SLUB: allocate smallest object size if the user asks for 0 bytes
Makes SLUB behave like SLAB in this area to avoid issues....
Throw a stack dump to alert people.
At some point the behavior should be switched back. NULL is no memory as
far as I can tell and if the use asked for 0 bytes then he need to get no
memory.
Signed-off-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Structures may contain u64 items on 32 bit platforms that are only able to
address 64 bit items on 64 bit boundaries. Change the mininum alignment of
slabs to conform to those expectations.
ARCH_KMALLOC_MINALIGN must be changed for good since a variety of structure
are mixed in the general slabs.
ARCH_SLAB_MINALIGN is changed because currently there is no consistent
specification of object alignment. We may have that in the future when the
KMEM_CACHE and related macros are used to generate slabs. These pass the
alignment of the structure generated by the compiler to the slab.
With KMEM_CACHE etc we could align structures that do not contain 64
bit values to 32 bit boundaries potentially saving some memory.
Signed-off-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is a new slab allocator which was motivated by the complexity of the
existing code in mm/slab.c. It attempts to address a variety of concerns
with the existing implementation.
A. Management of object queues
A particular concern was the complex management of the numerous object
queues in SLAB. SLUB has no such queues. Instead we dedicate a slab for
each allocating CPU and use objects from a slab directly instead of
queueing them up.
B. Storage overhead of object queues
SLAB Object queues exist per node, per CPU. The alien cache queue even
has a queue array that contain a queue for each processor on each
node. For very large systems the number of queues and the number of
objects that may be caught in those queues grows exponentially. On our
systems with 1k nodes / processors we have several gigabytes just tied up
for storing references to objects for those queues This does not include
the objects that could be on those queues. One fears that the whole
memory of the machine could one day be consumed by those queues.
C. SLAB meta data overhead
SLAB has overhead at the beginning of each slab. This means that data
cannot be naturally aligned at the beginning of a slab block. SLUB keeps
all meta data in the corresponding page_struct. Objects can be naturally
aligned in the slab. F.e. a 128 byte object will be aligned at 128 byte
boundaries and can fit tightly into a 4k page with no bytes left over.
SLAB cannot do this.
D. SLAB has a complex cache reaper
SLUB does not need a cache reaper for UP systems. On SMP systems
the per CPU slab may be pushed back into partial list but that
operation is simple and does not require an iteration over a list
of objects. SLAB expires per CPU, shared and alien object queues
during cache reaping which may cause strange hold offs.
E. SLAB has complex NUMA policy layer support
SLUB pushes NUMA policy handling into the page allocator. This means that
allocation is coarser (SLUB does interleave on a page level) but that
situation was also present before 2.6.13. SLABs application of
policies to individual slab objects allocated in SLAB is
certainly a performance concern due to the frequent references to
memory policies which may lead a sequence of objects to come from
one node after another. SLUB will get a slab full of objects
from one node and then will switch to the next.
F. Reduction of the size of partial slab lists
SLAB has per node partial lists. This means that over time a large
number of partial slabs may accumulate on those lists. These can
only be reused if allocator occur on specific nodes. SLUB has a global
pool of partial slabs and will consume slabs from that pool to
decrease fragmentation.
G. Tunables
SLAB has sophisticated tuning abilities for each slab cache. One can
manipulate the queue sizes in detail. However, filling the queues still
requires the uses of the spin lock to check out slabs. SLUB has a global
parameter (min_slab_order) for tuning. Increasing the minimum slab
order can decrease the locking overhead. The bigger the slab order the
less motions of pages between per CPU and partial lists occur and the
better SLUB will be scaling.
G. Slab merging
We often have slab caches with similar parameters. SLUB detects those
on boot up and merges them into the corresponding general caches. This
leads to more effective memory use. About 50% of all caches can
be eliminated through slab merging. This will also decrease
slab fragmentation because partial allocated slabs can be filled
up again. Slab merging can be switched off by specifying
slub_nomerge on boot up.
Note that merging can expose heretofore unknown bugs in the kernel
because corrupted objects may now be placed differently and corrupt
differing neighboring objects. Enable sanity checks to find those.
H. Diagnostics
The current slab diagnostics are difficult to use and require a
recompilation of the kernel. SLUB contains debugging code that
is always available (but is kept out of the hot code paths).
SLUB diagnostics can be enabled via the "slab_debug" option.
Parameters can be specified to select a single or a group of
slab caches for diagnostics. This means that the system is running
with the usual performance and it is much more likely that
race conditions can be reproduced.
I. Resiliency
If basic sanity checks are on then SLUB is capable of detecting
common error conditions and recover as best as possible to allow the
system to continue.
J. Tracing
Tracing can be enabled via the slab_debug=T,<slabcache> option
during boot. SLUB will then protocol all actions on that slabcache
and dump the object contents on free.
K. On demand DMA cache creation.
Generally DMA caches are not needed. If a kmalloc is used with
__GFP_DMA then just create this single slabcache that is needed.
For systems that have no ZONE_DMA requirement the support is
completely eliminated.
L. Performance increase
Some benchmarks have shown speed improvements on kernbench in the
range of 5-10%. The locking overhead of slub is based on the
underlying base allocation size. If we can reliably allocate
larger order pages then it is possible to increase slub
performance much further. The anti-fragmentation patches may
enable further performance increases.
Tested on:
i386 UP + SMP, x86_64 UP + SMP + NUMA emulation, IA64 NUMA + Simulator
SLUB Boot options
slub_nomerge Disable merging of slabs
slub_min_order=x Require a minimum order for slab caches. This
increases the managed chunk size and therefore
reduces meta data and locking overhead.
slub_min_objects=x Mininum objects per slab. Default is 8.
slub_max_order=x Avoid generating slabs larger than order specified.
slub_debug Enable all diagnostics for all caches
slub_debug=<options> Enable selective options for all caches
slub_debug=<o>,<cache> Enable selective options for a certain set of
caches
Available Debug options
F Double Free checking, sanity and resiliency
R Red zoning
P Object / padding poisoning
U Track last free / alloc
T Trace all allocs / frees (only use for individual slabs).
To use SLUB: Apply this patch and then select SLUB as the default slab
allocator.
[hugh@veritas.com: fix an oops-causing locking error]
[akpm@linux-foundation.org: various stupid cleanups and small fixes] Signed-off-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Andy Whitcroft [Sun, 6 May 2007 21:49:33 +0000 (14:49 -0700)]
tty_register_driver: only allocate tty instances when defined
If device->num is zero we attempt to kmalloc() zero bytes. When SLUB is
enabled this returns a null pointer and take that as an allocation failure
and fail the device register. Check for no devices and avoid the
allocation.
[akpm: opportunistic kzalloc() conversion] Signed-off-by: Andy Whitcroft <apw@shadowen.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
i386: use page allocator to allocate thread_info structure
i386 uses kmalloc to allocate the threadinfo structure assuming that the
allocations result in a page sized aligned allocation. That has worked so
far because SLAB exempts page sized slabs from debugging and aligns them in
special ways that goes beyond the restrictions imposed by
KMALLOC_ARCH_MINALIGN valid for other slabs in the kmalloc array.
SLUB also works fine without debugging since page sized allocations neatly
align at page boundaries. However, if debugging is switched on then SLUB
will extend the slab with debug information. The resulting slab is not
longer of page size. It will only be aligned following the requirements
imposed by KMALLOC_ARCH_MINALIGN. As a result the threadinfo structure may
not be page aligned which makes i386 fail to boot with SLUB debug on.
Replace the calls to kmalloc with calls into the page allocator.
An alternate solution may be to create a custom slab cache where the
alignment is set to PAGE_SIZE. That would allow slub debugging to be
applied to the threadinfo structure.
Signed-off-by: Christoph Lameter <clameter@sgi.com> Cc: William Lee Irwin III <wli@holomorphy.com> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Rientjes [Sun, 6 May 2007 21:49:32 +0000 (14:49 -0700)]
cpusets: allow TIF_MEMDIE threads to allocate anywhere
OOM killed tasks have access to memory reserves as specified by the
TIF_MEMDIE flag in the hopes that it will quickly exit. If such a task has
memory allocations constrained by cpusets, we may encounter a deadlock if a
blocking task cannot exit because it cannot allocate the necessary memory.
We allow tasks that have the TIF_MEMDIE flag to allocate memory anywhere,
including outside its cpuset restriction, so that it can quickly die
regardless of whether it is __GFP_HARDWALL.
Cc: Andi Kleen <ak@suse.de> Cc: Paul Jackson <pj@sgi.com> Cc: Christoph Lameter <clameter@engr.sgi.com> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Andrew Morton [Sun, 6 May 2007 21:49:31 +0000 (14:49 -0700)]
slab: mark set_up_list3s() __init
It is only ever used prior to free_initmem().
(It will cause a warning when we run the section checking, but that's a
false-positive and it simply changes the source of an existing warning, which
is also a false-positive)
Cc: Christoph Lameter <clameter@engr.sgi.com> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Mel Gorman [Sun, 6 May 2007 21:49:30 +0000 (14:49 -0700)]
Do not disable interrupts when reading min_free_kbytes
The sysctl handler for min_free_kbytes calls setup_per_zone_pages_min() on
read or write. This function iterates through every zone and calls
spin_lock_irqsave() on the zone LRU lock. When reading min_free_kbytes,
this is a total waste of time that disables interrupts on the local
processor. It might even be noticable machines with large numbers of zones
if a process started constantly reading min_free_kbytes.
This patch only calls setup_per_zone_pages_min() only on write. Tested on
an x86 laptop and it did the right thing.
Signed-off-by: Mel Gorman <mel@csn.ul.ie> Acked-by: Christoph Lameter <clameter@engr.sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Eric Dumazet [Sun, 6 May 2007 21:49:29 +0000 (14:49 -0700)]
slab: NUMA kmem_cache diet
Some NUMA machines have a big MAX_NUMNODES (possibly 1024), but fewer
possible nodes. This patch dynamically sizes the 'struct kmem_cache' to
allocate only needed space.
I moved nodelists[] field at the end of struct kmem_cache, and use the
following computation in kmem_cache_init()
On my two nodes x86_64 machine, kmem_cache.obj_size is now 192 instead of 704
(This is because on x86_64, MAX_NUMNODES is 64)
On bigger NUMA setups, this might reduce the gfporder of "cache_cache"
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Andy Whitcroft <apw@shadowen.org> Cc: Christoph Lameter <clameter@engr.sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Eric Dumazet [Sun, 6 May 2007 21:49:28 +0000 (14:49 -0700)]
SLAB: don't allocate empty shared caches
We can avoid allocating empty shared caches and avoid unecessary check of
cache->limit. We save some memory. We avoid bringing into CPU cache
unecessary cache lines.
All accesses to l3->shared are already checking NULL pointers so this patch is
safe.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Acked-by: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Christoph Lameter <clameter@engr.sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Eric Dumazet [Sun, 6 May 2007 21:49:27 +0000 (14:49 -0700)]
SLAB: use num_possible_cpus() in enable_cpucache()
The existing comment in mm/slab.c is *perfect*, so I reproduce it :
/*
* CPU bound tasks (e.g. network routing) can exhibit cpu bound
* allocation behaviour: Most allocs on one cpu, most free operations
* on another cpu. For these cases, an efficient object passing between
* cpus is necessary. This is provided by a shared array. The array
* replaces Bonwick's magazine layer.
* On uniprocessor, it's functionally equivalent (but less efficient)
* to a larger limit. Thus disabled by default.
*/
As most shiped linux kernels are now compiled with CONFIG_SMP, there is no way
a preprocessor #if can detect if the machine is UP or SMP. Better to use
num_possible_cpus().
This means on UP we allocate a 'size=0 shared array', to be more efficient.
Another patch can later avoid the allocations of 'empty shared arrays', to
save some memory.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Acked-by: Pekka Enberg <penberg@cs.helsinki.fi> Acked-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jan Kara [Sun, 6 May 2007 21:49:26 +0000 (14:49 -0700)]
readahead: code cleanup
Rename file_ra_state.prev_page to prev_index and file_ra_state.offset to
prev_offset. Also update of prev_index in do_generic_mapping_read() is now
moved close to the update of prev_offset.
[wfg@mail.ustc.edu.cn: fix it] Signed-off-by: Jan Kara <jack@suse.cz> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Cc: WU Fengguang <wfg@mail.ustc.edu.cn> Signed-off-by: Fengguang Wu <wfg@mail.ustc.edu.cn> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Introduce ra.offset and store in it an offset where the previous read
ended. This way we can detect whether reads are really sequential (and
thus we should not mark the page as accessed repeatedly) or whether they
are random and just happen to be in the same page (and the page should
really be marked accessed again).
Signed-off-by: Jan Kara <jack@suse.cz> Acked-by: Nick Piggin <nickpiggin@yahoo.com.au> Cc: WU Fengguang <wfg@mail.ustc.edu.cn> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Rientjes [Sun, 6 May 2007 21:49:24 +0000 (14:49 -0700)]
smaps: add clear_refs file to clear reference
Adds /proc/pid/clear_refs. When any non-zero number is written to this file,
pte_mkold() and ClearPageReferenced() is called for each pte and its
corresponding page, respectively, in that task's VMAs. This file is only
writable by the user who owns the task.
It is now possible to measure _approximately_ how much memory a task is using
by clearing the reference bits with
echo 1 > /proc/pid/clear_refs
and checking the reference count for each VMA from the /proc/pid/smaps output
at a measured time interval. For example, to observe the approximate change
in memory footprint for a task, write a script that clears the references
(echo 1 > /proc/pid/clear_refs), sleeps, and then greps for Pgs_Referenced and
extracts the size in kB. Add the sizes for each VMA together for the total
referenced footprint. Moments later, repeat the process and observe the
difference.
For example, using an efficient Mozilla:
accumulated time referenced memory
---------------- -----------------
0 s 408 kB
1 s 408 kB
2 s 556 kB
3 s 1028 kB
4 s 872 kB
5 s 1956 kB
6 s 416 kB
7 s 1560 kB
8 s 2336 kB
9 s 1044 kB
10 s 416 kB
This is a valuable tool to get an approximate measurement of the memory
footprint for a task.
Cc: Hugh Dickins <hugh@veritas.com> Cc: Paul Mundt <lethal@linux-sh.org> Cc: Christoph Lameter <clameter@sgi.com> Signed-off-by: David Rientjes <rientjes@google.com>
[akpm@linux-foundation.org: build fixes]
[mpm@selenic.com: rename for_each_pmd] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Rientjes [Sun, 6 May 2007 21:49:22 +0000 (14:49 -0700)]
smaps: add pages referenced count to smaps
Adds an additional unsigned long field to struct mem_size_stats called
'referenced'. For each pte walked in the smaps code, this field is
incremented by PAGE_SIZE if it has pte-reference bits.
An additional line was added to the /proc/pid/smaps output for each VMA to
indicate how many pages within it are currently marked as referenced or
accessed.
Cc: Hugh Dickins <hugh@veritas.com> Cc: Paul Mundt <lethal@linux-sh.org> Cc: Christoph Lameter <clameter@sgi.com> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Rientjes [Sun, 6 May 2007 21:49:21 +0000 (14:49 -0700)]
smaps: extract pmd walker from smaps code
Extracts the pmd walker from smaps-specific code in fs/proc/task_mmu.c.
The new struct pmd_walker includes the struct vm_area_struct of the memory to
walk over. Iteration begins at the vma->vm_start and completes at
vma->vm_end. A pointer to another data structure may be stored in the private
field such as struct mem_size_stats, which acts as the smaps accumulator. For
each pmd in the VMA, the action function is called with a pointer to its
struct vm_area_struct, a pointer to the pmd_t, its start and end addresses,
and the private field.
The interface for walking pmd's in a VMA for fs/proc/task_mmu.c is now:
void for_each_pmd(struct vm_area_struct *vma,
void (*action)(struct vm_area_struct *vma,
pmd_t *pmd, unsigned long addr,
unsigned long end,
void *private),
void *private);
Since the pmd walker is now extracted from the smaps code, smaps_one_pmd() is
invoked for each pmd in the VMA. Its behavior and efficiency is identical to
the existing implementation.
Cc: Hugh Dickins <hugh@veritas.com> Cc: Paul Mundt <lethal@linux-sh.org> Cc: Christoph Lameter <clameter@sgi.com> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Zachary Amsden [Sun, 6 May 2007 21:49:20 +0000 (14:49 -0700)]
i386: use pte_update_defer in ptep_test_and_clear_{dirty,young}
If you actually clear the bit, you need to:
+ pte_update_defer(vma->vm_mm, addr, ptep);
The reason is, when updating PTEs, the hypervisor must be notified. Using
atomic operations to do this is fine for all hypervisors I am aware of.
However, for hypervisors which shadow page tables, if these PTE
modifications are not trapped, you need a post-modification call to fulfill
the update of the shadow page table.
Acked-by: Zachary Amsden <zach@vmware.com> Cc: Hugh Dickins <hugh@veritas.com> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Rientjes [Sun, 6 May 2007 21:49:19 +0000 (14:49 -0700)]
i386: add ptep_test_and_clear_{dirty,young}
Add ptep_test_and_clear_{dirty,young} to i386. They advertise that they
have it and there is at least one place where it needs to be called without
the page table lock: to clear the accessed bit on write to
/proc/pid/clear_refs.
ptep_clear_flush_{dirty,young} are updated to use the new functions. The
overall net effect to current users of ptep_clear_flush_{dirty,young} is
that we introduce an additional branch.
Andy Whitcroft [Sun, 6 May 2007 21:49:14 +0000 (14:49 -0700)]
add pfn_valid_within helper for sub-MAX_ORDER hole detection
Generally we work under the assumption that memory the mem_map array is
contigious and valid out to MAX_ORDER_NR_PAGES block of pages, ie. that if we
have validated any page within this MAX_ORDER_NR_PAGES block we need not check
any other. This is not true when CONFIG_HOLES_IN_ZONE is set and we must
check each and every reference we make from a pfn.
Add a pfn_valid_within() helper which should be used when scanning pages
within a MAX_ORDER_NR_PAGES block when we have already checked the validility
of the block normally with pfn_valid(). This can then be optimised away when
we do not have holes within a MAX_ORDER_NR_PAGES block of pages.
Signed-off-by: Andy Whitcroft <apw@shadowen.org> Acked-by: Mel Gorman <mel@csn.ul.ie> Acked-by: Bob Picco <bob.picco@hp.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Heiko Carstens [Sun, 6 May 2007 21:49:09 +0000 (14:49 -0700)]
Introduce CONFIG_HAS_DMA
Architectures that don't support DMA can say so by adding a config NO_DMA
to their Kconfig file. This will prevent compilation of some dma specific
driver code. Also dma-mapping-broken.h isn't needed anymore on at least
s390. This avoids compilation and linking of otherwise dead/broken code.
Other architectures that include dma-mapping-broken.h are arm26, h8300,
m68k, m68knommu and v850. If these could be converted as well we could get
rid of the header file.
If the badness of a process is zero then oom_adj>0 has no effect. This
patch makes sure that the oom_adj shift actually increases badness points
appropriately.
Signed-off-by: Joshua N. Pritikin <jpritikin@pobox.com> Cc: Andrea Arcangeli <andrea@novell.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Nick Piggin [Sun, 6 May 2007 21:49:05 +0000 (14:49 -0700)]
fs: buffer don't PageUptodate without page locked
__block_write_full_page is calling SetPageUptodate without the page locked.
This is unusual, but not incorrect, as PG_writeback is still set.
However the next patch will require that SetPageUptodate always be called with
the page locked. Simply don't bother setting the page uptodate in this case
(it is unusual that the write path does such a thing anyway). Instead just
leave it to the read side to bring the page uptodate when it notices that all
buffers are uptodate.
Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Nick Piggin [Sun, 6 May 2007 21:49:04 +0000 (14:49 -0700)]
mm: make read_cache_page synchronous
Ensure pages are uptodate after returning from read_cache_page, which allows
us to cut out most of the filesystem-internal PageUptodate calls.
I didn't have a great look down the call chains, but this appears to fixes 7
possible use-before uptodate in hfs, 2 in hfsplus, 1 in jfs, a few in
ecryptfs, 1 in jffs2, and a possible cleared data overwritten with readpage in
block2mtd. All depending on whether the filler is async and/or can return
with a !uptodate page.
Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pekka Enberg [Sun, 6 May 2007 21:49:03 +0000 (14:49 -0700)]
slab: ensure cache_alloc_refill terminates
If slab->inuse is corrupted, cache_alloc_refill can enter an infinite
loop as detailed by Michael Richardson in the following post:
<http://lkml.org/lkml/2007/2/16/292>. This adds a BUG_ON to catch
those cases.
Cc: Michael Richardson <mcr@sandelman.ca> Acked-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use ZVC counters to establish exact size of dirtyable pages
We can use the global ZVC counters to establish the exact size of the LRU
and the free pages. This allows a more accurate determination of the dirty
ratio.
This patch will fix the broken ratio calculations if large amounts of
memory are allocated to huge pags or other consumers that do not put the
pages on to the LRU.
Notes:
- I did not add NR_SLAB_RECLAIMABLE to the calculation of the
dirtyable pages. Those may be reclaimable but they are at this
point not dirtyable. If NR_SLAB_RECLAIMABLE would be considered
then a huge number of reclaimable pages would stop writeback
from occurring.
- This patch used to be in mm as the last one in a series of patches.
It was removed when Linus updated the treatment of highmem because
there was a conflict. I updated the patch to follow Linus' approach.
This patch is neede to fulfill the claims made in the beginning of the
patchset that is now in Linus' tree.
Signed-off-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Safer nr_node_ids and nr_node_ids determination and initial values
The nr_cpu_ids value is currently only calculated in smp_init. However, it
may be needed before (SLUB needs it on kmem_cache_init!) and other kernel
components may also want to allocate dynamically sized per cpu array before
smp_init. So move the determination of possible cpus into sched_init()
where we already loop over all possible cpus early in boot.
Also initialize both nr_node_ids and nr_cpu_ids with the highest value they
could take. If we have accidental users before these values are determined
then the current valud of 0 may cause too small per cpu and per node arrays
to be allocated. If it is set to the maximum possible then we only waste
some memory for early boot users.
Signed-off-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add apply_to_page_range() which applies a function to a pte range
Add a new mm function apply_to_page_range() which applies a given function to
every pte in a given virtual address range in a given mm structure. This is a
generic alternative to cut-and-pasting the Linux idiomatic pagetable walking
code in every place that a sequence of PTEs must be accessed.
Although this interface is intended to be useful in a wide range of
situations, it is currently used specifically by several Xen subsystems, for
example: to ensure that pagetables have been allocated for a virtual address
range, and to construct batched special pagetable update requests to map I/O
memory (in ioremap()).
[akpm@linux-foundation.org: fix warning, unpleasantly] Signed-off-by: Ian Pratt <ian.pratt@xensource.com> Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk> Signed-off-by: Chris Wright <chrisw@sous-sol.org> Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Cc: Christoph Lameter <clameter@sgi.com> Cc: Matt Mackall <mpm@waste.org> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Dave Jiang [Sun, 6 May 2007 21:48:50 +0000 (14:48 -0700)]
MPSC serial driver tx locking
The MPSC serial driver assumes that interrupt is always on to pick up the
DMA transmit ops that aren't submitted while the DMA engine is active.
However when irqs are off for a period of time such as operations under
kernel crash dump console messages do not show up due to additional DMA ops
are being dropped. This makes console writes to process through all the tx
DMAs queued up before submitting a new request.
Also, the current locking mechanism does not protect the hardware registers
and ring buffer when a printk is done during the serial write operations.
The additional per port transmit lock provides a finer granular locking and
protects registers being clobbered while printks are nested within UART
writes.
Signed-off-by: Dave Jiang <djiang@mvista.com> Signed-off-by: Mark A. Greer <mgreer@mvista.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Gibson [Sun, 6 May 2007 21:48:49 +0000 (14:48 -0700)]
serial: define FIXED_PORT flag for serial_core
At present, the serial core always allows setserial in userspace to change the
port address, irq and base clock of any serial port. That makes sense for
legacy ISA ports, but not for (say) embedded ns16550 compatible serial ports
at peculiar addresses. In these cases, the kernel code configuring the ports
must know exactly where they are, and their clocking arrangements (which can
be unusual on embedded boards). It doesn't make sense for userspace to change
these settings.
Therefore, this patch defines a UPF_FIXED_PORT flag for the uart_port
structure. If this flag is set when the serial port is configured, any
attempts to alter the port's type, io address, irq or base clock with
setserial are ignored.
In addition this patch uses the new flag for on-chip serial ports probed in
arch/powerpc/kernel/legacy_serial.c, and for other hard-wired serial ports
probed by drivers/serial/of_serial.c.
Signed-off-by: David Gibson <dwg@au1.ibm.com> Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Marc St-Jean [Sun, 6 May 2007 21:48:45 +0000 (14:48 -0700)]
serial driver PMC MSP71xx
Serial driver patch for the PMC-Sierra MSP71xx devices.
There are three different fixes:
1 Fix for DesignWare APB THRE errata: In brief, this is a non-standard
16550 in that the THRE interrupt will not re-assert itself simply by
disabling and re-enabling the THRI bit in the IER, it is only re-enabled
if a character is actually sent out.
It appears that the "8250-uart-backup-timer.patch" in the "mm" tree
also fixes it so we have dropped our initial workaround. This patch now
needs to be applied on top of that "mm" patch.
2 Fix for Busy Detect on LCR write: The DesignWare APB UART has a feature
which causes a new Busy Detect interrupt to be generated if it's busy
when the LCR is written. This fix saves the value of the LCR and
rewrites it after clearing the interrupt.
3 Workaround for interrupt/data concurrency issue: The SoC needs to
ensure that writes that can cause interrupts to be cleared reach the UART
before returning from the ISR. This fix reads a non-destructive register
on the UART so the read transaction completion ensures the previously
queued write transaction has also completed.
Signed-off-by: Marc St-Jean <Marc_St-Jean@pmc-sierra.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Bernhard Walle [Sun, 6 May 2007 21:48:44 +0000 (14:48 -0700)]
add new_id to PCMCIA drivers
PCI drivers have the new_id file in sysfs which allows new IDs to be added
at runtime. The advantage is to avoid re-compilation of a driver that
works for a new device, but it's ID table doesn't contain the new device.
This mechanism is only meant for testing, after the driver has been tested
successfully, the ID should be added in source code so that new revisions
of the kernel automatically detect the device.
The implementation follows the PCI implementation. The interface is documented
in Documentation/pcmcia/driver.txt. Computations should be done in userspace,
so the sysfs string contains the raw structure members for matching.
Signed-off-by: Bernhard Walle <bwalle@suse.de> Cc: Dominik Brodowski <linux@dominikbrodowski.net> Cc: Greg KH <greg@kroah.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Brownell [Sun, 6 May 2007 21:48:42 +0000 (14:48 -0700)]
at91_cf, minor fix
This is a minor correctness fix: since the at91_cf driver probe() routine
is in the init section, it should use platform_driver_probe() instead of
leaving that pointer around in the driver struct after init section
removal.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Dominik Brodowski <linux@dominikbrodowski.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pekka Enberg [Sun, 6 May 2007 21:48:40 +0000 (14:48 -0700)]
slab: introduce krealloc
This introduce krealloc() that reallocates memory while keeping the contents
unchanged. The allocator avoids reallocation if the new size fits the
currently used cache. I also added a simple non-optimized version for
mm/slob.c for compatibility.
[akpm@linux-foundation.org: fix warnings] Acked-by: Josef Sipek <jsipek@fsl.cs.sunysb.edu> Acked-by: Matt Mackall <mpm@selenic.com> Acked-by: Christoph Lameter <clameter@sgi.com> Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Mon, 7 May 2007 15:44:24 +0000 (08:44 -0700)]
Revert "[PATCH] x86: __pa and __pa_symbol address space separation"
This was broken. It adds complexity, for no good reason. Rather than
separate __pa() and __pa_symbol(), we should deprecate __pa_symbol(),
and preferably __pa() too - and just use "virt_to_phys()" instead, which
is more readable and has nicer semantics.
However, right now, just undo the separation, and make __pa_symbol() be
the exact same as __pa(). That fixes the bugs this patch introduced,
and we can do the fairly obvious cleanups later.
Do the new __phys_addr() function (which is now the actual workhorse for
the unified __pa()/__pa_symbol()) as a real external function, that way
all the potential issues with compile/link-time optimizations of
constant symbol addresses go away, and we can also, if we choose to, add
more sanity-checking of the argument.
Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Vivek Goyal <vgoyal@in.ibm.com> Cc: Andi Kleen <ak@suse.de> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild: (38 commits)
kconfig: fix mconf segmentation fault
kbuild: enable use of code from a different dir
kconfig: error out if recursive dependencies are found
kbuild: scripts/basic/fixdep segfault on pathological string-o-death
kconfig: correct minor typo in Kconfig warning message.
kconfig: fix path to modules.txt in Kconfig help
usr/Kconfig: fix typo
kernel-doc: alphabetically-sorted entries in index.html of 'htmldocs'
kbuild: be more explicit on missing .config file
kbuild: clarify the creation of the LOCALVERSION_AUTO string.
kbuild: propagate errors from find in scripts/gen_initramfs_list.sh
kconfig: refer to qt3 if we cannot find qt libraries
kbuild: handle compressed cpio initramfs-es
kbuild: ignore section mismatch warning for references from .paravirtprobe to .init.text
kbuild: remove stale comment in modpost.c
kbuild/mkuboot.sh: allow spaces in CROSS_COMPILE
kbuild: fix make mrproper for Documentation/DocBook/man
kbuild: remove kconfig binaries during make mrproper
kconfig/menuconfig: do not hardcode '.config'
kbuild: override build timestamp & version
...
Linus Torvalds [Sun, 6 May 2007 20:21:18 +0000 (13:21 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm: (66 commits)
KVM: Remove unused 'instruction_length'
KVM: Don't require explicit indication of completion of mmio or pio
KVM: Remove extraneous guest entry on mmio read
KVM: SVM: Only save/restore MSRs when needed
KVM: fix an if() condition
KVM: VMX: Add lazy FPU support for VT
KVM: VMX: Properly shadow the CR0 register in the vcpu struct
KVM: Don't complain about cpu erratum AA15
KVM: Lazy FPU support for SVM
KVM: Allow passing 64-bit values to the emulated read/write API
KVM: Per-vcpu statistics
KVM: VMX: Avoid unnecessary vcpu_load()/vcpu_put() cycles
KVM: MMU: Avoid heavy ASSERT at non debug mode.
KVM: VMX: Only save/restore MSR_K6_STAR if necessary
KVM: Fold drivers/kvm/kvm_vmx.h into drivers/kvm/vmx.c
KVM: VMX: Don't switch 64-bit msrs for 32-bit guests
KVM: VMX: Reduce unnecessary saving of host msrs
KVM: Handle guest page faults when emulating mmio
KVM: SVM: Report hardware exit reason to userspace instead of dmesg
KVM: Retry sleeping allocation if atomic allocation fails
...
Sam Ravnborg [Sun, 6 May 2007 07:23:45 +0000 (09:23 +0200)]
kbuild: enable use of code from a different dir
To introduce support for source in one directory but output files
in another directory during a non O= build prefix all paths
with $(src) repsectively $(obj).
* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
[CIFS] Fix typo in cifs readme from previous commit
[CIFS] Make sec=none force an anonymous mount
[CIFS] Change semaphore to mutex for cifs lock_sem
[CIFS] Fix oops in reset_cifs_unix_caps on reconnect
[CIFS] UID/GID override on CIFS mounts to Samba
[CIFS] prefixpath mounts to servers supporting posix paths used wrong slash
[CIFS] Update cifs version to 1.49
[CIFS] Replace kmalloc/memset combination with kzalloc
[CIFS] Add IPv6 support
[CIFS] New CIFS POSIX mkdir performance improvement (part 2)
[CIFS] New CIFS POSIX mkdir performance improvement
[CIFS] Add write perm for usr to file on windows should remove r/o dos attr
[CIFS] Remove unnecessary parm to cifs_reopen_file
[CIFS] Switch cifsd to kthread_run from kernel_thread
[CIFS] Remove unnecessary checks
Linus Torvalds [Sat, 5 May 2007 21:55:20 +0000 (14:55 -0700)]
Merge branch 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6
* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6: (231 commits)
[PATCH] i386: Don't delete cpu_devs data to identify different x86 types in late_initcall
[PATCH] i386: type may be unused
[PATCH] i386: Some additional chipset register values validation.
[PATCH] i386: Add missing !X86_PAE dependincy to the 2G/2G split.
[PATCH] x86-64: Don't exclude asm-offsets.c in Documentation/dontdiff
[PATCH] i386: avoid redundant preempt_disable in __unlazy_fpu
[PATCH] i386: white space fixes in i387.h
[PATCH] i386: Drop noisy e820 debugging printks
[PATCH] x86-64: Fix allnoconfig error in genapic_flat.c
[PATCH] x86-64: Shut up warnings for vfat compat ioctls on other file systems
[PATCH] x86-64: Share identical video.S between i386 and x86-64
[PATCH] x86-64: Remove CONFIG_REORDER
[PATCH] x86-64: Print type and size correctly for unknown compat ioctls
[PATCH] i386: Remove copy_*_user BUG_ONs for (size < 0)
[PATCH] i386: Little cleanups in smpboot.c
[PATCH] x86-64: Don't enable NUMA for a single node in K8 NUMA scanning
[PATCH] x86: Use RDTSCP for synchronous get_cycles if possible
[PATCH] i386: Add X86_FEATURE_RDTSCP
[PATCH] i386: Implement X86_FEATURE_SYNC_RDTSC on i386
[PATCH] i386: Implement alternative_io for i386
...
Fix up trivial conflict in include/linux/highmem.h manually.
* master.kernel.org:/pub/scm/linux/kernel/git/jejb/voyager-2.6:
[VOYAGER] add smp alternatives
[VOYAGER] Use modern techniques to setup and teardown low identiy mappings.
[VOYAGER] Convert the monitor thread to use the kthread API
[VOYAGER] clockevents driver: bring voyager in to line
[VOYAGER] clockevents: correct boot cpu is zero assumption
[VOYAGER] add smp_call_function_single
The S3C2410_UDC_SETIX() macro is not used and won't be used by the udc
driver, so delete it.
Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org> Signed-off-by: Ben Dooks <ben-linux@fluff.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
linux/mmc/protocol.h header is gone, thus breaking the build of the
mach-qt2410.c file. As this header is not used, I'm removing it. The
right headers may still be added later if needed.
Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org> Signed-off-by: Ben Dooks <ben-linux@fluff.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This driver doesn't support SWDMA so use the correct ->swdma_mask.
While at it:
* no need to call config_chipset_for_pio() in config_chipset_for_dma(),
if DMA is not available config_chipset_for_pio() will be called
by siimage_config_drive_for_dma() and if DMA is available
config_siimage_chipset_for_pio() will be called by siimage_tune_chipset()
* limit max PIO mode to PIO4, this driver doesn't support PIO5 and attempt
to setup PIO5 by it821x_tuneproc() could result in incorrect PIO timings
+ incorrect base clock being set for controller in the passthrough mode
* move code limiting max PIO according to the pair device capabilities from
config_it821x_chipset_for_pio() to it821x_tuneproc() so the check is also
applied for mode change requests coming through ->tuneproc and ->speedproc
interfaces
* set device speed in it821x_tuneproc()
* in it821x_tune_chipset() call it821x_tuneproc() also if the controller is
in the smart mode (so the check for pair device max PIO is done)
* rename it821x_tuneproc() to it821x_tune_pio(), then add it821x_tuneproc()
wrapper which does the max PIO mode check; it worked by the pure luck
previously, pio[4] and pio_want[4] arrays were used with index == 255
so random PIO timings and base clock were set for the controller in the
passthrough mode, thankfully PIO timings and base clock were corrected
later by config_it821x_chipset_for_pio() call (but it was not called for
PIO-only devices during resume and for user requested PIO autotuning)
* remove config_it821x_chipset_for_pio() call from config_chipset_for_dma()
as the driver sets ->autotune to 1 and ->tuneproc does the proper job now
* convert the last user of config_it821x_chipset_for_pio() to use
it821x_tuneproc(drive, 255) and remove no longer needed function
Sergei Shtylyov [Sat, 5 May 2007 20:03:50 +0000 (22:03 +0200)]
aec62xx: fix PIO/DMA setup issues
Teach the driver's tuneproc() method to do PIO auto-runing properly since it
treated 5 instead of 255 as auto-tune request, and also passed the mode limit
of PIO5 to ide_get_best_pio_mode() despite supporting up to PIO4 only.
While at it, also:
- remove the driver's wrong claim about supporting SWDMA modes;
- stop hooking ide_dma_timeout() method as the handler clearly doesn't fit for
the task...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:50 +0000 (22:03 +0200)]
cmd64x: use interrupt status from MRDMODE register (take 2)
Fold the parts of the ide_dma_end() methods identical to __ide_dma_end() into a
mere call to it.
Start using faster versions of the ide_dma_end() and ide_dma_test_irq() methods
for the PCI0646U and newer chips that have the duplicate interrupt status bits
in the I/O mapped MRDMODE register, determing what methods to use at the driver
load time. Do some cleanup/renaming in the "old" ide_dma_test_irq() method too.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:50 +0000 (22:03 +0200)]
cmd64x: procfs code fixes/cleanups (take 2)
Fix several issues with the driver's procfs output:
- when testing if channel is enabled, the code looks at the "simplex" bits, not
at the real enable bits -- add #define for the primary channel enable bit;
- UltraDMA modes 0, 1, 3 for slave drive reported incorrectly due to using the
master drive's clock cycle resolution bit.
While at it, also perform the following cleanups:
- don't print extra newline before the first controller's dump;
- correct the chipset names (from CMDxxx to PCI-xxx)
- don't read from the registers which aren't used for dump;
- better align the table column sizes;
- rework UltraDMA mode dump code;
- remove PIO mode dump code that has never been finished;
- remove the duplicate interrupt status (the MRDMODE register bits mirror those
those in the CFR and ARTTIM23 registers) and fold the dump into single line;
- correct the style of the ?: operators...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
cmd64x: add/fix enablebits (take 2)
The IDE core looks at the wrong bit when checking if the secondary channel is
enabled on PCI0646 -- CNTRL register bit 7 is read-ahead disable, bit 3 is the
correct one.
Starting with PCI0646U chip, the primary channel can also be enabled/disabled --
so, add 'enablebits' initializers to each 'ide_pci_device_t' structure, handling
the original PCI0646 via adding the init_setup() method and clearing the 'reg'
field there if necessary...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
cmd64x: interrupt status fixes (take 2)
The driver's ide_dma_test_irq() method was reading the MRDMODE register even on
PCI0643/6 where it was write-only -- fix this by always reading the "backward-
compatible" interrupt bits, renaming dma_alt_stat to irq_stat as the interrupt
status bits are not coupled to DMA.
In addition, wrong interrupt bit was tested/cleared for the primary channel --
it's bit 2 in all the chip specs and the driver used bit 1... :-/
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
cmd64x: fix multiword and remove single-word DMA support
Fix the multiword DMA and drop the single-word DMA support (which nobody will
miss, I think). In order to do it, a number of changes was necessary:
- rename program_drive_counts() to program_cycle_times(), pass to it cycle's
total/active times instead of the clock counts, and convert them into the
active/recovery clocks there instead of cmd64x_tune_pio() -- this causes
quantize_timing() to also move;
- contrarywise, move all the code handling the address setup timing into
cmd64x_tune_pio(), so that setting MWDMA mode wouldn't change address setup;
- remove from the speedproc() method the bogus code pretending to set the DMA
timings by twiddling bits in the BMIDE status register, handle setting MWDMA
by just calling program_cycle_times(); while at it, improve the style of that
whole switch statement;
- stop fiddling with the DMA capable bits in the speedproc() method -- they do
not enable DMA, and are properly dealt with by the dma_host_{on,off} methods;
- don't set hwif->swdma_mask in the init_hwif() method anymore.
In addition to those changes, do the following:
- in cmd64x_tune_pio(), when writing to ARTTIM23 register preserve the interrupt
status bit, eliminate local_irq_{save|restore}() around this code as there's
*no* actual race with the interrupt handler, and move cmdprintk() to a more
fitting place -- after ide_get_best_pio_mode() call;
- make {arttim|drwtim}_regs arrays single-dimensional, indexed with drive->dn;
- rename {setup|recovery}_counts[] into more fitting {setup|recovery}_values[];
- in the speedproc() method, get rid of the duplicate reads/writes from/to the
UDIDETCRx registers and of the extra variable used to store the transfer mode
value after filtering, use another method of determining master/slave drive,
and cleanup useless parens;
- beautify cmdprintk() output here and there.
While at it, remove meaningless comment about the driver being used only on
UltraSPARC and long non-relevant RCS tag. :-)
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
sl82c105: DMA support code cleanup (take 4)
Fold the now equivalent code in the ide_dma_check() method into a mere call to
ide_use_dma(). Make config_for_dma() return non-zero if DMA mode has been set
and call it from the ide_dma_check() method instead of ide_dma_on().
Defer writing the DMA timings to the chip registers until DMA is really turned
on (and do not enable IORDY for DMA).
Remove unneeded code from the init_hwif() method, improve its overall looks.
Rename the dma_start(), ide_dma_check(), and ide_dma_lostirq() methods, and
also use more proper hwif->dma_command, fix printk() and comment in the latter
one as well. While at it, cleanup style in several places.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
sl82c105: rework PIO support (take 2)
Get rid of the 'pio_speed' member of 'ide_drive_t' that was only used by this
driver by storing the PIO mode timings in the 'drive_data' instead -- this
allows us to greatly simplify the process of "reloading" of the chip's timing
register and do it right in sl82c150_dma_off_quietly() and to get rid of two
extra arguments to config_for_pio() -- which got renamed to sl82c105_tune_pio()
and now returns a PIO mode selected, with ide_config_drive_speed() call moved
into the tuneproc() method, now called sl82c105_tune_drive() with the code to
set drive's 'io_32bit' and 'unmask' flags in its turn moved to its proper place
in the init_hwif() method.
Also, while at it, rename get_timing_sl82c105() into get_pio_timings() and get
rid of the code in it clamping cycle counts to 32 which was both incorrect and
never executed anyway...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Russell King [Sat, 5 May 2007 19:59:27 +0000 (20:59 +0100)]
[ARM] mm 10: allow memory type to be specified with ioremap
__ioremap() took a set of page table flags (specifically the cacheable
and bufferable bits) to control the mapping type. However, with
the advent of ARMv6, this is far too limited.
Replace the page table flags with a memory type index, so that the
desired attributes can be selected from the mem_type table.
Finally, to prevent silent miscompilation due to the differing
arguments, rename the __ioremap() and __ioremap_pfn() functions.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Matt Carlson [Sat, 5 May 2007 19:47:25 +0000 (12:47 -0700)]
[TG3]: Eliminate the TG3_FLAG_5701_REG_WRITE_BUG flag.
This patch removes the use of the TG3_FLAG_5701_REG_WRITE_BUG flag.
It's logic is only used to set a function pointer and thus the
logic can be collapsed and the flag removed.
[ Comment tidy by Christoph Hellwig. -DaveM ]
Signed-off-by: Matt Carlson <mcarlson@broadcom.com> Signed-off-by: Michael Chan <mchan@broadcom.com>
Michael Chan [Sat, 5 May 2007 19:10:20 +0000 (12:10 -0700)]
[TG3]: Remove reset during MAC address changes.
The reset was added a while back so that ASF could re-init whatever
MAC address it wanted to use after the MAC address was changed.
Instead of resetting, we can just keep MAC address 1 unchanged during
MAC address changes if MAC address 1 is different from MAC address 0.
This fixes 2 problems:
1. Bonding calls set_mac_address in contexts that cannot sleep.
It no longer sleeps with the chip reset removed.
2. When ASF shares the same MAC address as the NIC, it needs to
always do that even when the MAC address is changed.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gary Zambrano [Sat, 5 May 2007 18:52:19 +0000 (11:52 -0700)]
[TG3]: WoL fixes.
Change TG3_FLAG_SERDES_WOL_CAP to TG3_FLAG_WOL_CAP to make it easier
to manage WoL. This flag is now used consistently during ethtool WoL
setup and power setting changes.
Signed-off-by: Gary Zambrano <zambrano@broadcom.com> Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gary Zambrano [Sat, 5 May 2007 18:51:45 +0000 (11:51 -0700)]
[TG3]: Clear GPIO mask before storing.
The GPIO settings may change during reset and so the stored values in
tp->grc_local_ctrl should be cleared first.
Signed-off-by: Gary Zambrano <zambrano@broadcom.com> Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Matt Carlson [Sat, 5 May 2007 18:51:05 +0000 (11:51 -0700)]
[TG3]: Improve NVRAM sizing.
This patch changes the NVRAM sizing procedure so that the driver can
take advantage of devices with 1:1 NVRAM strapping configurations. This
is useful in cases where the traditional NVRAM sizing method fails. In
the event that the flash size cannot be determined, the largest known
NVRAM size is used. The patch also removes support for 5755 NVRAM
devices that are not supported by Broadcom and adds explicit sizing for
this device.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com> Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Matt Carlson [Sat, 5 May 2007 18:50:04 +0000 (11:50 -0700)]
[TG3]: Fix TSO bugs.
1. Remove the check for skb->len greater than MTU when doing TSO.
When the destination has a smaller MSS than the source, a TSO packet
may be smaller than the MTU and we still need to process it as a TSO
packet.
2. On 5705A3 devices with TSO enabled, the DMA engine can hang due to a
hardware bug. This patch avoids the hanging condition by reducing the
DMA burst size.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com> Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Sat, 5 May 2007 18:42:03 +0000 (11:42 -0700)]
[NETLINK]: Remove references to process ID
People treating the *_pid fields in netlink as a process ID has caused
endless confusion over the years. The fact that our own netlink.h
does this only adds to the confusion.
So here is a patch to change the comments to refer to it as the port
ID which hopefully will make it clear what the purpose of the fields
really is.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Heiko Carstens [Sat, 5 May 2007 18:41:18 +0000 (11:41 -0700)]
[AF_IUCV]: Compile fix - adopt to skbuff changes.
From: Heiko Carstens <heiko.carstens@de.ibm.com>
CC [M] net/iucv/af_iucv.o
net/iucv/af_iucv.c: In function `iucv_fragment_skb':
net/iucv/af_iucv.c:984: error: structure has no member named `h'
net/iucv/af_iucv.c:985: error: structure has no member named `nh'
net/iucv/af_iucv.c:988: error: incompatible type for argument 1 of
`skb_queue_tail'
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Brownell [Mon, 2 Apr 2007 19:46:47 +0000 (12:46 -0700)]
ARM: OMAP: fix OMAP1 mpuio suspend/resume oops
Fix oops in omap16xx mpuio suspend/resume code; field wasn't initialized
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
David Brownell [Thu, 7 Dec 2006 01:14:11 +0000 (17:14 -0800)]
ARM: OMAP: MPUIO wake updates
GPIO and MPUIO wake updates:
- Hook MPUIOs into the irq wakeup framework too. This uses a platform
device to update irq enables during system sleep states, instead of
a sys_device, since the latter is no longer needed for such things.
- Also forward enable/disable irq wake requests to the relevant GPIO
controller, so the top level IRQ dispatcher can (eventually) handle
these wakeup events automatically if more than one GPIO pin needs to
be a wakeup event source.
- Minor tweak to the 24xx non-wakeup gpio stuff: no need to check such
read-only data under the spinlock.
This assumes (maybe wrongly?) that only 16xx can do GPIO wakeup; without
a 15xx I can't test such stuff.
Also this expects the top level IRQ dispatcher to properly handle requests
to enable/disable irq wake, which is currently known to be wrong: omap1
saves the flags but ignores them, omap2 doesn't even save it. (Wakeup
events are, wrongly, hardwired in the relevant mach-omapX/pm.c file ...)
So MPUIO irqs won't yet trigger system wakeup.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
David Brownell [Thu, 7 Dec 2006 01:14:10 +0000 (17:14 -0800)]
ARM: OMAP: speed up gpio irq handling
Speedup and shrink GPIO irq handling code, by using a pointer
that's available in the irq_chip structure instead of calling
the get_gpio_bank() function. On OMAP1 this saves 44 words,
most of which were in IRQ critical path methods. Hey, every
few instructions help.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This patch adds minimal OMAP2430 support to plat-omap files to
get the kernel booting on 2430SDP.
Signed-off-by: Syed Mohammed Khasim <x0khasim@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
David Brownell [Thu, 7 Dec 2006 01:13:59 +0000 (17:13 -0800)]
ARM: OMAP: gpio object shrinkage, cleanup
More GPIO/IRQ cleanup:
- compile-time removal of much useless code
* mpuio support on non-OMAP1.
* 15xx/730/24xx gpio support on 1610
* 15xx/730/16xx gpio support on 24xx
* etc
- remove all BUG() calls, which are always bad news ... replaced some
with normal fault reports for that call, others with WARN_ON(1).
- small mpuio bugfix: add missing set_type() method
Oh, and fix a minor merge issue: inode->u.generic_ip is now gone.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>