opensuse:kernel.git
15 years ago- fix mapped-pte usage - do not access it after unmapping.
Ingo Molnar [Tue, 19 Feb 2002 22:42:52 +0000 (23:42 +0100)]
- fix mapped-pte usage - do not access it after unmapping.

15 years ago- cleanup from Andi Kleen: schedule_tail() does not have to take 'prev' as a paramete...
Ingo Molnar [Tue, 19 Feb 2002 17:46:39 +0000 (18:46 +0100)]
- cleanup from Andi Kleen: schedule_tail() does not have to take 'prev' as a parameter anymore.

15 years agoadds simple support for atomically-mapped PTEs. On highmem systems this enables the...
Ingo Molnar [Tue, 19 Feb 2002 17:27:40 +0000 (18:27 +0100)]
adds simple support for atomically-mapped PTEs. On highmem systems this enables the allocation of the pagetables in highmem.

15 years ago- the new vmalloc_to_page() interface should be used to determine the physical page...
Ingo Molnar [Tue, 19 Feb 2002 17:10:45 +0000 (18:10 +0100)]
- the new vmalloc_to_page() interface should be used to determine the physical page a given vmalloc() area virtual address is mapped to.

15 years agoMerge k:/home/torvalds/BK/linux-2.5/
Ingo Molnar [Tue, 19 Feb 2002 16:59:30 +0000 (17:59 +0100)]
Merge k:/home/torvalds/BK/linux-2.5/
into elte.hu:/home/mingo/BK/mine/linux-2.5

15 years agoallow the boot-time sizing of the highmem pool.
Ingo Molnar [Mon, 18 Feb 2002 21:21:32 +0000 (22:21 +0100)]
allow the boot-time sizing of the highmem pool.

15 years agoMerge master.kernel.org:/home/mochel/BK/linux-2.5
Linus Torvalds [Mon, 18 Feb 2002 01:11:44 +0000 (17:11 -0800)]
Merge master.kernel.org:/home/mochel/BK/linux-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] meye driver update
Stelian Pop [Mon, 18 Feb 2002 01:10:34 +0000 (17:10 -0800)]
[PATCH] meye driver update

This patch (+ BK changeset) converts the meye driver to the new DMA API,
this is necessary for the driver to be used in 2.5.

Stelian.

ChangeSet@1.332, 2002-02-15 16:35:31+01:00, stelian@popies.net
  Convert to the new DMA API and allocate separate DMA pages instead of one big buffer.

15 years agoMerge home.transmeta.com:/home/torvalds/v2.5/dalecki
Linus Torvalds [Mon, 18 Feb 2002 01:09:55 +0000 (17:09 -0800)]
Merge home.transmeta.com:/home/torvalds/v2.5/dalecki
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] PATCH 2.5.5-pre1 dead arrays.
Martin Dalecki [Mon, 18 Feb 2002 01:08:27 +0000 (17:08 -0800)]
[PATCH] PATCH 2.5.5-pre1 dead arrays.

Just the usual removal of the dead global arrays and associated cruft.
(Thistime not affecting lvm, which BTW. doesn't compile currently anyway
;-).

15 years ago[PATCH] Re: IDE cleanup for 2.5.4-pre3
Martin Dalecki [Mon, 18 Feb 2002 01:05:04 +0000 (17:05 -0800)]
[PATCH] Re: IDE cleanup for 2.5.4-pre3

The end_request() function familiy (not the global one, but the IDE
specific ones), did bear a permuted parameter ordering.  After fixing
this it turned out that at all places the huk parameter wasn't the
hwgroup, but just the drive in question itself.  I have changed this to
be more sane, which allowed to remove many unneccessary code
duplication, or rather obfuscation, in between the __ide_end_request()
and ide_end_request() functions.  This simplification is actually the
"spreading" part of the game.

15 years agoMerge home.transmeta.com:/home/torvalds/v2.5/morton
Linus Torvalds [Mon, 18 Feb 2002 01:02:13 +0000 (17:02 -0800)]
Merge home.transmeta.com:/home/torvalds/v2.5/morton
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] ENOSPC correctness
Andrew Morton [Mon, 18 Feb 2002 01:01:56 +0000 (17:01 -0800)]
[PATCH] ENOSPC correctness

A forward-port.  This is the code which prevents ENOSPC
errors from exposing stale data within filesystems.

- in generic_file_write(), if prepare_write() fails, truncate
  the file to drop any part-added blocks.

- in __block_write_full_page(), if we hit an error,  push
  whatever buffers we _have_ mapped into the file out to disk.

- in __block_prepare_write(), if we hit an error, zero out
  any blocks which we did manage to map into the file.  This
  is because the caller won't be doing any writing to those
  blocks due to the error.

15 years ago[PATCH] msync correctness
Andrew Morton [Mon, 18 Feb 2002 01:01:52 +0000 (17:01 -0800)]
[PATCH] msync correctness

A forward port.  At present, msync() does not report errors
from EIO or ENOSPC.  fsync() has the same bug for mapped pages
against the affected fd.

The patch correctly propagates these errors back up from
writepage so that fsync and msync correctly report errors.

It's fairly important - msync is the only way we have
of reporting ENOSPC against sparse mappings.

Of course, you can still silently lose your data if it's kswapd who
gets ENOSPC during writepage.  I have 3/4 of a patch for that.  It
records the data loss so that a later msync() will report the bad
news.

This patch also adds an implementation of msync(MS_ASYNC), because
it was easy.

15 years ago[PATCH] IS_SYNC diretory handling
Andrew Morton [Mon, 18 Feb 2002 01:01:48 +0000 (17:01 -0800)]
[PATCH] IS_SYNC diretory handling

A forward-port.  ext2, minix and sysv aren't handling directories
correctly when IS_SYNC is in place.  They call waitfor_one_page(),
but forgot to start the I/O.

The patch also moves waitfor_one_page and writeout_one_page
into fs/buffer.c, so mm/filemap.c now does not mention buffer_head
at all.

15 years ago[PATCH] BUG register preservation
Andrew Morton [Mon, 18 Feb 2002 00:44:37 +0000 (16:44 -0800)]
[PATCH] BUG register preservation

Here's the x86 BUG() implementation we discussed the other day.

I also have the rework of the header files which avoids
instantiation of strings in headers and saves 100-200k.
However that is only needed for gcc 2.X.  I assume that
by the time 2.6 is in use, gcc-3.x will be the preferred
compiler.

15 years agoMerge home.transmeta.com:/home/torvalds/v2.5/knfsd
Linus Torvalds [Mon, 18 Feb 2002 00:43:08 +0000 (16:43 -0800)]
Merge home.transmeta.com:/home/torvalds/v2.5/knfsd
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] PATCH 7/7: knfsd cleanups - module initialisation
Neil Brown [Mon, 18 Feb 2002 00:41:08 +0000 (16:41 -0800)]
[PATCH] PATCH 7/7: knfsd cleanups - module initialisation

Tidyup init/exit for nfsd module

move nfsd_init into an initcall with other module
startup.  This means that "initialized" isn't needed for
any of the files that use it, as the bits are always initialised if
in use.

15 years ago[PATCH] PATCH 6/7: knfsd cleanups - syscall cleanup
Neil Brown [Mon, 18 Feb 2002 00:41:04 +0000 (16:41 -0800)]
[PATCH] PATCH 6/7: knfsd cleanups - syscall cleanup

Cleanup the syscall interface to nfsd

1/ add an "owner" field to the nfsd_linkage structure
2/ grab a reference to that module before calling the syscall
3/ Remove the reference counting from inside the module
4/ Always allow nfsd module to be called, even if compile with
CONFIG_NFSD == N
   ( but not if CONFIG_MODULES also == N)

15 years ago[PATCH] PATCH 5/7: knfsd cleanups - mkdev
Neil Brown [Mon, 18 Feb 2002 00:40:10 +0000 (16:40 -0800)]
[PATCH] PATCH 5/7: knfsd cleanups - mkdev

Use MKDEV for making device number from components

This patch is thanks to GOTO Masanori <gotom@debian.or.jp>

15 years ago[PATCH] PATCH 4/7: knfsd cleanups - vfs.c improvements
Neil Brown [Mon, 18 Feb 2002 00:40:06 +0000 (16:40 -0800)]
[PATCH] PATCH 4/7: knfsd cleanups - vfs.c improvements

Tidy up some vfs calls in nfsd

1/ changes to sys_fsync had not also been made to nfsd_sync.
   Now nfsd_sync calls filemap_fdatasync and filemap_fdatawait.
2/ change nfsd_readdir to use vfs_readdir instead of i_op->readdir

15 years ago[PATCH] PATCH 3/7: knfsd cleanups - incorrect use of inode_change_ok
Neil Brown [Mon, 18 Feb 2002 00:40:02 +0000 (16:40 -0800)]
[PATCH] PATCH 3/7: knfsd cleanups - incorrect use of inode_change_ok

Get nfsd_setattr to not put too much weight on inode_change_ok

nfsd_currently calls inode_change_ok and does not try setattr if this fails.
However this is wrong.  If a filesystem defines it's own i_op->setattr, then
it might use a completely different mechanisim for determining what is ok.
nfsd shouldn't assume...

We still use inode_change_ok when normalising NFSv2 "touch" requests, but
only in passing.

15 years ago[PATCH] PATCH 2/7: knfsd cleanup - FAT shouldn't return NULL
Neil Brown [Mon, 18 Feb 2002 00:39:58 +0000 (16:39 -0800)]
[PATCH] PATCH 2/7: knfsd cleanup - FAT shouldn't return NULL

Stop fat_fh_to_dentry returning NULL

the fh_to_dentry routines should never return NULL.
The caller expects an ERR_PTR or a valid (possibly negative)
dentry.  fat did the wrong thing and so could oops.

15 years ago[PATCH] PATCH 1/7: knfsd cleanups - big fixes
Neil Brown [Mon, 18 Feb 2002 00:39:53 +0000 (16:39 -0800)]
[PATCH] PATCH 1/7:  knfsd cleanups - big fixes

Fix bugs recently introduced into kNFSd

When searching a list.h list, we cannot export
to find NULL at the end.  Instead we return a pointer
when found, or NULL if nothing found.  Same bug, 4 times.

The seq_file improvements to /proc/fs/nfs/exports got the counting
wrong so that some clients would get reported twice, always the last.

15 years ago[PATCH] dnotify race fix
Alexander Viro [Mon, 18 Feb 2002 00:35:06 +0000 (16:35 -0800)]
[PATCH] dnotify race fix

A bunch of places dereferences ->d_parent->d_inode with no
protection whatsoever (e.g. on return from read()).  It's an
SMP race on all boxen and pretty wide UP race if we have dnotify
set on parent (race between read() and rename() and similar beasts).

Patch below is the first one in a series of ->d_parent-related
fixes.  It adds a helper (dnotify_parent(dentry, event)) and converts
places that did inode_dir_notify(dentry->d_parent->d_inode,...) to it.

Please, apply.  Notice that problem exists in 2.4 and unlike 2.5
there we can't switch to saner API (basically, reporting file events on
file, not on a parent directory).

Some of the further fixes depend on Pat's and Greg's (driverfs and usbdevfs
resp.) patches, so getting them merged would make life easier.  And yes,
some of further chunks (e.g. smbfs ->revalidate() and friends) will also
have to go into 2.4 - they are independent from any threading projects ;-/

15 years ago[PATCH] BKL shifting - ->rename()
Alexander Viro [Mon, 18 Feb 2002 00:35:02 +0000 (16:35 -0800)]
[PATCH] BKL shifting - ->rename()

15 years ago[PATCH] Re: [PATCH] Re: 2.5: further llseek cleanup (3/3)
Robert Love [Mon, 18 Feb 2002 00:32:49 +0000 (16:32 -0800)]
[PATCH] Re: [PATCH] Re: 2.5: further llseek cleanup (3/3)

Ugh, another one.  Linus, please apply.

Robert Love

15 years ago[PATCH] Re: 2.5: further llseek cleanup (3/3)
Robert Love [Mon, 18 Feb 2002 00:32:45 +0000 (16:32 -0800)]
[PATCH] Re: 2.5: further llseek cleanup (3/3)

15 years agoFix from Mike Kravetz: use the nr_running return value
Ingo Molnar [Fri, 15 Feb 2002 13:14:37 +0000 (14:14 +0100)]
Fix from Mike Kravetz: use the nr_running return value
of double_lock_balance().

15 years agoMerge hera.kernel.org:/home/torvalds/BK/linux-2.5
Patrick Mochel [Fri, 15 Feb 2002 03:37:22 +0000 (19:37 -0800)]
Merge hera.kernel.org:/home/torvalds/BK/linux-2.5
into hera.kernel.org:/home/mochel/BK/linux-2.5

15 years agoFix driverfs deadlock on file/directory removal
Patrick Mochel [Fri, 15 Feb 2002 03:31:53 +0000 (19:31 -0800)]
Fix driverfs deadlock on file/directory removal

15 years agoFix sound config files (makes xconfig happy)
Linus Torvalds [Thu, 14 Feb 2002 09:55:22 +0000 (01:55 -0800)]
Fix sound config files (makes xconfig happy)

15 years ago[PATCH] (5/5) more BKL shifting
Alexander Viro [Thu, 14 Feb 2002 09:46:47 +0000 (01:46 -0800)]
[PATCH] (5/5) more BKL shifting

old_inode is locked by vfs_link().

15 years ago[PATCH] (4/5) more BKL shifting
Alexander Viro [Thu, 14 Feb 2002 09:46:43 +0000 (01:46 -0800)]
[PATCH] (4/5) more BKL shifting

BKL shifted into ->symlink() (+ fixes for {ramfs,shmem}_symlink())

15 years ago[PATCH] (3/5) more BKL shifting
Alexander Viro [Thu, 14 Feb 2002 09:46:38 +0000 (01:46 -0800)]
[PATCH] (3/5) more BKL shifting

BKL shifted into ->link(), check for S_ISDIR moved into vfs_link().

15 years ago[PATCH] (2/5) more BKL shifting
Alexander Viro [Thu, 14 Feb 2002 09:46:33 +0000 (01:46 -0800)]
[PATCH] (2/5) more BKL shifting

BKL shifted into ->rmdir()

15 years ago[PATCH] (1/5) more BKL shifting
Alexander Viro [Thu, 14 Feb 2002 09:46:28 +0000 (01:46 -0800)]
[PATCH] (1/5) more BKL shifting

BKL shifted into ->mkdir()

15 years ago[PATCH] pid allocator bugfix
J. A. Magallon [Thu, 14 Feb 2002 09:33:14 +0000 (01:33 -0800)]
[PATCH] pid allocator bugfix

This patch fixes a bug in the Linux process ID allocator.  It isn't quite
SMP safe since it references "last_pid" after releasing the lock protecting
it.  This can result in two processes getting assigned the same process ID.

15 years ago[PATCH] NeoMagic FPU fix (2.5.5-pre1)
Denis Oliver Kropp [Thu, 14 Feb 2002 09:30:48 +0000 (01:30 -0800)]
[PATCH] NeoMagic FPU fix (2.5.5-pre1)

this is the NeoMagic floating foint removal patch
I posted previously for Linux 2.4.18-pre9-ac3.

Best regards,
  Denis Oliver Kropp

15 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.isdn
Linus Torvalds [Thu, 14 Feb 2002 07:09:39 +0000 (23:09 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.isdn
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

15 years agoMerge linux-isdn@linux-isdn.bkbits.net:linux-2.5.isdn
Kai Germaschewski [Thu, 14 Feb 2002 00:55:44 +0000 (01:55 +0100)]
Merge linux-isdn@linux-isdn.bkbits.net:linux-2.5.isdn
into tp1.ruhr-uni-bochum.de:/home/kai/kernel/linux-2.5.isdn

15 years ago[PATCH] xattr updates (minor, 4/4)
Nathan Scott [Wed, 13 Feb 2002 13:45:51 +0000 (05:45 -0800)]
[PATCH] xattr updates (minor, 4/4)

This final patch is a simple documentation fixup patch filling
in the blanks for the extended attribute interfaces; and fixes
a typo in the same spot: "permission".

Nathan

15 years ago[PATCH] xattr updates (minor, 3/4)
Nathan Scott [Wed, 13 Feb 2002 13:45:45 +0000 (05:45 -0800)]
[PATCH] xattr updates (minor, 3/4)

This next incremental patch tidies up the data types passed back
from the `list' and `get' extended attribute syscalls - these now
match the design (ie. using ssize_t) rather than simply using long
or int everywhere; also now use const types in the VFS interface,
where appropriate.

Nathan

15 years ago[PATCH] xattr updates (minor, 2/4)
Nathan Scott [Wed, 13 Feb 2002 13:45:40 +0000 (05:45 -0800)]
[PATCH] xattr updates (minor, 2/4)

This incremental patch fixes copying out of an extended attribute
value or name list.  Previously we copied out the entire buffer
passed in from userspace, now we only copy out the size which the
underlying filesystem tells us to (ie. we will no longer include
potentialy-initialised data as well).

Nathan

15 years ago[PATCH] xattr updates (minor, 1/4)
Nathan Scott [Wed, 13 Feb 2002 13:45:35 +0000 (05:45 -0800)]
[PATCH] xattr updates (minor, 1/4)

This first patch fixes the copying in of extended attribute
names from userspace in the extended attribute syscalls (a
problem found when using electric fence on the user tools).

Nathan

15 years ago[PATCH] alsa + x86-64
Peter Samuelson [Wed, 13 Feb 2002 13:43:33 +0000 (05:43 -0800)]
[PATCH] alsa + x86-64

Obvious patch to make x86_64 pick up new sound location

15 years agoMerge bk://linuxusb.bkbits.net/linus-2.5
Linus Torvalds [Wed, 13 Feb 2002 13:41:26 +0000 (05:41 -0800)]
Merge bk://linuxusb.bkbits.net/linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agoMerge master.kernel.org:/home/mochel/BK/linux-2.5
Linus Torvalds [Wed, 13 Feb 2002 13:38:52 +0000 (05:38 -0800)]
Merge master.kernel.org:/home/mochel/BK/linux-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agomem_flags added to usb_submit_urb().
Greg Kroah-Hartman [Wed, 13 Feb 2002 09:39:56 +0000 (01:39 -0800)]
mem_flags added to usb_submit_urb().
This fixes usb drivers outside of the drivers/usb directory.

15 years agoadded mem_flags to usb_submit_urb().
Greg Kroah-Hartman [Wed, 13 Feb 2002 09:38:41 +0000 (01:38 -0800)]
added mem_flags to usb_submit_urb().
Fixes drivers/usb/storage.
Patch done by Oliver Neukum.

15 years agoadded mem_flags to usb_submit_urb().
Greg Kroah-Hartman [Wed, 13 Feb 2002 09:37:10 +0000 (01:37 -0800)]
added mem_flags to usb_submit_urb().
This fixes the drivers in drivers/usb/serial.
Patch done by Oliver Neukum.

15 years agoAdded mem_flags to usb_submit_urb().
Greg Kroah-Hartman [Wed, 13 Feb 2002 09:36:19 +0000 (01:36 -0800)]
Added mem_flags to usb_submit_urb().
This modifies the drivers in drivers/usb.
Patch done by Oliver Neukum.

15 years agoAdded mem_flags to usb_submit_urb(), this is the USB core code changes.
Greg Kroah-Hartman [Wed, 13 Feb 2002 09:34:58 +0000 (01:34 -0800)]
Added mem_flags to usb_submit_urb(), this is the USB core code changes.

15 years agousb vicam driver:
Oliver Neukum [Wed, 13 Feb 2002 09:18:52 +0000 (01:18 -0800)]
usb vicam driver:
- fix for memory leak.

15 years agousb hpusbscsi driver fixes:
Oliver Neukum [Wed, 13 Feb 2002 09:15:28 +0000 (01:15 -0800)]
usb hpusbscsi driver fixes:
- special case for REQUEST_SENSE
- reset handling won't work properly -> disabled
- error reporting corrected

15 years ago[PATCH] USB OHCI powerbook fix (v2.5.4)
Paul Mackerras [Wed, 13 Feb 2002 09:12:38 +0000 (01:12 -0800)]
[PATCH] USB OHCI powerbook fix (v2.5.4)

The patch below fixes a compile problem in the USB OHCI HCD driver on
powerbooks, namely that the ohci_hcd structure doesn't have an irq
member.

Paul.

15 years agodevice model/driverfs updates
Patrick Mochel [Wed, 13 Feb 2002 07:27:07 +0000 (23:27 -0800)]
device model/driverfs updates

15 years agoupdate version
Linus Torvalds [Wed, 13 Feb 2002 04:53:55 +0000 (20:53 -0800)]
update version

15 years ago[PATCH] BKL shifted inside ->unlink()
Alexander Viro [Wed, 13 Feb 2002 04:31:22 +0000 (20:31 -0800)]
[PATCH] BKL shifted inside ->unlink()

Next one in the series - this time it's ->unlink()...

15 years ago[PATCH] misc fixes
Alexander Viro [Wed, 13 Feb 2002 04:31:17 +0000 (20:31 -0800)]
[PATCH] misc fixes

1) proc/<pid>/fd readdir() has broken locking - sometimes it ends up with
task->files->file_lock locked.

2) missing variable definition in arch/s390x/kernel/linux32.c::cp_new_stat32().

Please, apply - the first one is really serious.

15 years ago[PATCH] BKL shifted into ->mknod()
Alexander Viro [Wed, 13 Feb 2002 04:31:13 +0000 (20:31 -0800)]
[PATCH] BKL shifted into ->mknod()

15 years ago[PATCH] SMP-safe ext2/namei.c
Alexander Viro [Wed, 13 Feb 2002 04:31:09 +0000 (20:31 -0800)]
[PATCH] SMP-safe ext2/namei.c

BKL held around ->i_nlink changes in ext2/namei.c.  That makes
namei.c SMP-safe (dir.c already is).  BKL dropped around ext2_free_inode() -
it's already SMP-safe (the only place that needed BKL was DQUOT_FREE_INODE()
and it got BKL in the previous patch).

15 years ago[PATCH] BKL shifted into ->create()
Alexander Viro [Wed, 13 Feb 2002 04:31:05 +0000 (20:31 -0800)]
[PATCH] BKL shifted into ->create()

BKL shifted into ->create().  lock_kernel()/unlock_kernel() added
in dquot_{alloc,free}_inode() - that makes {ext2,minix,sysv,ufs}_read_inode()
SMP-safe.

15 years ago[PATCH] ext2/inode.c cleanup.
Alexander Viro [Wed, 13 Feb 2002 04:31:00 +0000 (20:31 -0800)]
[PATCH] ext2/inode.c cleanup.

ext2_read_inode() and ext2_update_inode() share large chunk
of code - getting a buffer_head with raw inode in it.
Moved into a helper function (ext2_get_inode()), the rest
of ext2_{read,update}_inode() slightly cleaned up.

15 years ago[PATCH] x86_64 merge: fs/proc/inode.c #include fix
Andi Kleen [Wed, 13 Feb 2002 04:17:43 +0000 (20:17 -0800)]
[PATCH] x86_64 merge: fs/proc/inode.c #include fix

fs/proc/inode.c is using __init, but for some reason missing an
#include <linux/init.h>. Add this.

15 years ago[PATCH] x86-64 MAINTAINERS
Andi Kleen [Wed, 13 Feb 2002 04:17:39 +0000 (20:17 -0800)]
[PATCH] x86-64 MAINTAINERS

Add Andi Kleen as x86-64 maintainer.

15 years ago[PATCH] x86_64 merge: arch + asm
Andi Kleen [Wed, 13 Feb 2002 04:17:35 +0000 (20:17 -0800)]
[PATCH] x86_64 merge: arch + asm

This adds the x86_64 arch and asm directories and a Documentation/x86_64.

It took a bit longer because I first had to make preemption and thread_info
work and also found some other bugs while doing this. The port has been
tested for a long time on UP.

I'm not sure what I should describe.  A lot is based on i386 with
a lot of cleanups. I wrote a paper about it for last year's OLS that describes
most of the changes (ftp://ftp.firstfloor.org/pub/ak/x86_64.ps.gz). It is
a bit outdated now, but should give a good overview.

It currently has a completely cut'n'pasted from others+hacked 32bit
emulation. I hope to clean that up in the future by merging the generic
core of this with other 64bit archs.

Thanks,
-Andi

15 years ago[PATCH] x86_64-merge file.c warning
Andi Kleen [Wed, 13 Feb 2002 04:17:29 +0000 (20:17 -0800)]
[PATCH] x86_64-merge file.c warning

Just an gcc 3.1 warning fix. It now warns about __FUNCTION__ string
concatenation. Also remove the check because it does not seem to trigger
ever.

-Andi

15 years agoAvoid pci driver warnings on 64-bit hosts
Linus Torvalds [Wed, 13 Feb 2002 04:09:34 +0000 (20:09 -0800)]
Avoid pci driver warnings on 64-bit hosts

15 years agoupdate version
Linus Torvalds [Wed, 13 Feb 2002 03:34:48 +0000 (19:34 -0800)]
update version

15 years agoMerge http://gkernel.bkbits.net/misc-2.5
Linus Torvalds [Wed, 13 Feb 2002 05:02:27 +0000 (21:02 -0800)]
Merge http://gkernel.bkbits.net/misc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agoInclude linux/compiler.h in include/asm-i386/bitops.h,
Jeff Garzik [Wed, 13 Feb 2002 09:05:47 +0000 (04:05 -0500)]
Include linux/compiler.h in include/asm-i386/bitops.h,
for the definition of unlikely().

15 years ago[PATCH] flush_icache_user_range (v2.5.4)
Paul Mackerras [Wed, 13 Feb 2002 04:27:37 +0000 (20:27 -0800)]
[PATCH] flush_icache_user_range (v2.5.4)

The patch below changes access_process_vm to use a new architecture
hook, flush_icache_user_range, instead of flush_icache_page, and adds
a definition of flush_icache_user_range which does the same thing as
flush_icache_page for all architectures except PPC.  (The PPC update
that is in Linus' BK tree already includes a suitable definition of
flush_icache_user_range.)

The reason for doing this is that when flush_icache_page is called
from do_no_page or do_swap_page, I want to be able to do the flush
conditionally, based on the state of the page.  In contrast,
access_process_vm needs to do the flush unconditionally since it has
just modified the page.  In the access_process_vm case it is useful to
have the information about the user address and length that have been
modified since then we can just flush the affected cache lines rather
than the whole page.

This patch should make it easy to improve performance on alpha, since
there (as I understand it) the icache flush is not needed at all in
do_no_page or do_swap_page, but is needed in access_process_vm.  All
that is needed is to make flush_icache_page a noop on alpha.  The
patch below doesn't do this, I'll let the alpha maintainers push that
change if they want.

15 years ago[PATCH] fix sd_find_target (v2.5.4)
Paul Mackerras [Wed, 13 Feb 2002 04:27:31 +0000 (20:27 -0800)]
[PATCH] fix sd_find_target (v2.5.4)

This patch fixes a compile error on PPC.  It's in sd_find_target, a
function that returns a kdev_t.

15 years agoMerge bk://linuxusb.bkbits.net/linus-2.5
Linus Torvalds [Wed, 13 Feb 2002 04:24:15 +0000 (20:24 -0800)]
Merge bk://linuxusb.bkbits.net/linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] menuconfig: fix error exit if awk fails
Michael Elizabeth Chastain [Wed, 13 Feb 2002 04:23:42 +0000 (20:23 -0800)]
[PATCH] menuconfig: fix error exit if awk fails

This one-liner fixes an error case in Menuconfig when awk fails.
Written by Andrew Church (achurch@achurch.org).
Reviewed and tested by Michael Elizabeth Chastain (mec@shout.net).

Michael Elizabeth Chastain

===

15 years agoMake Jaroslav the sound maintainer, remove Alan on his request.
Linus Torvalds [Wed, 13 Feb 2002 04:05:43 +0000 (20:05 -0800)]
Make Jaroslav the sound maintainer, remove Alan on his request.

15 years ago[PATCH] 2.5.4, add help texts to drivers/net/pcmcia/Config.help
Steven Cole [Wed, 13 Feb 2002 03:58:48 +0000 (19:58 -0800)]
[PATCH] 2.5.4, add help texts to drivers/net/pcmcia/Config.help

Add help texts for CONFIG_PCMCIA_AXNET and CONFIG_PCMCIA_XIRCOM

15 years agoMerge http://gkernel.bkbits.net/net-drivers-2.5
Linus Torvalds [Wed, 13 Feb 2002 03:57:41 +0000 (19:57 -0800)]
Merge http://gkernel.bkbits.net/net-drivers-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agoMerge http://gkernel.bkbits.net/misc-2.5
Linus Torvalds [Wed, 13 Feb 2002 03:56:50 +0000 (19:56 -0800)]
Merge http://gkernel.bkbits.net/misc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agoMerge home.transmeta.com:/home/torvalds/v2.5/alsa
Linus Torvalds [Wed, 13 Feb 2002 03:34:08 +0000 (19:34 -0800)]
Merge home.transmeta.com:/home/torvalds/v2.5/alsa
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years ago[PATCH] ALSA patch for 2.5.4
Jaroslav Kysela [Wed, 13 Feb 2002 03:32:11 +0000 (19:32 -0800)]
[PATCH] ALSA patch for 2.5.4

Integrate ALSA into v2.5.4

            Jaroslav

15 years agoMerge master.kernel.org:BK/linux-2.5
Linus Torvalds [Wed, 13 Feb 2002 02:57:46 +0000 (18:57 -0800)]
Merge master.kernel.org:BK/linux-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

15 years agocleanups.
Ingo Molnar [Wed, 13 Feb 2002 20:32:25 +0000 (21:32 +0100)]
cleanups.

15 years ago- make the preempt-enable test cheaper - only test for the (very rare) TIF_NEED_RESCHED
Ingo Molnar [Wed, 13 Feb 2002 15:01:58 +0000 (16:01 +0100)]
- make the preempt-enable test cheaper - only test for the (very rare) TIF_NEED_RESCHED
  condition, we test the preemption count in preempt_schedule(). This reduces the icache
  footprint and the overhead of preemption.

- plus optimize the irq-path preemption check a bit.

15 years agothis is a fragile piece of the ptrace code, the code relies on a single wakeup coming...
Ingo Molnar [Wed, 13 Feb 2002 14:32:22 +0000 (15:32 +0100)]
this is a fragile piece of the ptrace code, the code relies on a single wakeup coming from the parent.
This fix is necessery after the preempt_schedule() cleanups, it unbreaks 'strace strace ...'.

15 years agodo not unlock irqs before calling schedule() - besides being a small exit() speedup...
Ingo Molnar [Wed, 13 Feb 2002 03:45:03 +0000 (04:45 +0100)]
do not unlock irqs before calling schedule() - besides being a small exit() speedup, this also
fixes a preemption race that was introduced by my removal of PREEMPT_ACTIVE.

15 years agoMerge hostme.bitkeeper.com:/ua/repos/g/gkernel/linus-2.5
Jeff Garzik [Wed, 13 Feb 2002 02:01:25 +0000 (18:01 -0800)]
Merge hostme.bitkeeper.com:/ua/repos/g/gkernel/linus-2.5
into hostme.bitkeeper.com:/ua/repos/g/gkernel/net-drivers-2.5

15 years agoMerge hostme.bitkeeper.com:/ua/repos/g/gkernel/linus-2.5
Jeff Garzik [Wed, 13 Feb 2002 01:57:20 +0000 (17:57 -0800)]
Merge hostme.bitkeeper.com:/ua/repos/g/gkernel/linus-2.5
into hostme.bitkeeper.com:/ua/repos/g/gkernel/misc-2.5

15 years agoMerge tp1.ruhr-uni-bochum.de:/home/kai/kernel/linux-2.5.isdn.hisax_netjet
Kai Germaschewski [Wed, 13 Feb 2002 00:05:22 +0000 (01:05 +0100)]
Merge tp1.ruhr-uni-bochum.de:/home/kai/kernel/linux-2.5.isdn.hisax_netjet
into tp1.ruhr-uni-bochum.de:/home/kai/kernel/linux-2.5.isdn

15 years agoHiSax netjet driver update
Kai Germaschewski [Wed, 13 Feb 2002 00:04:51 +0000 (01:04 +0100)]
HiSax netjet driver update

Move netjet driver to new PCI DMA API.

15 years agoHiSax HFC PCI driver udpate
Kai Germaschewski [Tue, 12 Feb 2002 23:14:24 +0000 (00:14 +0100)]
HiSax HFC PCI driver udpate

Move to PCI DMA API.
Make it work on big endian machines.

Contributed by Benjamin Herrenschmidt.

15 years agoa cleanup and a bugfix in the preemptive kernel:
Ingo Molnar [Tue, 12 Feb 2002 19:57:41 +0000 (20:57 +0100)]
a cleanup and a bugfix in the preemptive kernel:

- the PREEMPT_ACTIVE trick is not needed

- schedule() should check for need_resched, we might miss a
  reschedule otherwise.

the cleanup also fixes the bug. The only reason why i kept
preempt_schedule() was to fix up p->state to TASK_RUNNING,
to make it possible to preempt from places that mark the
task TASK_UNINTERRUPTIBLE before adding the task to a waitqueue,
and thus a preemption in that small window could cause the
task to be removed from the runqueue erroneously.

15 years agomove sched_find_first_bit() from mmu_context.h to bitops.h, it belongs there.
Ingo Molnar [Tue, 12 Feb 2002 19:05:41 +0000 (20:05 +0100)]
move sched_find_first_bit() from mmu_context.h to bitops.h, it belongs there.

15 years agooptimization, cleanup: switch_to(3 parameter) => switch_to(2 parameter).
Ingo Molnar [Tue, 12 Feb 2002 18:53:21 +0000 (19:53 +0100)]
optimization, cleanup: switch_to(3 parameter) => switch_to(2 parameter).

15 years ago[PATCH] usb_set_interface: correct toggle reset
Martin Diehl [Tue, 12 Feb 2002 14:55:33 +0000 (06:55 -0800)]
[PATCH] usb_set_interface: correct toggle reset

this is a patch to prevent usb_set_interface() from erroneously resetting
the toggles for all endpoints instead of only the affected ones from the
requested interface/altsetting. I've also added some missing parentheses
to related macros in usb.h as I prefered not to take special care for
nasty side-effects ;-)

Patch below was created against 2.4.18-pre9 (with some lines of offset it
applies to 2.5.4-pre5 as well).

Tested in multi-interface configuration to provide evidence it:
* correctly identifies the affected endpoints and resets the toggles
* doesn't touch endpoints from other interfaces
* provides correct handling of shared EP0
* solves an issue I had with 2.4.18-pre9 where setting one interface
  occasionally caused transfers on other interface to hang due to lost
  toggle synchronisation

Despite being a pure bugfix, well localized and (IMHO) pretty obviously
correct wrt. USB-spec, I'd like to suggest including this in early
2.4.19-pre. Just in case some existing driver would somehow workaround
the currently wrong behavior and might break with this fix. And it's
not very urgent right now, as we are probably close to 2.4.18-rc1.

Regards,
Martin

15 years agousb hid driver:
Vojtech Pavlik [Tue, 12 Feb 2002 14:48:06 +0000 (06:48 -0800)]
usb hid driver:
- patch to fix bug where urbs were freed too soon.

15 years agoAdd Macrolink board PCI ids to pci.ids and pci_ids.h.
Jeff Garzik [Tue, 12 Feb 2002 03:40:11 +0000 (22:40 -0500)]
Add Macrolink board PCI ids to pci.ids and pci_ids.h.

Contributor: Ed Vance @ Macrolink

15 years agoAdd new pci id to 8139too net driver, for Allied Telesyn cardbus cards.
Jeff Garzik [Tue, 12 Feb 2002 03:24:42 +0000 (22:24 -0500)]
Add new pci id to 8139too net driver, for Allied Telesyn cardbus cards.

Contributor: Go Taniguchi

15 years agoMerge bug fixes and PPC-specific feature additions from 2.4.x
Jeff Garzik [Tue, 12 Feb 2002 03:22:45 +0000 (22:22 -0500)]
Merge bug fixes and PPC-specific feature additions from 2.4.x
into bmac and mace net drivers.

Via Dave Jones.