firefly-linux-kernel-4.4.55.git
19 years ago[PATCH] SELinux: cleanup ipc_has_perm
Stephen Smalley [Sun, 1 May 2005 15:58:39 +0000 (08:58 -0700)]
[PATCH] SELinux: cleanup ipc_has_perm

This patch removes the sclass argument from ipc_has_perm in the SELinux
module, as it can be obtained from the ipc security structure.  The use of
a separate argument was a legacy of the older precondition function
handling in SELinux and is obsolete.  Please apply.

Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: James Morris <jmorris@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] drop_buffers() oops fix
akpm@osdl.org [Sun, 1 May 2005 15:58:39 +0000 (08:58 -0700)]
[PATCH] drop_buffers() oops fix

In rare situations, drop_buffers() can be called for a page which has buffers,
but no ->mapping (it was truncated, but the buffers were left behind because
ext3 was still fiddling with them).

But if there was an I/O error in a buffer_head, drop_buffers() will try to get
at the address_space and will oops.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mpage_writepages() page locking fix
Nikita Danilov [Sun, 1 May 2005 15:58:39 +0000 (08:58 -0700)]
[PATCH] mpage_writepages() page locking fix

When ->writepage() returns WRITEPAGE_ACTIVATE, the page is still locked.
Explicitly unlock the page in mpage_writepages().

Signed-off-by: Nikita Danilov <nikita@clusterfs.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] add kmalloc_node, inline cleanup
Manfred Spraul [Sun, 1 May 2005 15:58:38 +0000 (08:58 -0700)]
[PATCH] add kmalloc_node, inline cleanup

The patch makes the following function calls available to allocate memory
on a specific node without changing the basic operation of the slab
allocator:

 kmem_cache_alloc_node(kmem_cache_t *cachep, unsigned int flags, int node);
 kmalloc_node(size_t size, unsigned int flags, int node);

in a similar way to the existing node-blind functions:

 kmem_cache_alloc(kmem_cache_t *cachep, unsigned int flags);
 kmalloc(size, flags);

kmem_cache_alloc_node was changed to pass flags and the node information
through the existing layers of the slab allocator (which lead to some minor
rearrangements).  The functions at the lowest layer (kmem_getpages,
cache_grow) are already node aware.  Also __alloc_percpu can call
kmalloc_node now.

Performance measurements (using the pageset localization patch) yields:

w/o patches:
Tasks    jobs/min  jti  jobs/min/task      real       cpu
    1      484.27  100       484.2736     12.02      1.97   Wed Mar 30 20:50:43 2005
  100    25170.83   91       251.7083     23.12    150.10   Wed Mar 30 20:51:06 2005
  200    34601.66   84       173.0083     33.64    294.14   Wed Mar 30 20:51:40 2005
  300    37154.47   86       123.8482     46.99    436.56   Wed Mar 30 20:52:28 2005
  400    39839.82   80        99.5995     58.43    580.46   Wed Mar 30 20:53:27 2005
  500    40036.32   79        80.0726     72.68    728.60   Wed Mar 30 20:54:40 2005
  600    44074.21   79        73.4570     79.23    872.10   Wed Mar 30 20:55:59 2005
  700    44016.60   78        62.8809     92.56   1015.84   Wed Mar 30 20:57:32 2005
  800    40411.05   80        50.5138    115.22   1161.13   Wed Mar 30 20:59:28 2005
  900    42298.56   79        46.9984    123.83   1303.42   Wed Mar 30 21:01:33 2005
 1000    40955.05   80        40.9551    142.11   1441.92   Wed Mar 30 21:03:55 2005

with pageset localization and slab API patches:
Tasks    jobs/min  jti  jobs/min/task      real       cpu
    1      484.19  100       484.1930     12.02      1.98   Wed Mar 30 21:10:18 2005
  100    27428.25   92       274.2825     21.22    149.79   Wed Mar 30 21:10:40 2005
  200    37228.94   86       186.1447     31.27    293.49   Wed Mar 30 21:11:12 2005
  300    41725.42   85       139.0847     41.84    434.10   Wed Mar 30 21:11:54 2005
  400    43032.22   82       107.5805     54.10    582.06   Wed Mar 30 21:12:48 2005
  500    42211.23   83        84.4225     68.94    722.61   Wed Mar 30 21:13:58 2005
  600    40084.49   82        66.8075     87.12    873.11   Wed Mar 30 21:15:25 2005
  700    44169.30   79        63.0990     92.24   1008.77   Wed Mar 30 21:16:58 2005
  800    43097.94   79        53.8724    108.03   1155.88   Wed Mar 30 21:18:47 2005
  900    41846.75   79        46.4964    125.17   1303.38   Wed Mar 30 21:20:52 2005
 1000    40247.85   79        40.2478    144.60   1442.21   Wed Mar 30 21:23:17 2005

Signed-off-by: Christoph Lameter <christoph@lameter.com>
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] sync_page() smp_mb() comment
William Lee Irwin III [Sun, 1 May 2005 15:58:38 +0000 (08:58 -0700)]
[PATCH] sync_page() smp_mb() comment

The smp_mb() is becaus sync_page() doesn't have PG_locked while it accesses
page_mapping(page).  The comments in the patch (the entire patch is the
addition of this comment) try to explain further how and why smp_mb() is
used.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] RLIMIT_MEMLOCK checking fix
Chris Wright [Sun, 1 May 2005 15:58:38 +0000 (08:58 -0700)]
[PATCH] RLIMIT_MEMLOCK checking fix

Always use page counts when doing RLIMIT_MEMLOCK checking to avoid possible
overflow.

Signed-off-by: Chris Wright <chrisw@osdl.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] count bounce buffer pages in vmstat
KAMEZAWA Hiroyuki [Sun, 1 May 2005 15:58:37 +0000 (08:58 -0700)]
[PATCH] count bounce buffer pages in vmstat

This is a patch for counting the number of pages for bounce buffers.  It's
shown in /proc/vmstat.

Currently, the number of bounce pages are not counted anywhere.  So, if
there are many bounce pages, it seems that there are leaked pages.  And
it's difficult for a user to imagine the usage of bounce pages.  So, it's
meaningful to show # of bouce pages.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] doc: Locking update
Nikita Danilov [Sun, 1 May 2005 15:58:37 +0000 (08:58 -0700)]
[PATCH] doc: Locking update

Make the Locking document truer.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mm: use __GFP_NOMEMALLOC
Nick Piggin [Sun, 1 May 2005 15:58:37 +0000 (08:58 -0700)]
[PATCH] mm: use __GFP_NOMEMALLOC

Use the new __GFP_NOMEMALLOC to simplify the previous handling of
PF_MEMALLOC.

Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mempool: simplify alloc
Nick Piggin [Sun, 1 May 2005 15:58:37 +0000 (08:58 -0700)]
[PATCH] mempool: simplify alloc

Mempool is pretty clever.  Looks too clever for its own good :) It
shouldn't really know so much about page reclaim internals.

- don't guess about what effective page reclaim might involve.

- don't randomly flush out all dirty data if some unlikely thing
  happens (alloc returns NULL). page reclaim can (sort of :P) handle
  it.

I think the main motivation is trying to avoid pool->lock at all costs.
However the first allocation is attempted with __GFP_WAIT cleared, so it
will be 'can_try_harder' if it hits the page allocator.  So if allocation
still fails, then we can probably afford to hit the pool->lock - and what's
the alternative?  Try page reclaim and hit zone->lru_lock?

A nice upshot is that we don't need to do any fancy memory barriers or do
(intentionally) racy access to pool-> fields outside the lock.

Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mempool: NOMEMALLOC and NORETRY
Nick Piggin [Sun, 1 May 2005 15:58:36 +0000 (08:58 -0700)]
[PATCH] mempool: NOMEMALLOC and NORETRY

Mempools have 2 problems.

The first is that mempool_alloc can possibly get stuck in __alloc_pages
when they should opt to fail, and take an element from their reserved pool.

The second is that it will happily eat emergency PF_MEMALLOC reserves
instead of going to their reserved pools.

Fix the first by passing __GFP_NORETRY in the allocation calls in
mempool_alloc.  Fix the second by introducing a __GFP_MEMPOOL flag which
directs the page allocator not to allocate from the reserve pool.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mm: pcp use non powers of 2 for batch size
Nick Piggin [Sun, 1 May 2005 15:58:36 +0000 (08:58 -0700)]
[PATCH] mm: pcp use non powers of 2 for batch size

Jack Steiner reported this to have fixed his problem (bad colouring):
"The patches fix both problems that I found - bad
 coloring & excessive pages in pagesets."

In most workloads this is not likely to be such a pronounced problem,
however it should help corner cases.  And avoiding powers of 2 in these
types of memory operations is always a good idea.

Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mm: rmap.c cleanup
Nikita Danilov [Sun, 1 May 2005 15:58:36 +0000 (08:58 -0700)]
[PATCH] mm: rmap.c cleanup

mm/rmap.c:page_referenced_one() and mm/rmap.c:try_to_unmap_one() contain
identical code that

 - takes mm->page_table_lock;

 - drills through page tables;

 - checks that correct pte is reached.

Coalesce this into page_check_address()

Signed-off-by: Nikita Danilov <nikita@clusterfs.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] RLIMIT_AS checking fix
akpm@osdl.org [Sun, 1 May 2005 15:58:35 +0000 (08:58 -0700)]
[PATCH] RLIMIT_AS checking fix

Address bug #4508: there's potential for wraparound in the various places
where we perform RLIMIT_AS checking.

(I'm a bit worried about acct_stack_growth().  Are we sure that vma->vm_mm is
always equal to current->mm?  If not, then we're comparing some other
process's total_vm with the calling process's rlimits).

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] generic_file_buffered_write fixes
akpm@osdl.org [Sun, 1 May 2005 15:58:35 +0000 (08:58 -0700)]
[PATCH] generic_file_buffered_write fixes

Anton Altaparmakov <aia21@cam.ac.uk> points out:

- It calls fault_in_pages_readable() which is completely bogus if @nr_segs >
  1.  It needs to be replaced by a to be written
  "fault_in_pages_readable_iovec()".

- It increments @buf even in the iovec case thus @buf can point to random
  memory really quickly (in the iovec case) and then it calls
  fault_in_pages_readable() on this random memory.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ultrastor build fix
akpm@osdl.org [Sun, 1 May 2005 15:58:35 +0000 (08:58 -0700)]
[PATCH] ultrastor build fix

Fix a typo.

James Bottomley <James.Bottomley@steeleye.com>

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kbuild: Set NOSTDINC_FLAGS late to speed up compile (a little)
Sam Ravnborg [Sat, 30 Apr 2005 23:51:42 +0000 (16:51 -0700)]
[PATCH] kbuild: Set NOSTDINC_FLAGS late to speed up compile (a little)

Move definition of NOSTDINC_FLAGS below inclusion of arch Makefile, so
any arch specific settings to $(CC) takes effect before looking up the
compiler include directory.

The previous solution that replaced ':=' with '=' caused gcc to be
invoked one additional time for each directory visited.

This decreases kernel compile time with 0.1 second (3.6 -> 3.5 seconds) when
running make on a fully built kernel

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kbuild/ppc: tell when uimage was not built
Sam Ravnborg [Sat, 30 Apr 2005 23:51:42 +0000 (16:51 -0700)]
[PATCH] kbuild/ppc: tell when uimage was not built

Tom Rini said:
  Note that there is still a trivial'ish change to make.  When mkimage
  doesn't exist on the host we should say "uImage not made" or
  something similar.

So I did like Tom asked.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kbuild/i386: re-introduce dependency on vmlinux for install target, and add...
Sam Ravnborg [Sat, 30 Apr 2005 23:51:42 +0000 (16:51 -0700)]
[PATCH] kbuild/i386: re-introduce dependency on vmlinux for install target, and add kernel_install

Removing the dependency on vmlinux for the install target raised a few
complaints, so instead a new target i added: kernel_install.

kernel_install will install the kernel just like the ordinary install target.
The only difference is that install has a dependency on vmlinux,
kernel_install does not. Therefore kernel_install is the best choice
when accessing the kernel over a NFS mount or as another user.

kernel_install is similar to modules_install in the fact that neither does
a full kernel compile before performing the install.
In this way they are good for root use. Also added back the
dependency on vmlinux for the install target so peoples scripts are no
longer broken.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge of master.kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Sat, 30 Apr 2005 22:42:33 +0000 (15:42 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] ARM: IntegratorCP: Fix CLCD MUX selection values
Russell King [Sat, 30 Apr 2005 22:32:38 +0000 (23:32 +0100)]
[PATCH] ARM: IntegratorCP: Fix CLCD MUX selection values

The documentation on these values seems to be rather wrong.
These values have been determined by mere trial and error.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: IntegratorCP: 16bpp is RGB565 not RGB555
Russell King [Sat, 30 Apr 2005 22:28:47 +0000 (23:28 +0100)]
[PATCH] ARM: IntegratorCP: 16bpp is RGB565 not RGB555

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: AMBA CLCD: program palette for pseudocolor visuals
Russell King [Sat, 30 Apr 2005 21:39:51 +0000 (22:39 +0100)]
[PATCH] ARM: AMBA CLCD: program palette for pseudocolor visuals

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years agoMerge of master.kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Sat, 30 Apr 2005 20:34:21 +0000 (13:34 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] cifs: Update cifs todo list
Steve French [Sat, 30 Apr 2005 18:10:58 +0000 (11:10 -0700)]
[PATCH] cifs: Update cifs todo list

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: append \* properly on ASCII servers
Steve French [Sat, 30 Apr 2005 18:10:57 +0000 (11:10 -0700)]
[PATCH] cifs: append \* properly on ASCII servers

For older servers which do not support Unicode

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: fix 32-bit signal frame back link
Paul Mackerras [Sat, 30 Apr 2005 17:01:40 +0000 (10:01 -0700)]
[PATCH] ppc64: fix 32-bit signal frame back link

When the kernel creates a signal frame on the user stack, it puts the
old stack pointer value at the beginning so that the signal frame is
linked into the chain of stack frames like any other frame.
Unfortunately, for 32-bit processes we are writing the old stack
pointer as a 64-bit value rather than a 32-bit value, and the process
sees that as a null pointer, since it only looks at the first 32 bits,
which are zero since ppc is bigendian and the stack pointer is below
4GB.  This bug is in SLES9 and RHEL4 too, hence the ccs.

This patch fixes the bug by making the signal code write the old stack
pointer as a u32 instead of an unsigned long.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ARM: PXA I2C: add platform device
Russell King [Sat, 30 Apr 2005 12:26:06 +0000 (13:26 +0100)]
[PATCH] ARM: PXA I2C: add platform device

Add the PXA I2C platform device.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: RTC: allow driver methods to return error
Russell King [Sat, 30 Apr 2005 11:19:28 +0000 (12:19 +0100)]
[PATCH] ARM: RTC: allow driver methods to return error

Allow RTC drivers to return error codes from their read_time
or read_alarm methods.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years agoMerge of master.kernel.org:/home/rmk/linux-2.6-serial.git
Linus Torvalds [Fri, 29 Apr 2005 22:08:34 +0000 (15:08 -0700)]
Merge ... /home/rmk/linux-2.6-serial.git

19 years agoMerge of master.kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Fri, 29 Apr 2005 22:06:00 +0000 (15:06 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] ARM: 2654/1: i.MX UART initialization sets and honors UFCR value
Sascha Hauer [Fri, 29 Apr 2005 21:46:40 +0000 (22:46 +0100)]
[PATCH] ARM: 2654/1: i.MX UART initialization sets and honors UFCR value

Patch from Sascha Hauer

This patch adds UCFR_RFDIV setting into i.MX serial driver.
This is required, if loader does not fully agree with Linux kernel
about UART setup manner. Linux only blindly expected some values until
now. This should enable to use even serial ports not recognized by
boot-loader as for example third UART found in the bluethoot module.
Patch also enables to detect original setup baudrate in more cases.

Signed-off-by: Pavel Pisa
Signed-off-by: Sascha Hauer
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2660/2: fix ixdp2800 boot and pci init
Lennert Buytenhek [Fri, 29 Apr 2005 21:13:57 +0000 (22:13 +0100)]
[PATCH] ARM: 2660/2: fix ixdp2800 boot and pci init

Patch from Lennert Buytenhek

The IXDP2800 is an evalution platform for the IXP2800 processor that
has two IXP2800s connected to the same PCI bus.  This is problematic
as both CPUs will try to configure the PCI bus as they boot linux.
Contrary to on the other IXP2000 platforms, the boot loader on the
IXDP2800 doesn't configure the PCI bus properly, so we do want the
linux instance on one of the CPUs to do that.
Making one of the CPUs ignore the PCI bus (and thus act like a pure
PCI slave device) is not an option because there is a 82559 NIC on
the PCI bus for each of the CPUs.
The chosen solution is to have the master CPU configure the PCI bus
while the slave is kept in a quiescent state, and then to have the
slave CPU scan the PCI bus (without assigning resources) while the
master is kept in a quiescent state.  After this ritual, the master
deletes the slave NIC from its PCI device list, the slave deletes
the master NIC from its device list, and (almost) all is well.
There's still one little problem: each of the CPUs has a 1G SDRAM
BAR, but the IXP2000 only has 512M of outbound PCI memory window.
We solve this by hand-assigning the master and slave SDRAM BARs to
a location outside each of the IXP's outbound PCI windows, and by
having the rest of the BARs autoconfigured in the outbound PCI
windows, in the range [e0000000..ffffffff], so that there is a 1:1
pci:phys mapping between them.
Even with this patch, a number of issues still remain -- just imagine
what happens if one of the CPUs is rebooted, by watchdog or by hand,
but the other one isn't.  But those issues are not easily fixable
given the strange PCI layout of this board and the behavior of the
boot loader shipped with the platform.

Signed-off-by: Lennert Buytenhek
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2656/1: Access permission bits are wrong for kernel XIP sections on...
George G. Davis [Fri, 29 Apr 2005 21:08:35 +0000 (22:08 +0100)]
[PATCH] ARM: 2656/1: Access permission bits are wrong for kernel XIP sections on ARMv6

Patch from George G. Davis

This patch is required for kernel XIP support on ARMv6 machines.  It ensures that the access permission bits for kernel XIP section descriptors are APX=1 and AP[1:0]=01, which is Kernel read-only/User no access permissions.  Prior to this change, kernel XIP section descriptor access permissions were set to Kernel no access/User no access on ARMv6 machines and the kernel would therefore hang upon entry to userspace when set_fs(USER_DS) was executed.

Signed-off-by: Steve Longerbeam
Signed-off-by: George G. Davis
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2649/1: Fix 'sparse -Wbitwise' warnings from MMIO macros
Olav Kongas [Fri, 29 Apr 2005 21:08:34 +0000 (22:08 +0100)]
[PATCH] ARM: 2649/1: Fix 'sparse -Wbitwise' warnings from MMIO macros

Patch from Olav Kongas

On ARM, the outX() and writeX() families of macros take the
result of cpu_to_leYY(), which is of restricted type __leYY,
and feed it to __raw_writeX(), which expect an argument of
unrestricted type. This results in 'sparse -Wbitwise'
warnings about incorrect types in assignments. Analogous
type mismatch warnings are issued for inX() and readX()
counterparts. The below patch resolves these warnings by
adding forced typecasts.

Signed-off-by: Olav Kongas
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2651/3: kernel helpers for NPTL support
Nicolas Pitre [Fri, 29 Apr 2005 21:08:33 +0000 (22:08 +0100)]
[PATCH] ARM: 2651/3: kernel helpers for NPTL support

Patch from Nicolas Pitre

This patch entirely reworks the kernel assistance for NPTL on ARM.
In particular this provides an efficient way to retrieve the TLS
value and perform atomic operations without any instruction emulation
nor special system call.  This even allows for pre ARMv6 binaries to
be forward compatible with SMP systems without any penalty.
The problematic and performance critical operations are performed
through segment of kernel provided user code reachable from user space
at a fixed address in kernel memory.  Those fixed entry points are
within the vector page so we basically get it for free as no extra
memory page is required and nothing else may be mapped at that
location anyway.
This is different from (but doesn't preclude) a full blown VDSO
implementation, however a VDSO would prevent some assembly tricks with
constants that allows for efficient branching to those code segments.
And since those code segments only use a few cycles before returning to
user code, the overhead of a VDSO far call would add a significant
overhead to such minimalistic operations.
The ARM_NR_set_tls syscall also changed number.  This is done for two
reasons:
1) this patch changes the way the TLS value was previously meant to be
   retrieved, therefore we ensure whatever library using the old way
   gets fixed (they only exist in private tree at the moment since the
   NPTL work is still progressing).
2) the previous number was allocated in a range causing an undefined
   instruction trap on kernels not supporting that syscall and it was
   determined that allocating it in a range returning -ENOSYS would be
   much nicer for libraries trying to determine if the feature is
   present or not.

Signed-off-by: Nicolas Pitre
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2655/1: ARM1136 SWP instruction abort handler fix
George G. Davis [Fri, 29 Apr 2005 21:08:33 +0000 (22:08 +0100)]
[PATCH] ARM: 2655/1: ARM1136 SWP instruction abort handler fix

Patch from George G. Davis

As noted in http://www.arm.com/linux/patch-2.6.9-arm1.gz, the "Faulty SWP instruction on 1136 doesn't set bit 11 in DFSR." So the v6_early_abort handler does not report the correct rd/wr direction for the SWP instruction which may result in SEGVS or hangs. In order to work around this problem, this patch merely updates the fix contained in the ARM Ltd. patch to use the macroised abort handler fixups.

Signed-off-by: George G. Davis
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2659/1: do not assign PCI I/O address zero on IXP2000
Lennert Buytenhek [Fri, 29 Apr 2005 20:58:16 +0000 (21:58 +0100)]
[PATCH] ARM: 2659/1: do not assign PCI I/O address zero on IXP2000

Patch from Lennert Buytenhek

Assigning the address zero to a PCI device BAR causes some part of the
PCI subsystem to believe that resource allocation for that BAR failed
due to resource conflicts, which will make attempts to enable the
device fail.  Work around this by assigning I/O addresses starting
from 00010000.
While we're at it, make the PCI I/O resource end at 0001ffff, since we
only have 64k of outbound I/O window on the IXP2000, and we don't do
bank switching.

Signed-off-by: Lennert Buytenhek
Signed-off-by: Deepak Saxena
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2658/1: start ixp2000 pci memory resource at 0xe0000000
Lennert Buytenhek [Fri, 29 Apr 2005 20:58:15 +0000 (21:58 +0100)]
[PATCH] ARM: 2658/1: start ixp2000 pci memory resource at 0xe0000000

Patch from Lennert Buytenhek

On the IXDP2800, the bootloader does an awful job of configuring
the PCI bus, so we make linux reconfigure everything.  Having a 1:1
pci:phys address mapping generally simplifies everything, so try to
allocate PCI addresses from the [e0000000..ffffffff] range, which is
the physical address range of the outbound PCI window on the IXP2000.
This does not affect any of the other IXP2000 platforms since they
all use their bootloader's PCI resource assignment.

Signed-off-by: Lennert Buytenhek
Signed-off-by: Deepak Saxena
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] ARM: 2657/1: export ixp2000_pci_config_addr
Lennert Buytenhek [Fri, 29 Apr 2005 20:58:15 +0000 (21:58 +0100)]
[PATCH] ARM: 2657/1: export ixp2000_pci_config_addr

Patch from Lennert Buytenhek

Export ixp2000_pci_config_addr, to be used by the IXDP2800 platform
setup code to coordinate booting the master and slave NPU.

Signed-off-by: Lennert Buytenhek
Signed-off-by: Deepak Saxena
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years agox86: make traps on 'iret' be debuggable in user space
Linus Torvalds [Fri, 29 Apr 2005 16:38:44 +0000 (09:38 -0700)]
x86: make traps on 'iret' be debuggable in user space

This makes a trap on the 'iret' that returns us to user space
cause a nice clean SIGSEGV, instead of just a hard (and silent)
exit.

That way a debugger can actually try to see what happened, and
we also properly notify everybody who might be interested about
us being gone.

This loses the error code, but tells the debugger what happened
with ILL_BADSTK in the siginfo.

19 years agoRemove bogus BUG() in kernel/exit.c
Linus Torvalds [Fri, 29 Apr 2005 16:37:07 +0000 (09:37 -0700)]
Remove bogus BUG() in kernel/exit.c

It's old sanity checking that may have been useful for debugging, but
is just bogus these days.

Noticed by Mattia Belletti.

19 years ago[PATCH] Add suspend method to cpufreq core
Benjamin Herrenschmidt [Fri, 29 Apr 2005 14:40:12 +0000 (07:40 -0700)]
[PATCH] Add suspend method to cpufreq core

In order to properly fix some issues with cpufreq vs. sleep on
PowerBooks, I had to add a suspend callback to the pmac_cpufreq driver.
I must force a switch to full speed before sleep and I switch back to
previous speed on resume.

I also added a driver flag to disable the warnings in suspend/resume
since it is expected in this case to have different speed (and I want it
to fixup the jiffies properly).

Signed-off-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] x86_64: fix PT_NOTE addition to IA32 vDSO
Roland McGrath [Fri, 29 Apr 2005 05:47:29 +0000 (22:47 -0700)]
[PATCH] x86_64: fix PT_NOTE addition to IA32 vDSO

The addition of the PT_NOTE didn't take in the x86_64 version of the i386
vDSO, because I forgot the linker script bit in that copy.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Do not sleep interruptible after socket connect failure
Steve French [Fri, 29 Apr 2005 05:41:11 +0000 (22:41 -0700)]
[PATCH] cifs: Do not sleep interruptible after socket connect failure

.. since it can be due to pending kill.

Update readme information to better describe cifs umount

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Do not init smb requests or block when sending requests
Steve French [Fri, 29 Apr 2005 05:41:11 +0000 (22:41 -0700)]
[PATCH] cifs: Do not init smb requests or block when sending requests

if cifsd thread is no longer running to demultixplex responses.

Do not send FindClose request when FindFirst failed without reaching end
of search.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: handle termination of cifs oplockd kernel thread
Steve French [Fri, 29 Apr 2005 05:41:10 +0000 (22:41 -0700)]
[PATCH] cifs: handle termination of cifs oplockd kernel thread

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Fix mapping of EMLINK case
Steve French [Fri, 29 Apr 2005 05:41:10 +0000 (22:41 -0700)]
[PATCH] cifs: Fix mapping of EMLINK case

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Handle case of multiple trans2 responses for one SMB request (part...
Steve French [Fri, 29 Apr 2005 05:41:10 +0000 (22:41 -0700)]
[PATCH] cifs: Handle case of multiple trans2 responses for one SMB request (part 2 of 2)

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: cleanup various long lines
Steve French [Fri, 29 Apr 2005 05:41:10 +0000 (22:41 -0700)]
[PATCH] cifs: cleanup various long lines

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Handle multiple response transact2 part 1 of 2
Steve French [Fri, 29 Apr 2005 05:41:09 +0000 (22:41 -0700)]
[PATCH] cifs: Handle multiple response transact2 part 1 of 2

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Ease memory pressure, do not use large buffers in byte range lock requests.
Steve French [Fri, 29 Apr 2005 05:41:09 +0000 (22:41 -0700)]
[PATCH] cifs: Ease memory pressure, do not use large buffers in byte range lock requests.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: missing semicolon from previous fix
Steve French [Fri, 29 Apr 2005 05:41:09 +0000 (22:41 -0700)]
[PATCH] cifs: missing semicolon from previous fix

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Better handle errors on second socket recv message call
Steve French [Fri, 29 Apr 2005 05:41:09 +0000 (22:41 -0700)]
[PATCH] cifs: Better handle errors on second socket recv message call

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: improve check for search entry going beyond end of SMB transact
Steve French [Fri, 29 Apr 2005 05:41:08 +0000 (22:41 -0700)]
[PATCH] cifs: improve check for search entry going beyond end of SMB transact

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Fix caching problem
Steve French [Fri, 29 Apr 2005 05:41:08 +0000 (22:41 -0700)]
[PATCH] cifs: Fix caching problem

pointed out by Dave Stahl and Vince Negri in which cifs can update the
last modify time on a server modified file without invalidating the
local cached data due to an intervening readdir.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: remove cifs_kcalloc and check for NULL return on kcalloc in session...
Steve French [Fri, 29 Apr 2005 05:41:08 +0000 (22:41 -0700)]
[PATCH] cifs: remove cifs_kcalloc and check for NULL return on kcalloc in session initialization

Suggested by: Adrian Bunk and Dave Miller

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Missing initialization for largeBuf flag left out of previous changeset
Steve French [Fri, 29 Apr 2005 05:41:07 +0000 (22:41 -0700)]
[PATCH] cifs: Missing initialization for largeBuf flag left out of previous changeset

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Do not use large smb buffers in response path
Steve French [Fri, 29 Apr 2005 05:41:07 +0000 (22:41 -0700)]
[PATCH] cifs: Do not use large smb buffers in response path

unless response is larger than 256 bytes.  This cuts more than 1/3 of
the large memory allocations that cifs does and should be a huge help to
memory pressure under stress.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: cleanup of ifdefs usage so it is more consistent
Steve French [Fri, 29 Apr 2005 05:41:07 +0000 (22:41 -0700)]
[PATCH] cifs: cleanup of ifdefs usage so it is more consistent

And fix to not needlessly send new POSIX QFSInfo when server does not
explicitly claim support for the new protocol extensions.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: CIFS ioctl needed by umount.cifs utility
Steve French [Fri, 29 Apr 2005 05:41:07 +0000 (22:41 -0700)]
[PATCH] cifs: CIFS ioctl needed by umount.cifs utility

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Do not interpret oplock break responses as responses to an unrelated...
Steve French [Fri, 29 Apr 2005 05:41:07 +0000 (22:41 -0700)]
[PATCH] cifs: Do not interpret oplock break responses as responses to an unrelated command

.. even if the multiplex ids match.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Fix PPC64 compile error
Steve French [Fri, 29 Apr 2005 05:41:06 +0000 (22:41 -0700)]
[PATCH] cifs: Fix PPC64 compile error

.. and do not double endian convert the special characters whem mounted
with mapchars mount parm.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: character mapping of special characters (part 3 of 3)
Steve French [Fri, 29 Apr 2005 05:41:06 +0000 (22:41 -0700)]
[PATCH] cifs: character mapping of special characters (part 3 of 3)

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: finish up of special character mapping capable unicode conversion routi...
Steve French [Fri, 29 Apr 2005 05:41:06 +0000 (22:41 -0700)]
[PATCH] cifs: finish up of special character mapping capable unicode conversion routine part 2 of 3

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: remove a few redundant null pointer checks, and cleanup misc source...
Steve French [Fri, 29 Apr 2005 05:41:05 +0000 (22:41 -0700)]
[PATCH] cifs: remove a few redundant null pointer checks, and cleanup misc source formatting

Mostly suggested by Jesper Juhl

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Add new mount parm mapchars
Steve French [Fri, 29 Apr 2005 05:41:05 +0000 (22:41 -0700)]
[PATCH] cifs: Add new mount parm mapchars

For handling seven special characters that shells use for filenames.

This first parts implements conversions from Unicode.

Signed-off-by: Steve French
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: fix rare oops in cifs_close
Steve French [Fri, 29 Apr 2005 05:41:05 +0000 (22:41 -0700)]
[PATCH] cifs: fix rare oops in cifs_close

Protect access to cifs file list in cifs_close path

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Fix multiuser packet signing to use the right sequence number and mac...
Steve French [Fri, 29 Apr 2005 05:41:05 +0000 (22:41 -0700)]
[PATCH] cifs: Fix multiuser packet signing to use the right sequence number and mac session key

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Enable ioctl support in POSIX extensions to handle lsattr
Steve French [Fri, 29 Apr 2005 05:41:04 +0000 (22:41 -0700)]
[PATCH] cifs: Enable ioctl support in POSIX extensions to handle lsattr

remove sparse warnings, unnecessary pad in QueryFileInfo and redundant
function define.

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Gracefully turn off serverino (when serverino is enabled on mount)
Steve French [Fri, 29 Apr 2005 05:41:04 +0000 (22:41 -0700)]
[PATCH] cifs: Gracefully turn off serverino (when serverino is enabled on mount)

Old servers such as NT4 do not support this level of FindFirst (and
retry with a lower infolevel)

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: add support for chattr/lsattr in new CIFS POSIX extensions
Steve French [Fri, 29 Apr 2005 05:41:04 +0000 (22:41 -0700)]
[PATCH] cifs: add support for chattr/lsattr in new CIFS POSIX extensions

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cifs: Only send POSIX ACL calls to server if server claims to support that...
Steve French [Fri, 29 Apr 2005 05:41:04 +0000 (22:41 -0700)]
[PATCH] cifs: Only send POSIX ACL calls to server if server claims to support that capability bit

Signed-off-by: Steve French (sfrench@us.ibm.com)
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoAutomatic merge of master.kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Thu, 28 Apr 2005 23:09:57 +0000 (16:09 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] Fix error recovery path for arch_setup_additional_pages
Roland McGrath [Thu, 28 Apr 2005 22:17:19 +0000 (15:17 -0700)]
[PATCH] Fix error recovery path for arch_setup_additional_pages

If arch_setup_additional_pages fails, the error path will do some double-frees.
This fixes it.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[NET]: /proc/net/stat/* header cleanup
Olaf Rempel [Thu, 28 Apr 2005 19:16:08 +0000 (12:16 -0700)]
[NET]: /proc/net/stat/* header cleanup

Signed-off-by: Olaf Rempel <razzor@kopf-tisch.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PKT_SCHED]: Fix range in psched_tod_diff() to 0..bound
Nicolas Dichtel [Thu, 28 Apr 2005 19:14:37 +0000 (12:14 -0700)]
[PKT_SCHED]: Fix range in psched_tod_diff() to 0..bound

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IPV6]: Incorrect permissions on route flush sysctl
Dave Jones [Thu, 28 Apr 2005 19:11:49 +0000 (12:11 -0700)]
[IPV6]: Incorrect permissions on route flush sysctl

On Mon, Apr 25, 2005 at 12:01:13PM -0400, Dave Jones wrote:
 > This has been brought up before.. http://lkml.org/lkml/2000/1/21/116
 > but didnt seem to get resolved.  This morning I got someone
 > file a bugzilla about it breaking sysctl(8).

And here's its ipv6 counterpart.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IPV4]: Incorrect permissions on route flush sysctl
Dave Jones [Thu, 28 Apr 2005 19:11:03 +0000 (12:11 -0700)]
[IPV4]: Incorrect permissions on route flush sysctl

This has been brought up before.. http://lkml.org/lkml/2000/1/21/116
but didnt seem to get resolved.  This morning I got someone
file a bugzilla about it breaking sysctl(8).

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Fix SCTP sendbuffer accouting.
Neil Horman [Thu, 28 Apr 2005 19:02:04 +0000 (12:02 -0700)]
[SCTP] Fix SCTP sendbuffer accouting.

- Include chunk and skb sizes in sendbuffer accounting.
- 2 policies are supported. 0: per socket accouting, 1: per association
  accounting

DaveM: I've made the default per-socket.

Signed-off-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Replace incorrect use of dev_alloc_skb with alloc_skb in sctp_packet_transmit().
Sridhar Samudrala [Thu, 28 Apr 2005 19:00:23 +0000 (12:00 -0700)]
[SCTP] Replace incorrect use of dev_alloc_skb with alloc_skb in sctp_packet_transmit().

Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Fix bug in sctp_init() error handling code.
Neil Horman [Thu, 28 Apr 2005 18:59:49 +0000 (11:59 -0700)]
[SCTP] Fix bug in sctp_init() error handling code.

Signed-off-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Use ipv6_addr_any() rather than ipv6_addr_type() in sctp_v6_is_any().
Brian Haley [Thu, 28 Apr 2005 18:59:16 +0000 (11:59 -0700)]
[SCTP] Use ipv6_addr_any() rather than ipv6_addr_type() in sctp_v6_is_any().

Signed-off-by: Brian Haley <Brian.Haley@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Implement Sec 2.41 of SCTP Implementers guide.
Jerome Forissier [Thu, 28 Apr 2005 18:58:43 +0000 (11:58 -0700)]
[SCTP] Implement Sec 2.41 of SCTP Implementers guide.

- Fixed sctp_vtag_verify_either() to comply with impguide 2.41 B) and C).
- Make sure vtag is reflected when T-bit is set in SHUTDOWN-COMPLETE sent
  due to an OOTB SHUTDOWN-ACK and in ABORT sent due to an OOTB packet.
- Do not set T-Bit in ABORT chunk in response to INIT.
- Fixed some comments to reflect the new meaning of the T-Bit.

Signed-off-by: Jerome Forissier <jerome.forissier@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP] Fix SCTP_ASSOCINFO getsockopt for 1-1 style
Vladislav Yasevich [Thu, 28 Apr 2005 18:57:54 +0000 (11:57 -0700)]
[SCTP] Fix SCTP_ASSOCINFO getsockopt for 1-1 style

Signed-off-by: Vladislav Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TG3]: Set SA_SAMPLE_RANDOM in request_irq() calls.
David S. Miller [Thu, 28 Apr 2005 18:33:20 +0000 (11:33 -0700)]
[TG3]: Set SA_SAMPLE_RANDOM in request_irq() calls.

Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] time interpolator: Fix settimeofday inaccuracy
Christoph Lameter [Thu, 28 Apr 2005 15:13:58 +0000 (08:13 -0700)]
[PATCH] time interpolator: Fix settimeofday inaccuracy

settimeofday will set the time a little bit too early on systems using
time interpolation since it subtracts the current interpolator offset
from the time. This used to be necessary with the code in 2.6.9 and earlier
but the new code resets the time interpolator after setting the time.
Thus the time is set too early and gettimeofday will return a time slightly
before the time specified with settimeofday if invoked immeditely after
settimeofday.

This removes the obsolete subtraction of the time interpolator offset
and makes settimeofday set the time accurately.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ARM: AMBA CLCD: X resolutions must be multiples of 16
Russell King [Thu, 28 Apr 2005 09:46:15 +0000 (10:46 +0100)]
[PATCH] ARM: AMBA CLCD: X resolutions must be multiples of 16

We ignore the bottom 4 bits of the X resolution, so we should
round X resolutions up to the nearest multiple of 16.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Fix AMBA CLCD fb driver for 32bpp
Russell King [Thu, 28 Apr 2005 09:43:52 +0000 (10:43 +0100)]
[PATCH] ARM: Fix AMBA CLCD fb driver for 32bpp

We were supporting 24bpp.  However, the pixel organisation in
memory was 0RGB, so it was 24bpp in 32bit words.  This means
we're actually supporting 32bpp and not 24bpp.

Also, add a check to ensure that we don't exceed the available
framebuffer when changing display resolutions.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Fix AMBA CLCD fb driver for 1bpp/STN mono panels
Russell King [Thu, 28 Apr 2005 09:38:19 +0000 (10:38 +0100)]
[PATCH] ARM: Fix AMBA CLCD fb driver for 1bpp/STN mono panels

Fix the AMBA CLCD framebuffer driver for 1bpp modes and STN
monochrome LCD panels.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ppc64: Fix return value of some vDSO calls
Benjamin Herrenschmidt [Thu, 28 Apr 2005 01:04:45 +0000 (18:04 -0700)]
[PATCH] ppc64: Fix return value of some vDSO calls

The ppc vDSO would not properly clear the return value for some calls,
which will be a problem when interfacing those calls with glibc. This
should be fixed before 2.6.12 is released (as it is the first kernel
with the ppc vDSO) so that we don't have to play with symbol versioning
and ugly workarounds.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] NFS4: Don't use __user with compat_uptr_t
David Howells [Wed, 27 Apr 2005 22:39:03 +0000 (15:39 -0700)]
[PATCH] NFS4: Don't use __user with compat_uptr_t

The attached patch removes __user from compat_uptr_t types in the NFS4 mount
32-bit->64-bit compatibility structures.

Signed-Off-By: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ARM: Add further explaination for clk_get()
Russell King [Wed, 27 Apr 2005 17:19:55 +0000 (18:19 +0100)]
[PATCH] ARM: Add further explaination for clk_get()

clk_get() comments can be confusing.  Add extra explaination of
the dev and id parameters to ensure correct usage.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years agoAutomatic merge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2...
Linus Torvalds [Wed, 27 Apr 2005 17:05:42 +0000 (10:05 -0700)]
Merge ... /linux/kernel/git/aegl/linux-2.6.git

19 years ago[PATCH] PC300 pci_enable_device fix
Marcelo Tosatti [Tue, 26 Apr 2005 20:09:35 +0000 (13:09 -0700)]
[PATCH] PC300 pci_enable_device fix

Call pci_enable_device() before looking at IRQ and resources,
and pci_disable_device() when shutting the interface down.

The driver requires this fix or the "pci=routeirq" workaround
on 2.6.10 and later kernels.

Reported and tested by Artur Lipowski.

From: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: trivial user annotations
Al Viro [Tue, 26 Apr 2005 18:26:53 +0000 (11:26 -0700)]
[PATCH] ppc64: trivial user annotations

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] size_t portability fixes - drivers/usb
Al Viro [Tue, 26 Apr 2005 18:26:53 +0000 (11:26 -0700)]
[PATCH] size_t portability fixes - drivers/usb

size_t is zu, ssize_t is zd...

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[IA64] re-enable preempt before page allocation for pgtable quicklist
Martin Hicks [Tue, 26 Apr 2005 16:04:31 +0000 (09:04 -0700)]
[IA64] re-enable preempt before page allocation for pgtable quicklist

This is a fix to the pgtable_quicklist code.  There is a GFP_KERNEL
allocation in pgtable_quicklist_alloc(), which spews the usual warnings
if the kernel is under heavy VM pressure and the reclaim code is
invoked.  re-enable preempt before we allocate the new page.

This patch is against 2.6.12-rc2-mm2

Signed-off-by: Martin Hicks <mort@sgi.com>
Signed-off-by: Tony Luck <tony.luckintel.com>
19 years agoAutomatic merge of kernel.org:/home/rmk/linux-2.6-serial.git
Linus Torvalds [Tue, 26 Apr 2005 16:02:40 +0000 (09:02 -0700)]
Merge ... /home/rmk/linux-2.6-serial.git

19 years agoAutomatic merge of kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Tue, 26 Apr 2005 15:58:22 +0000 (08:58 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git