min_interrupt_in_interval and
min_interrupt_out_interval are increased to 2 ms (see the corresponding
module parameters). This way, I can use two devices simultaneously at
the same hub.

Signed-off-by: Michael Hund <mhund@ld-didactic.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: fix usb reference count bug in cdc-acm driver
brian@murphy.dk [Wed, 29 Jun 2005 23:53:29 +0000 (16:53 -0700)] 
[PATCH] USB: fix usb reference count bug in cdc-acm driver

This increases the reference count on the usb cdc acm control interface
which is referred to by the tty interface provided by the driver. This
allows the deferred removal of the tty after the physical device is
disconnected if the tty is held open at the time of disconnection.

Signed-off-by: brian@murphy.dk
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: export usb_get_intf() and usb_put_intf()
brian@murphy.dk [Wed, 29 Jun 2005 23:53:29 +0000 (16:53 -0700)] 
[PATCH] USB: export usb_get_intf() and usb_put_intf()

Export usb_get_intf and usb_put_intf so that modules can increase
usb interface reference counts.

Signed-off-by: brian@murphy.dk
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: another cdc descriptor
david-b@pacbell.net [Wed, 29 Jun 2005 14:04:14 +0000 (07:04 -0700)] 
[PATCH] USB: another cdc descriptor

This adds another CDC descriptor type to <linux/usb_cdc.h>; the main claim
to fame for this is that some Motorola phones include it.  It's not currently
needed by any driver code; included for completeness.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: fix ohci merge glitch
david-b@pacbell.net [Wed, 29 Jun 2005 14:03:10 +0000 (07:03 -0700)] 
[PATCH] USB: fix ohci merge glitch

A patch re-organizing some parts of root hub initialization deleted the
code initializing the bus-neutral reboot/shutdown notifier for OHCI.
This patch just restores that deleted code.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: ohci-omap pm updates
david-b@pacbell.net [Wed, 29 Jun 2005 13:59:14 +0000 (06:59 -0700)] 
[PATCH] USB: ohci-omap pm updates

The recent "pm_message_t" changes removed functionality from the Linux
PM framework.  This patch removes it from the OMAP OHCI too, removing
the distinction between (previous) PM_SUSPEND_MEM and PM_SUSPEND_DISK
state transitions ... now the only suspend semantics supportable are
what was previously PM_SUSPEND_DISK (4) and is now "PMSG_SUSPEND" (3).

From: Todd Poynor <tpoynor@mvista.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>