]> git.hungrycats.org Git - linux/log
linux
22 years agoLinux v2.5.33 v2.5.33
Linus Torvalds [Sat, 31 Aug 2002 08:07:38 +0000 (01:07 -0700)]
Linux v2.5.33

22 years agoClean up insane floppy driver CURRENT handling, make the
Linus Torvalds [Sat, 31 Aug 2002 08:00:16 +0000 (01:00 -0700)]
Clean up insane floppy driver CURRENT handling, make the
driver remove the ftont of the queue from the request list
and cache it in 'current_req'.

This simplifies the driver and should be a lot more robust.

Still, I'll probably go blind just for _looking_ at the dang
sources of this thing. Whee.

22 years agoMake block device initialization initialize the request queue
Linus Torvalds [Sat, 31 Aug 2002 07:58:00 +0000 (00:58 -0700)]
Make block device initialization initialize the request queue
pointer before the device is opened.

This should finally fix the 2.5.x floppy driver.

22 years agoCall con_init_devfs() to initialize VT subsystem for devfs.
Linus Torvalds [Sat, 31 Aug 2002 04:22:42 +0000 (21:22 -0700)]
Call con_init_devfs() to initialize VT subsystem for devfs.

22 years agoMerge http://linux-ntfs.bkbits.net/ntfs-2.5
Linus Torvalds [Sat, 31 Aug 2002 04:03:08 +0000 (21:03 -0700)]
Merge http://linux-ntfs.bkbits.net/ntfs-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5
Anton Altaparmakov [Sat, 31 Aug 2002 14:42:10 +0000 (15:42 +0100)]
Merge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5

22 years agoMerge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5
Anton Altaparmakov [Sat, 31 Aug 2002 06:42:30 +0000 (07:42 +0100)]
Merge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5

22 years ago[PATCH] Fix panic if pnpbios is enabled and speed up its check in
Luca Barbieri [Sat, 31 Aug 2002 03:21:42 +0000 (20:21 -0700)]
[PATCH] Fix panic if pnpbios is enabled and speed up its check in

This fixes the pnpbios CS check to check for the correct values (it
wasn't up to date with the various GDT reshuffles), moves it inside the
kernel mode check, modifies it so that it takes less instructions and
marks it with unlikely().

Note that the 2.5.32 version of this check will cause the kernel to
always panic since it checks for the kernel segments and will thus
decide to jump to the pnpbios fault handler without being in pnpbios.
pnpbios_core.c instead seems to use the correct values.

22 years ago[PATCH] ptrace exit fix
Daniel Jacobowitz [Sat, 31 Aug 2002 02:52:25 +0000 (19:52 -0700)]
[PATCH] ptrace exit fix

Gotta ptrace_unlink before calling release_task, and instead of fiddling
with the real_parent directly.

22 years agoACPI: fix needed due to previous pci_ops change
Greg Kroah-Hartman [Fri, 30 Aug 2002 15:33:07 +0000 (08:33 -0700)]
ACPI: fix needed due to previous pci_ops change

22 years agoThe SCSI layer should _not_ try to decide about non-existent
Linus Torvalds [Fri, 30 Aug 2002 09:59:19 +0000 (02:59 -0700)]
The SCSI layer should _not_ try to decide about non-existent
partitions. The higher layers do a better job of it.

22 years ago[PATCH] PATCH - md - Fix a typo in a recent patchset for raid5
Neil Brown [Fri, 30 Aug 2002 09:00:04 +0000 (02:00 -0700)]
[PATCH] PATCH - md - Fix a typo in a recent patchset for raid5

22 years ago[PATCH] PATCH - kNFSd - More small fixes for TCP nfsd
Neil Brown [Fri, 30 Aug 2002 09:00:00 +0000 (02:00 -0700)]
[PATCH] PATCH - kNFSd - More small fixes for TCP nfsd

sk_inuse should be bigger than "char" as we can
have more than 255 server threads.  Due to the way the count
is used, this is unlikely to actually cause a problem, but it
should nonetheless be fixed.

Also, two printk generate more noise than we would like,
so turn them into dprintk (debugging printk).

22 years ago[PATCH] sock_writeable not appropriate for TCP sockets, for 2.5.32
Chuck Lever [Fri, 30 Aug 2002 08:58:33 +0000 (01:58 -0700)]
[PATCH] sock_writeable not appropriate for TCP sockets, for 2.5.32

sock_writeable determines whether there is space in a socket's output
buffer.  socket write_space callbacks use it to determine whether to wake
up those that are waiting for more output buffer space.

however, sock_writeable is not appropriate for TCP sockets.  because the
RPC layer's write_space callback uses it for TCP sockets, the RPC layer
hammers on sock_sendmsg with dozens of write requests that are only a few
hundred bytes long when it is trying to send a large write RPC request.
this patch adds logic to the RPC layer's write_space callback that
properly handles TCP sockets.

patch reviewed by Trond and Alexey.

22 years ago[PATCH] prevent oops in xprt_lock_write, against 2.5.32
Chuck Lever [Fri, 30 Aug 2002 08:58:29 +0000 (01:58 -0700)]
[PATCH] prevent oops in xprt_lock_write, against 2.5.32

when several RPC requests want to reconnect a TCP transport socket at
once, xprt_lock_write serializes the tasks to prevent multiple socket
connects.  however, TCP connects are always done by a RPC child task that
has no request slot.  xprt_lock_write can oops if there is no request slot
allocated to the invoking RPC task.  reviewed and accepted by Trond.

the xprt_lock_write changes are not yet in 2.4, so this patch does not
apply to 2.4.

22 years ago[PATCH] TLS boot-initialization bugfix on SMP, 2.5.32-BK
Ingo Molnar [Fri, 30 Aug 2002 08:56:43 +0000 (01:56 -0700)]
[PATCH] TLS boot-initialization bugfix on SMP, 2.5.32-BK

This fixes a bad TLS initialization bug found by Andi Kleen.  x86/SMP
only worked due to luck.

22 years ago[PATCH] scheduler fixes, 2.5.32-BK
Ingo Molnar [Fri, 30 Aug 2002 08:56:01 +0000 (01:56 -0700)]
[PATCH] scheduler fixes, 2.5.32-BK

This adds two scheduler related fixes:

 - changes the migration code to use struct completion. Andrew pointed out
   that there might be a small window in where the up() touches the
   semaphore while the waiting task goes on and frees its stack. And
   completion is more suited for this kind of stuff anyway.

 - removes two unneeded exports, pointed out by Andrew.

22 years ago[PATCH] clone-cleanup 2.5.32-BK
Ingo Molnar [Fri, 30 Aug 2002 08:55:56 +0000 (01:55 -0700)]
[PATCH] clone-cleanup 2.5.32-BK

This moves CLONE_SETTID and CLONE_CLEARTID handling into kernel/fork.c,
where it belongs.  [the CLONE_SETTLS is x86-specific and thus remains in
the per-arch process.c] This makes support for these two new flags much
easier: architectures only have to pass in the user_tid pointer.

22 years ago[PATCH] include/asm-i386/msr.h
Dominik Brodowski [Fri, 30 Aug 2002 08:54:06 +0000 (01:54 -0700)]
[PATCH] include/asm-i386/msr.h

It would be helpful if these msr.h #defines could get in.

22 years ago[PATCH] efi.h move
David Mosberger [Fri, 30 Aug 2002 08:52:33 +0000 (01:52 -0700)]
[PATCH] efi.h move

It makes no sense to keep efi.h as an ia64-specific header (there really
are x86 machines coming out with optional EFI BIOS support).

22 years agoMerge http://lia64.bkbits.net/to-linus-2.5
Linus Torvalds [Fri, 30 Aug 2002 08:51:31 +0000 (01:51 -0700)]
Merge http://lia64.bkbits.net/to-linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years ago[PATCH] ext3 __FUNCTION__ pasting fix
Andrew Morton [Fri, 30 Aug 2002 08:49:35 +0000 (01:49 -0700)]
[PATCH] ext3 __FUNCTION__ pasting fix

Fix a __FUNCTION__ paste in revoke.c

22 years ago[PATCH] O_DIRECT for ext3
Andrew Morton [Fri, 30 Aug 2002 08:49:31 +0000 (01:49 -0700)]
[PATCH] O_DIRECT for ext3

O_DIRECT support for ext3.

It works OK in all journalling modes.

Updates to the file metadata and inode are journalled as usual.

If the system crashes during an appending O_DIRECT write then journal
recovery will truncate the written-to file back to the length which it
had on entry to that write.

If the system crashes during a file overwrite to existing blocks then
the file contents will be an unknown mixture of old and new.

If the system crashes during a file overwrite which instantiates new
blocks in the middle of the file then there is a possibility of
uninitialised disk blocks being present in the file post-recovery.

22 years ago[PATCH] fix an ext3 deadlock
Andrew Morton [Fri, 30 Aug 2002 08:49:26 +0000 (01:49 -0700)]
[PATCH] fix an ext3 deadlock

mpage_writepages() does a lock_page() on pages to be written back, even
when it is being used for page reclaim writeback.

This is normally OK, because the page is unlocked quickly - pages are
unlocked during writeback and nobody should be performing __GFP_FS
allocations inside lock_page().

But it has introduced a ranking problem in ext3:

generic_file_write
->lock_page
  ->ext3_prepare_write
    ->journal_start (waits for a commit)

versus

ext3_create()
->journal_start()
  ->ext3_new_inode(GFP_KERNEL)
    ->page reclaim
      ->mpage_writepages
        ->lock_page (locks up, transaction is held open)

Maybe sometime, I'll have to turn mpage_writepages' lock_page into a
trylock if the caller is PF_MEMALLOC.  But for now, let's make ext3's
inside-transaction allocations use GFP_NOFS.  There is only one of them.

22 years ago[PATCH] writeback correctness and efficiency changes
Andrew Morton [Fri, 30 Aug 2002 08:49:22 +0000 (01:49 -0700)]
[PATCH] writeback correctness and efficiency changes

This is a performance and correctness fix against the writeback paths.

The writeback code has competing requirements.  Sometimes it is used
for "memory cleansing": kupdate, bdflush, writer throttling, page
allocator writeback, etc.  And sometimes this same code is used for
data integrity pruposes: fsync, msync, fdatasync, sync, umount, various
other kernel-internal uses.

The problem is: how to handle a dirty buffer or page which is currently
under writeback.

For memory cleansing, we just want to skip that buffer/page and go onto
the next one.  But for sync, we must wait on the old writeback and then
start new writeback.

mpage_writepages() is current correct for cleansing, but incorrect for
sync.  block_write_full_page() is currently correct for sync, but
inefficient for cleansing.

The fix is fairly simple.

- In mpage_writepages(), don't skip the page is it's a sync
operation.

- In block_write_full_page(), skip the buffer if it is a sync
operation.  And return -EAGAIN to tell the caller that the writeout
didn't work out.  The caller must then set the page dirty again and
move it onto mapping->dirty_pages.

This is an extension of the writepage API: writepage can now return
EAGAIN.  There are only three callers, and they have been updated.

fail_writepage() and ext3_writepage() were actually doing this by
hand.  They have been changed to return -EAGAIN.  NTFS will want to
be able to return -EAGAIN from its writepage as well.

- A sticky question is: how to tell the writeout code which mode it
is operating in?  Cleansing or sync?

It's such a tiny code change that I didn't have the heart to go and
propagate a `mode' argument down every instance of writepages() and
writepage() in the kernel.  So I passed it in via current->flags.

Incidentally, the occurrence of a locked-and-dirty buffer in
block_write_full_page() is fairly rare: normally the collision avoidance
happens at the address_space level, via PageWriteback.  But some
mappings (blockdevs, ext3 files, etc) have their dirty buffers written
out via submit_bh().  It is these buffers which can stall
block_write_full_page().

This wart will be pretty intrusive to fix.  ext3 needs to become fully
page-based (ugh.  It's a block-based journalling filesystem, and pages
are unnatural).  blockdev mappings are still written out by buffers
because that's how filesystems use them.  Putting _all_ metadata
(indirects, inodes, superblocks, etc) into standalone address_spaces
would fix that up.

- filemap_fdatawrite() sets PF_SYNC.  So filemap_fdatawrite() is the
kernel function which will start writeback against a mapping for
"data integrity" purposes, whereas the unexported, internal-only
do_writepages() is the writeback function which is used for memory
cleansing.  This difference is the reason why I didn't consolidate
those functions ages ago...

- Lots of code paths had a bogus extra call to filemap_fdatawait(),
which I previously added in a moment of weak-headedness.  They have
all been removed.

22 years ago[PATCH] batched freeing of anon pages
Andrew Morton [Fri, 30 Aug 2002 08:49:17 +0000 (01:49 -0700)]
[PATCH] batched freeing of anon pages

A reworked version of the batched page freeing and lock amortisation
for VMA teardown.

It walks the existing 507-page list in the mmu_gather_t in 16-page
chunks, drops their refcounts in 16-page chunks, and de-LRUs and
frees any resulting zero-count pages in up-to-16 page chunks.

22 years ago[PATCH] put_page() consolidation
Andrew Morton [Fri, 30 Aug 2002 08:49:13 +0000 (01:49 -0700)]
[PATCH] put_page() consolidation

Clean up put_page() and page_cache_release().  It's pretty simple now:

#define page_cache_get(page)           get_page(page)
#define page_cache_release(page)       put_page(page)

22 years ago[PATCH] remove pagevec_lru_del()
Andrew Morton [Fri, 30 Aug 2002 08:49:08 +0000 (01:49 -0700)]
[PATCH] remove pagevec_lru_del()

it was only being used in invalidate_inode_pages(), and from there,
pagevec_release() does the same thing.

22 years ago[PATCH] debug check in put_page_testzero()
Andrew Morton [Fri, 30 Aug 2002 08:49:04 +0000 (01:49 -0700)]
[PATCH] debug check in put_page_testzero()

As suggested by Daniel - it's a bug to run put_page_testzero
against a zero-ref page.

22 years ago[PATCH] MAINTAINERS patch
Ingo Molnar [Fri, 30 Aug 2002 08:45:57 +0000 (01:45 -0700)]
[PATCH] MAINTAINERS patch

please apply this patch (Robert ACK-ed it). While there is a preemptible
kernel entry already, i think listing this at the scheduler entry is
justfied, preemption has a number of scheduler interactions.

22 years ago[PATCH] ldt-fix-2.5.32-A3
Ingo Molnar [Fri, 30 Aug 2002 08:45:53 +0000 (01:45 -0700)]
[PATCH] ldt-fix-2.5.32-A3

this is an updated version of the LDT fixes. It fixes the following kinds
of problems:

 - fix a possible gcc optimization causing a race causing the loading of a
   corrupt LDT descriptor upon context switch. [this fix got simplified
   over previous versions.]

 - remove an unconditional OOM printk, and there's no need to set ->size
   in the OOM path.

 - fix preemption bugs, load_LDT()/clear_LDT() was not preemption-safe,
   when it was used outside of spinlocks.

the context-switch race is the following. 'LDT modification' is the
following operation: the seg->ldt pointer is modified, then seg->size is
modified. In theory gcc is free to reschedule the two modifications, and
first modify ->size, then ->ldt. Thus if this modification is not
synchronized with context-switches, another thread might see a temporary
state of the new ->size [which was increased], but still the old pointer.
Ie.:

CPU0 CPU1

pc->size = newsize;
load_LDT(); // (oldptr, newsize)
pc->ldt = newptr;

the corrupt LDT is loaded until the SMP cross-call is sent, leaving the
window open for many usecs.

the fix is to put a wmb() after ->ldt modifications. [this is also in
preparation of not-write-ordered SMP x86 designs.]

22 years agoMerge bk://linux-input.bkbits.net/linux-input
Linus Torvalds [Fri, 30 Aug 2002 08:43:13 +0000 (01:43 -0700)]
Merge bk://linux-input.bkbits.net/linux-input
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoIgnore error 0xff - 'general error' in AUX wire test in i8042.c,
Vojtech Pavlik [Fri, 30 Aug 2002 13:22:23 +0000 (15:22 +0200)]
Ignore error 0xff - 'general error' in AUX wire test in i8042.c,
some mainboards (Andrew Morton's Dell) report that even everything
is okay with AUX. Also remove a check for very old AMI i8042's, which
could generate false positives on modern buggy mainboards.

22 years agoMerge bk://jfs.bkbits.net/linux-2.5
Linus Torvalds [Fri, 30 Aug 2002 08:42:01 +0000 (01:42 -0700)]
Merge bk://jfs.bkbits.net/linux-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years ago[PATCH] oss/gus_card.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:31:22 +0000 (01:31 -0700)]
[PATCH] oss/gus_card.c - convert cli to spinlocks

22 years ago[PATCH] oss/nm256.h - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:31:18 +0000 (01:31 -0700)]
[PATCH] oss/nm256.h - convert cli to spinlocks

22 years ago[PATCH] oss/pas2_card.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:31:14 +0000 (01:31 -0700)]
[PATCH] oss/pas2_card.c - convert cli to spinlocks

22 years ago[PATCH] oss/vwsnd.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:31:08 +0000 (01:31 -0700)]
[PATCH] oss/vwsnd.c - convert cli to spinlocks

22 years ago[PATCH] oss/trident.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:31:03 +0000 (01:31 -0700)]
[PATCH] oss/trident.c - convert cli to spinlocks

22 years ago[PATCH] oss/midi_synth.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:59 +0000 (01:30 -0700)]
[PATCH] oss/midi_synth.c - convert cli to spinlocks

22 years ago[PATCH] oss/sonicvibes.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:55 +0000 (01:30 -0700)]
[PATCH] oss/sonicvibes.c - convert cli to spinlocks

22 years ago[PATCH] oss/esssolo1.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:51 +0000 (01:30 -0700)]
[PATCH] oss/esssolo1.c - convert cli to spinlocks

22 years ago[PATCH] oss/rme96xx.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:46 +0000 (01:30 -0700)]
[PATCH] oss/rme96xx.c - convert cli to spinlocks

22 years ago[PATCH] oss/cmpci.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:42 +0000 (01:30 -0700)]
[PATCH] oss/cmpci.c - convert cli to spinlocks

22 years ago[PATCH] oss/waveartist.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:38 +0000 (01:30 -0700)]
[PATCH] oss/waveartist.c - convert cli to spinlocks

22 years ago[PATCH] oss/soundcard.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:33 +0000 (01:30 -0700)]
[PATCH] oss/soundcard.c - convert cli to spinlocks

22 years ago[PATCH] oss/wavfront.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:29 +0000 (01:30 -0700)]
[PATCH] oss/wavfront.c - convert cli to spinlocks

22 years ago[PATCH] oss/opl3sa2.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:25 +0000 (01:30 -0700)]
[PATCH] oss/opl3sa2.c - convert cli to spinlocks

22 years ago[PATCH] oss/opl3sa.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:20 +0000 (01:30 -0700)]
[PATCH] oss/opl3sa.c - convert cli to spinlocks

22 years ago[PATCH] oss/dev_table.h - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:16 +0000 (01:30 -0700)]
[PATCH] oss/dev_table.h - convert cli to spinlocks

22 years ago[PATCH] oss/sys_timer.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:12 +0000 (01:30 -0700)]
[PATCH] oss/sys_timer.c - convert cli to spinlocks

22 years ago[PATCH] oss/mad16.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:07 +0000 (01:30 -0700)]
[PATCH] oss/mad16.c - convert cli to spinlocks

22 years ago[PATCH] oss/nec_vrc5477.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:30:03 +0000 (01:30 -0700)]
[PATCH] oss/nec_vrc5477.c - convert cli to spinlocks

22 years ago[PATCH] oss/sound_timer.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:29:58 +0000 (01:29 -0700)]
[PATCH] oss/sound_timer.c - convert cli to spinlocks

22 years ago[PATCH] oss/msnd_pinnacle.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:29:54 +0000 (01:29 -0700)]
[PATCH] oss/msnd_pinnacle.c - convert cli to spinlocks

22 years ago[PATCH] oss/es1370.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:29:50 +0000 (01:29 -0700)]
[PATCH] oss/es1370.c - convert cli to spinlocks

22 years ago[PATCH] oss/ite8172.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:29:46 +0000 (01:29 -0700)]
[PATCH] oss/ite8172.c - convert cli to spinlocks

22 years ago[PATCH] oss/maestro.c - convert cli to spinlocks
Peter Wächtler [Fri, 30 Aug 2002 08:29:41 +0000 (01:29 -0700)]
[PATCH] oss/maestro.c - convert cli to spinlocks

22 years agoProper implementation of jfs_get_blocks
Dave Kleikamp [Fri, 30 Aug 2002 08:21:28 +0000 (03:21 -0500)]
Proper implementation of jfs_get_blocks

jfs_get_blocks should return up to the number of blocks in the
extent rather than limiting itself to one block, as the initial,
trivial implementation did.  This greatly reduces the overhead of
O_DIRECT reads and writes.

Submitted by Badari Pulavarty (pbadari@us.ibm.com)

22 years agoMerge http://linuxusb.bkbits.net/pci-2.5
Linus Torvalds [Fri, 30 Aug 2002 08:19:14 +0000 (01:19 -0700)]
Merge http://linuxusb.bkbits.net/pci-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoJFS: Add write_super_lockfs() and unlock_fs() for snapshot.
Dave Kleikamp [Fri, 30 Aug 2002 08:09:16 +0000 (03:09 -0500)]
JFS: Add write_super_lockfs() and unlock_fs() for snapshot.

Submitted by Steve Best.

22 years agoPCI: compile time fix for the pci pool patch.
Greg Kroah-Hartman [Fri, 30 Aug 2002 07:56:40 +0000 (00:56 -0700)]
PCI: compile time fix for the pci pool patch.

22 years ago[PATCH] show pci_pool stats in driverfs]
David Brownell [Fri, 30 Aug 2002 07:52:18 +0000 (00:52 -0700)]
[PATCH] show pci_pool stats in driverfs]

This patch exposes basic allocation statistics for pci pools,
very much like /proc/slabinfo but applying to DMA-consistent
memory.  A file "pools" is created in the driverfs directory
for the relevant pci device when the first pool is created, and
removed when the last pool is destroyed.

Please merge to 2.5.latest.  If it matters, DaveM said it
looks fine.  It produces sane output for all the 2.5.30
USB host controller drivers.

22 years ago[PATCH] Fixed NUMA-Q PCI patch
Matthew Dobson [Fri, 30 Aug 2002 07:34:21 +0000 (00:34 -0700)]
[PATCH] Fixed NUMA-Q PCI patch

This patch fixes a bug in NUMA-Q PCI code where the kernel can't find PCI
devices on any node other than the first.

22 years ago[PATCH] PCI: sh pci_ops changes
Hanna V. Linder [Fri, 30 Aug 2002 07:30:06 +0000 (00:30 -0700)]
[PATCH] PCI: sh pci_ops changes

sh pci ops changes

22 years ago[PATCH] PCI: mips pci_ops changes
Hanna V. Linder [Fri, 30 Aug 2002 07:29:38 +0000 (00:29 -0700)]
[PATCH] PCI: mips pci_ops changes

mips pci ops changes

22 years ago[PATCH] PCI: ia64 pci_ops changes
Hanna V. Linder [Fri, 30 Aug 2002 07:29:11 +0000 (00:29 -0700)]
[PATCH] PCI: ia64 pci_ops changes

ia64 pci ops changes

22 years ago[PATCH] PCI: alpha pci_ops changes
Greg Kroah-Hartman [Fri, 30 Aug 2002 07:28:47 +0000 (00:28 -0700)]
[PATCH] PCI: alpha pci_ops changes

pci_ops update for most of the alpha ports.

22 years ago[PATCH] PCI: x86-64 pci_ops changes
Greg Kroah-Hartman [Fri, 30 Aug 2002 07:28:21 +0000 (00:28 -0700)]
[PATCH] PCI: x86-64 pci_ops changes

x86-64 pci changes

22 years ago[PATCH] PCI Hotplug: removed the pci_*_nodev functions
Greg Kroah-Hartman [Fri, 30 Aug 2002 07:27:56 +0000 (00:27 -0700)]
[PATCH] PCI Hotplug: removed the pci_*_nodev functions

removed the pci_*_nodev functions, as the pci_bus_* functions should be used instead.

22 years ago[PATCH] PCI: add pci_bus_* functions to replace the pci_read_* and pci_write_* funct...
Greg Kroah-Hartman [Fri, 30 Aug 2002 07:27:34 +0000 (00:27 -0700)]
[PATCH] PCI:  add pci_bus_* functions to replace the pci_read_* and pci_write_* functions.

add pci_bus_* functions to replace the pci_read_* and pci_write_* functions.

22 years ago[PATCH] PCI Cleanup
Matthew Dobson [Fri, 30 Aug 2002 07:27:12 +0000 (00:27 -0700)]
[PATCH] PCI Cleanup

The patch removes the pci_confN_(read|write)_config_(byte|word|dword) mess and
pares it down to pci_confN_(read|write).  This change is reflected in the
pci_ops structure, which only has read and write function pointers rather than
the byte, word, and dword versions.  These changes happen in the pci_conf(1|2)
and pci_bios read and write calls.

This patch also removes the pci_config_(read|write) function pointers.  People
shouldn't be using these (I don't think) and should be using the pci_ops
structure linked through the pci_dev structure.  These end up calling the same
functions that the pci_config_(read|write) pointers refer to anyway.

22 years agoMerge kleikamp.austin.ibm.com:/home/shaggy/bk/jfs-2.5
Dave Kleikamp [Fri, 30 Aug 2002 04:33:35 +0000 (23:33 -0500)]
Merge kleikamp.austin.ibm.com:/home/shaggy/bk/jfs-2.5
into kleikamp.austin.ibm.com:/home/shaggy/bk/acl-2.5

22 years agoJFS extended attributes
Dave Kleikamp [Fri, 30 Aug 2002 04:09:38 +0000 (23:09 -0500)]
JFS extended attributes

22 years agoMerge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5
Anton Altaparmakov [Thu, 29 Aug 2002 23:40:25 +0000 (00:40 +0100)]
Merge cantab.net:/usr/src/bklinux-2.5 into cantab.net:/usr/src/ntfs-2.5

22 years agoCreate dummy file include/asm-ia64/mc146818rtc.h since ide-geometry.c continues to
David Mosberger [Thu, 29 Aug 2002 14:12:06 +0000 (07:12 -0700)]
Create dummy file include/asm-ia64/mc146818rtc.h since ide-geometry.c continues to
insist on it.

22 years agoJFS: rework extent invalidation
Dave Kleikamp [Thu, 29 Aug 2002 11:39:47 +0000 (06:39 -0500)]
JFS: rework extent invalidation

All callers of invalidate_metapages() actually have a dxd_t or pxd_t to
invalidate, so add invalidate_pxd_metapages() and invalidate_dxd_metapages()
routines with a common __invalidate_metapages() backend instead.

Start to invalidate the EA/ACL extents, we'll need that soon.

Submitted by Christoph Hellwig

22 years agoMerge master.kernel.org:/home/davem/BK/sctp-rename-2.5
Linus Torvalds [Thu, 29 Aug 2002 10:39:52 +0000 (03:39 -0700)]
Merge master.kernel.org:/home/davem/BK/sctp-rename-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge penguin:v2.5/linux
Linus Torvalds [Thu, 29 Aug 2002 10:38:52 +0000 (03:38 -0700)]
Merge penguin:v2.5/linux
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years ago[SCTP]: Rename sctp_foo.[ch] to foo.[ch] and kill CVS tags on authors request.
David S. Miller [Thu, 29 Aug 2002 09:55:45 +0000 (02:55 -0700)]
[SCTP]: Rename sctp_foo.[ch] to foo.[ch] and kill CVS tags on authors request.

22 years agoMerge http://linux-acpi.bkbits.net/linux-acpi
Linus Torvalds [Thu, 29 Aug 2002 09:11:27 +0000 (02:11 -0700)]
Merge http://linux-acpi.bkbits.net/linux-acpi
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoia64: Sync with 2.5.32 to get a working kernel.
David Mosberger [Thu, 29 Aug 2002 09:02:35 +0000 (02:02 -0700)]
ia64: Sync with 2.5.32 to get a working kernel.

22 years ago[PATCH] oss/v_midi.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:29:02 +0000 (01:29 -0700)]
[PATCH] oss/v_midi.c - convert cli to spinlocks

22 years ago[PATCH] oss/uart401.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:28:50 +0000 (01:28 -0700)]
[PATCH] oss/uart401.c - convert cli to spinlocks

22 years ago[PATCH] oss/uart6850.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:28:38 +0000 (01:28 -0700)]
[PATCH] oss/uart6850.c - convert cli to spinlocks

22 years ago[PATCH] oss/pas2_midi.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:28:26 +0000 (01:28 -0700)]
[PATCH] oss/pas2_midi.c - convert cli to spinlocks

22 years ago[PATCH] oss/nm256_audio.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:28:15 +0000 (01:28 -0700)]
[PATCH] oss/nm256_audio.c - convert cli to spinlocks

22 years ago[PATCH] oss/pas2_pcm.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:28:02 +0000 (01:28 -0700)]
[PATCH] oss/pas2_pcm.c - convert cli to spinlocks

22 years ago[PATCH] oss/midibuf.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:27:50 +0000 (01:27 -0700)]
[PATCH] oss/midibuf.c - convert cli to spinlocks

22 years ago[PATCH] oss/audio.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:27:36 +0000 (01:27 -0700)]
[PATCH] oss/audio.c - convert cli to spinlocks

22 years ago[PATCH] oss/sequencer.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:27:26 +0000 (01:27 -0700)]
[PATCH] oss/sequencer.c - convert cli to spinlocks

22 years ago[PATCH] oss/mpu401.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:27:16 +0000 (01:27 -0700)]
[PATCH] oss/mpu401.c - convert cli to spinlocks

22 years ago[PATCH] oss/pss.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:27:04 +0000 (01:27 -0700)]
[PATCH] oss/pss.c - convert cli to spinlocks

22 years ago[PATCH] oss/sscape.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:26:53 +0000 (01:26 -0700)]
[PATCH] oss/sscape.c - convert cli to spinlocks

22 years ago[PATCH] oss/ad1816.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:26:43 +0000 (01:26 -0700)]
[PATCH] oss/ad1816.c - convert cli to spinlocks

22 years ago[PATCH] oss/ad1848.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:26:30 +0000 (01:26 -0700)]
[PATCH] oss/ad1848.c - convert cli to spinlocks

22 years ago[PATCH] oss/dmabuf.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:26:19 +0000 (01:26 -0700)]
[PATCH] oss/dmabuf.c - convert cli to spinlocks

22 years ago[PATCH] oss/maestro3.c - convert cli to spinlocks
Peter Wächtler [Thu, 29 Aug 2002 08:26:09 +0000 (01:26 -0700)]
[PATCH] oss/maestro3.c - convert cli to spinlocks

This is the first of a series of patches against 2.5.32
converting almost all remaining OSS sound drivers to use
spin_lock_irqsave&co instead of cli() which is nowadays
gone when compiling for SMP.

The more complicated ones were:
dmabuf.c
maestro3.c
ad1848.c

I hope I haven't broken anything - but I think some of the
drivers are simply not SMP safe.

One day OSS and some of the older soundcards are obsolete - so what?

22 years agoMerge bk://linux-input.bkbits.net/linux-input
Linus Torvalds [Thu, 29 Aug 2002 07:15:42 +0000 (00:15 -0700)]
Merge bk://linux-input.bkbits.net/linux-input
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

22 years agoIn mousedev.c, don't send a zero mouse movement after a command if
Vojtech Pavlik [Fri, 30 Aug 2002 00:27:35 +0000 (02:27 +0200)]
In mousedev.c, don't send a zero mouse movement after a command if
requested, also fix a possible race with two processes using the same
file descriptor.