GitHub/mt8127/android_kernel_alcatel_ttab.git
16 years agohugetlb: move hugetlb_acct_memory()
Mel Gorman [Thu, 24 Jul 2008 04:27:22 +0000 (21:27 -0700)]
hugetlb: move hugetlb_acct_memory()

This is a patchset to give reliable behaviour to a process that
successfully calls mmap(MAP_PRIVATE) on a hugetlbfs file.  Currently, it
is possible for the process to be killed due to a small hugepage pool size
even if it calls mlock().

MAP_SHARED mappings on hugetlbfs reserve huge pages at mmap() time.  This
guarantees all future faults against the mapping will succeed.  This
allows local allocations at first use improving NUMA locality whilst
retaining reliability.

MAP_PRIVATE mappings do not reserve pages.  This can result in an
application being SIGKILLed later if a huge page is not available at fault
time.  This makes huge pages usage very ill-advised in some cases as the
unexpected application failure cannot be detected and handled as it is
immediately fatal.  Although an application may force instantiation of the
pages using mlock(), this may lead to poor memory placement and the
process may still be killed when performing COW.

This patchset introduces a reliability guarantee for the process which
creates a private mapping, i.e.  the process that calls mmap() on a
hugetlbfs file successfully.  The first patch of the set is purely
mechanical code move to make later diffs easier to read.  The second patch
will guarantee faults up until the process calls fork().  After patch two,
as long as the child keeps the mappings, the parent is no longer
guaranteed to be reliable.  Patch 3 guarantees that the parent will always
successfully COW by unmapping the pages from the child in the event there
are insufficient pages in the hugepage pool in allocate a new page, be it
via a static or dynamic pool.

Existing hugepage-aware applications are unlikely to be affected by this
change.  For much of hugetlbfs's history, pages were pre-faulted at mmap()
time or mmap() failed which acts in a reserve-like manner.  If the pool is
sized correctly already so that parent and child can fault reliably, the
application will not even notice the reserves.  It's only when the pool is
too small for the application to function perfectly reliably that the
reserves come into play.

Credit goes to Andy Whitcroft for cleaning up a number of mistakes during
review before the patches were released.

This patch:

A later patch in this set needs to call hugetlb_acct_memory() before it is
defined.  This patch moves the function without modification.  This makes
later diffs easier to read.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: Adam Litke <agl@us.ibm.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: William Lee Irwin III <wli@holomorphy.com>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: drop unneeded pgdat argument from free_area_init_node()
Johannes Weiner [Thu, 24 Jul 2008 04:27:20 +0000 (21:27 -0700)]
mm: drop unneeded pgdat argument from free_area_init_node()

free_area_init_node() gets passed in the node id as well as the node
descriptor.  This is redundant as the function can trivially get the node
descriptor itself by means of NODE_DATA() and the node's id.

I checked all the users and NODE_DATA() seems to be usable everywhere
from where this function is called.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomapping_set_error: add unlikely()
Andrew Morton [Thu, 24 Jul 2008 04:27:19 +0000 (21:27 -0700)]
mapping_set_error: add unlikely()

This is called on a per-page basis and in the vast majority of cases
`error' is zero.

Cc: Guillaume Chazarain <guichaz@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoslob: record page flag overlays explicitly
Andy Whitcroft [Thu, 24 Jul 2008 04:27:19 +0000 (21:27 -0700)]
slob: record page flag overlays explicitly

SLOB reuses two page bits for internal purposes, it overlays PG_active and
PG_private.  This is hidden away in slob.c.  Document these overlays
explicitly in the main page-flags enum along with all the others.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoslub: record page flag overlays explicitly
Andy Whitcroft [Thu, 24 Jul 2008 04:27:18 +0000 (21:27 -0700)]
slub: record page flag overlays explicitly

SLUB reuses two page bits for internal purposes, it overlays PG_active and
PG_error.  This is hidden away in slub.c.  Document these overlays
explicitly in the main page-flags enum along with all the others.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopage-flags: record page flag overlays explicitly
Andy Whitcroft [Thu, 24 Jul 2008 04:27:16 +0000 (21:27 -0700)]
page-flags: record page flag overlays explicitly

With the recent page flag reorganisation we have a single enum which
defines the valid page flags and their values, nice and clear.  However
there are a number of bits which are overloaded by different subsystems.
Firstly there is PG_owner_priv_1 which is used by filesystems and by XEN.
Secondly both SLOB and SLUB use a couple of extra page bits to manage
internal state for pages they own; both overlay other bits.  All of these
"aliases" are scattered about the source making it very hard for a reader
to know if the bits are safe to rely on in all contexts; confusion here is
bad.

As we now have a single place where the bits are clearly assigned it makes
sense to clarify the reuse of bits by making the aliases explicit and
visible with the original bit assignments.  This patch creates explicit
aliases within the enum itself for the overloaded bits, creates standard
bit accessors PageFoo etc.  and uses those throughout.

This version pulls the bit manipulation out to standard named page bit
accessors as suggested by Christoph, it retains the explicit mapping to
the overlayed bits.  A fusion of both ideas.  This has been SLUB and SLOB
have been compile tested on x86_64 only, and SLUB boot tested.  If people
feel this is worth doing then I can run a fuller set of testing.

This patch:

Some page flags are used for more than one purpose, for example
PG_owner_priv_1.  Currently there are individual accessors for each user,
each built using the common flag name far away from the bit definitions.
This makes it hard to see all possible uses of these bits.

Now that we have a single enum to generate the bit orders it makes sense
to express overlays in the same place.  So create per use aliases for this
bit in the main page-flags enum and use those in the accessors.

[akpm@linux-foundation.org: fix xen]
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofix soft lock up at NFS mount via per-SB LRU-list of unused dentries
Kentaro Makita [Thu, 24 Jul 2008 04:27:13 +0000 (21:27 -0700)]
fix soft lock up at NFS mount via per-SB LRU-list of unused dentries

[Summary]

 Split LRU-list of unused dentries to one per superblock to avoid soft
 lock up during NFS mounts and remounting of any filesystem.

 Previously I posted here:
 http://lkml.org/lkml/2008/3/5/590

[Descriptions]

- background

  dentry_unused is a list of dentries which are not referenced.
  dentry_unused grows up when references on directories or files are
  released.  This list can be very long if there is huge free memory.

- the problem

  When shrink_dcache_sb() is called, it scans all dentry_unused linearly
  under spin_lock(), and if dentry->d_sb is differnt from given
  superblock, scan next dentry.  This scan costs very much if there are
  many entries, and very ineffective if there are many superblocks.

  IOW, When we need to shrink unused dentries on one dentry, but scans
  unused dentries on all superblocks in the system.  For example, we scan
  500 dentries to unmount a filesystem, but scans 1,000,000 or more unused
  dentries on other superblocks.

  In our case , At mounting NFS*, shrink_dcache_sb() is called to shrink
  unused dentries on NFS, but scans 100,000,000 unused dentries on
  superblocks in the system such as local ext3 filesystems.  I hear NFS
  mounting took 1 min on some system in use.

* : NFS uses virtual filesystem in rpc layer, so NFS is affected by
  this problem.

  100,000,000 is possible number on large systems.

  Per-superblock LRU of unused dentried can reduce the cost in
  reasonable manner.

- How to fix

  I found this problem is solved by David Chinner's "Per-superblock
  unused dentry LRU lists V3"(1), so I rebase it and add some fix to
  reclaim with fairness, which is in Andrew Morton's comments(2).

  1) http://lkml.org/lkml/2006/5/25/318
  2) http://lkml.org/lkml/2006/5/25/320

  Split LRU-list of unused dentries to each superblocks.  Then, NFS
  mounting will check dentries under a superblock instead of all.  But
  this spliting will break LRU of dentry-unused.  So, I've attempted to
  make reclaim unused dentrins with fairness by calculate number of
  dentries to scan on this sb based on following way

  number of dentries to scan on this sb =
  count * (number of dentries on this sb / number of dentries in the machine)

- ToDo
 - I have to measuring performance number and do stress tests.

 - When unmount occurs during prune_dcache(), scanning on same
  superblock, It is unable to reach next superblock because it is gone
  away.  We restart scannig superblock from first one, it causes
  unfairness of reclaim unused dentries on first superblock.  But I think
  this happens very rarely.

- Test Results

  Result on 6GB boxes with excessive unused dentries.

Without patch:

$ cat /proc/sys/fs/dentry-state
10181835        10180203        45      0       0       0
# mount -t nfs 10.124.60.70:/work/kernel-src nfs
real    0m1.830s
user    0m0.001s
sys     0m1.653s

 With this patch:
$ cat /proc/sys/fs/dentry-state
10236610        10234751        45      0       0       0
# mount -t nfs 10.124.60.70:/work/kernel-src nfs
real    0m0.106s
user    0m0.002s
sys     0m0.032s

[akpm@linux-foundation.org: fix comments]
Signed-off-by: Kentaro Makita <k-makita@np.css.fujitsu.com>
Cc: Neil Brown <neilb@suse.de>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: David Chinner <dgc@sgi.com>
Cc: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobuddy: clarify comments describing buddy merge
Andy Whitcroft [Thu, 24 Jul 2008 04:27:11 +0000 (21:27 -0700)]
buddy: clarify comments describing buddy merge

In __free_one_page(), the comment "Move the buddy up one level" appears
attached to the break and by implication when the break is taken we are
moving it up one level:

if (!page_is_buddy(page, buddy, order))
break;          /* Move the buddy up one level. */

In reality the inverse is true, we break out when we can no longer merge
this page with its buddy.  Looking back into pre-history (into the full
git history) it appears that these two lines accidentally got joined as
part of another change.

Move the comment down where it belongs below the if and clarify its
language.

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>
16 years agomm: remove double indirection on tlb parameter to free_pgd_range() & Co
Jan Beulich [Thu, 24 Jul 2008 04:27:10 +0000 (21:27 -0700)]
mm: remove double indirection on tlb parameter to free_pgd_range() & Co

The double indirection here is not needed anywhere and hence (at least)
confusing.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: "David S. Miller" <davem@davemloft.net>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agospufs: use new vm_ops->access to allow local state access from gdb
Benjamin Herrenschmidt [Thu, 24 Jul 2008 04:27:09 +0000 (21:27 -0700)]
spufs: use new vm_ops->access to allow local state access from gdb

This uses the new vm_ops->access to allow gdb to access the SPU local
store.  We currently prevent access to problem state registers, this can
be done later if really needed but it's safer not to.

[akpm@linux-foundation.org: fix typo]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopowerpc ioremap_prot
Benjamin Herrenschmidt [Thu, 24 Jul 2008 04:27:08 +0000 (21:27 -0700)]
powerpc ioremap_prot

This adds ioremap_prot and pte_pgprot() so that one can extract protection
bits from a PTE and use them to ioremap_prot() (in order to support ptrace
of VM_IO | VM_PFNMAP as per Rik's patch).

This moves a couple of flag checks around in the ioremap implementations
of arch/powerpc.  There's a side effect of allowing non-cacheable and
non-guarded mappings on ppc32 which before would always have _PAGE_GUARDED
set whenever _PAGE_NO_CACHE is.

(standard ioremap will still set _PAGE_GUARDED, but ioremap_prot will be
capable of setting such a non guarded mapping).

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agouse generic_access_phys for /dev/mem mappings
Rik van Riel [Thu, 24 Jul 2008 04:27:07 +0000 (21:27 -0700)]
use generic_access_phys for /dev/mem mappings

Use generic_access_phys as the access_process_vm access function for
/dev/mem mappings.  This makes it possible to debug the X server.

[akpm@linux-foundation.org: repair all the architectures which broke]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: Benjamin Herrensmidt <benh@kernel.crashing.org>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoaccess_process_vm device memory infrastructure
Rik van Riel [Thu, 24 Jul 2008 04:27:05 +0000 (21:27 -0700)]
access_process_vm device memory infrastructure

In order to be able to debug things like the X server and programs using
the PPC Cell SPUs, the debugger needs to be able to access device memory
through ptrace and /proc/pid/mem.

This patch:

Add the generic_access_phys access function and put the hooks in place
to allow access_process_vm to access device or PPC Cell SPU memory.

[riel@redhat.com: Add documentation for the vm_ops->access function]
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Benjamin Herrensmidt <benh@kernel.crashing.org>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: remove nopfn
Nick Piggin [Thu, 24 Jul 2008 04:27:05 +0000 (21:27 -0700)]
mm: remove nopfn

There are no users of nopfn in the tree. Remove it.

[hugh@veritas.com: fix build error]
Signed-off-by: Nick Piggin <npiggin@suse.de>
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>
16 years agokill generic_file_direct_IO()
Christoph Hellwig [Thu, 24 Jul 2008 04:27:04 +0000 (21:27 -0700)]
kill generic_file_direct_IO()

generic_file_direct_IO is a common helper around the invocation of
->direct_IO.  But there's almost nothing shared between the read and write
side, so we're better off without this helper.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm/hugetlb.c: fix duplicate variable
Adrian Bunk [Thu, 24 Jul 2008 04:27:03 +0000 (21:27 -0700)]
mm/hugetlb.c: fix duplicate variable

It's confusing that set_max_huge_pages() contained two different
variables named "ret", and although the code works correctly this should
be fixed.

The inner of the two variables can simply be removed.

Spotted by sparse.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: "KOSAKI Motohiro" <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm/vmstat.c: proper externs
Adrian Bunk [Thu, 24 Jul 2008 04:27:03 +0000 (21:27 -0700)]
mm/vmstat.c: proper externs

This patch adds proper extern declarations for five variables in
include/linux/vmstat.h

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm/migrate.c should #include <linux/syscalls.h>
Adrian Bunk [Thu, 24 Jul 2008 04:27:02 +0000 (21:27 -0700)]
mm/migrate.c should #include <linux/syscalls.h>

Every file should include the headers containing the externs for its
global functions (in this case for sys_move_pages()).

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopage allocator: inline some __alloc_pages() wrappers
KOSAKI Motohiro [Thu, 24 Jul 2008 04:27:01 +0000 (21:27 -0700)]
page allocator: inline some __alloc_pages() wrappers

Two zonelist patch series rewrote __page_alloc() largely.  Now, it is just
a wrapper function.  Inlining them will save a function call.

[akpm@linux-foundation.org: export __alloc_pages_internal]
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomspec: convert nopfn to fault
Nick Piggin [Thu, 24 Jul 2008 04:27:00 +0000 (21:27 -0700)]
mspec: convert nopfn to fault

[akpm@linux-foundation.org: remove unused variable]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: unexport __alloc_bootmem_core()
Johannes Weiner [Thu, 24 Jul 2008 04:26:59 +0000 (21:26 -0700)]
mm: unexport __alloc_bootmem_core()

This function has no external callers, so unexport it.  Also fix its naming
inconsistency.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: normalize internal argument passing of bootmem data
Johannes Weiner [Thu, 24 Jul 2008 04:26:59 +0000 (21:26 -0700)]
mm: normalize internal argument passing of bootmem data

All _core functions only need the bootmem data, not the whole node descriptor.
Adjust the two functions that take the node descriptor unneededly.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: fix free_all_bootmem_core alignment check
Johannes Weiner [Thu, 24 Jul 2008 04:26:58 +0000 (21:26 -0700)]
mm: fix free_all_bootmem_core alignment check

The check for node_boot_start is bogus because we start freeing at the
corresponding pfn.  So check if the pfn is properly aligned instead in a more
readable way and adjust the documentation.

Also remove an unneeded accounting variable.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: move bootmem descriptors definition to a single place
Johannes Weiner [Thu, 24 Jul 2008 04:26:55 +0000 (21:26 -0700)]
mm: move bootmem descriptors definition to a single place

There are a lot of places that define either a single bootmem descriptor or an
array of them.  Use only one central array with MAX_NUMNODES items instead.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Kyle McMartin <kyle@parisc-linux.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoadd a helper function to test if an object is on the stack
FUJITA Tomonori [Thu, 24 Jul 2008 04:26:53 +0000 (21:26 -0700)]
add a helper function to test if an object is on the stack

lib/debugobjects.c has a function to test if an object is on the stack.
The block layer and ide needs it (they need to avoid DMA from/to stack
buffers).  This patch moves the function to include/linux/sched.h so that
everyone can use it.

lib/debugobjects.c uses current->stack but this patch uses a
task_stack_page() accessor, which is a preferable way to access the stack.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: print out the zonelists on request for manual verification
Mel Gorman [Thu, 24 Jul 2008 04:26:52 +0000 (21:26 -0700)]
mm: print out the zonelists on request for manual verification

This patch prints out the zonelists during boot for manual verification by the
user if the mminit_loglevel is MMINIT_VERIFY or higher.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: make defensive checks around PFN values registered for memory usage
Mel Gorman [Thu, 24 Jul 2008 04:26:52 +0000 (21:26 -0700)]
mm: make defensive checks around PFN values registered for memory usage

There are a number of different views to how much memory is currently active.
There is the arch-independent zone-sizing view, the bootmem allocator and
memory models view.

Architectures register this information at different times and is not
necessarily in sync particularly with respect to some SPARSEMEM limitations.

This patch introduces mminit_validate_memmodel_limits() which is able to
validate and correct PFN ranges with respect to the memory model.  It is only
SPARSEMEM that currently validates itself.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: verify the page links and memory model
Mel Gorman [Thu, 24 Jul 2008 04:26:51 +0000 (21:26 -0700)]
mm: verify the page links and memory model

Print out information on how the page flags are being used if mminit_loglevel
is MMINIT_VERIFY or higher and unconditionally performs sanity checks on the
flags regardless of loglevel.

When the page flags are updated with section, node and zone information, a
check are made to ensure the values can be retrieved correctly.  Finally we
confirm that pfn_to_page and page_to_pfn are the correct inverse functions.

[akpm@linux-foundation.org: fix printk warnings]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: add a basic debugging framework for memory initialisation
Mel Gorman [Thu, 24 Jul 2008 04:26:49 +0000 (21:26 -0700)]
mm: add a basic debugging framework for memory initialisation

Boot initialisation is very complex, with significant numbers of
architecture-specific routines, hooks and code ordering.  While significant
amounts of the initialisation is architecture-independent, it trusts the data
received from the architecture layer.  This is a mistake, and has resulted in
a number of difficult-to-diagnose bugs.

This patchset adds some validation and tracing to memory initialisation.  It
also introduces a few basic defensive measures.  The validation code can be
explicitly disabled for embedded systems.

This patch:

Add additional debugging and verification code for memory initialisation.

Once enabled, the verification checks are always run and when required
additional debugging information may be outputted via a mminit_loglevel=
command-line parameter.

The verification code is placed in a new file mm/mm_init.c.  Ideally other mm
initialisation code will be moved here over time.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoadd HAVE_CLK to Kconfig, for driver dependencies
David Brownell [Thu, 24 Jul 2008 04:26:48 +0000 (21:26 -0700)]
add HAVE_CLK to Kconfig, for driver dependencies

Flag platforms as HAVE_CLK (or not) in Kconfig, based on whether they
support <linux/clk.h> calls, so that otherwise portable drivers which need
those calls can list that dependency.

Something like this is a prerequisite for merging the musb_hdrc driver,
currently used on platforms including Davinci, OMAP2430, OMAP3xx ...  and
the discrete TUSB6010 chip, which doesn't have a natural platform
dependency.  (Used with OMAP 2420 in current Nokia N8x0 tablets.)

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Acked-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoremove is_tty()
Adrian Bunk [Thu, 24 Jul 2008 04:26:47 +0000 (21:26 -0700)]
remove is_tty()

This patch removes the no longer used is_tty().

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohpet: clarify maintainer entry
Clemens Ladisch [Thu, 24 Jul 2008 04:26:46 +0000 (21:26 -0700)]
hpet: clarify maintainer entry

The existing HPET maintainer entries are somewhat unclear about which one
applies to what part of the kernel.

Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomove memory_read_from_buffer() from fs.h to string.h
Akinobu Mita [Thu, 24 Jul 2008 04:26:44 +0000 (21:26 -0700)]
move memory_read_from_buffer() from fs.h to string.h

James Bottomley warns that inclusion of linux/fs.h in a low level
driver was always a danger signal.  This patch moves
memory_read_from_buffer() from fs.h to string.h and fixes includes in
existing memory_read_from_buffer() users.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: James Bottomley <James.Bottomley@hansenpartnership.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Bob Moore <robert.moore@intel.com>
Cc: Thomas Renninger <trenn@suse.de>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'x86/auditsc' of git://git.kernel.org/pub/scm/linux/kernel/git/frob...
Linus Torvalds [Thu, 24 Jul 2008 03:39:21 +0000 (20:39 -0700)]
Merge branch 'x86/auditsc' of git://git./linux/kernel/git/frob/linux-2.6-roland

* 'x86/auditsc' of git://git.kernel.org/pub/scm/linux/kernel/git/frob/linux-2.6-roland:
  i386 syscall audit fast-path
  x86_64 ia32 syscall audit fast-path
  x86_64 syscall audit fast-path
  x86_64: remove bogus optimization in sysret_signal

16 years agoMerge branch 'sched/for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip...
Linus Torvalds [Thu, 24 Jul 2008 02:36:53 +0000 (19:36 -0700)]
Merge branch 'sched/for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched/for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: hrtick_enabled() should use cpu_active()
  sched, x86: clean up hrtick implementation
  sched: fix build error, provide partition_sched_domains() unconditionally
  sched: fix warning in inc_rt_tasks() to not declare variable 'rq' if it's not needed
  cpu hotplug: Make cpu_active_map synchronization dependency clear
  cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2)
  sched: rework of "prioritize non-migratable tasks over migratable ones"
  sched: reduce stack size in isolated_cpu_setup()
  Revert parts of "ftrace: do not trace scheduler functions"

Fixed up conflicts in include/asm-x86/thread_info.h (due to the
TIF_SINGLESTEP unification vs TIF_HRTICK_RESCHED removal) and
kernel/sched_fair.c (due to cpu_active_map vs for_each_cpu_mask_nr()
introduction).

16 years agoMerge branch 'cpus4096-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 24 Jul 2008 01:37:44 +0000 (18:37 -0700)]
Merge branch 'cpus4096-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'cpus4096-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (31 commits)
  NR_CPUS: Replace NR_CPUS in speedstep-centrino.c
  cpumask: Provide a generic set of CPUMASK_ALLOC macros, FIXUP
  NR_CPUS: Replace NR_CPUS in cpufreq userspace routines
  NR_CPUS: Replace per_cpu(..., smp_processor_id()) with __get_cpu_var
  NR_CPUS: Replace NR_CPUS in arch/x86/kernel/genapic_flat_64.c
  NR_CPUS: Replace NR_CPUS in arch/x86/kernel/genx2apic_uv_x.c
  NR_CPUS: Replace NR_CPUS in arch/x86/kernel/cpu/proc.c
  NR_CPUS: Replace NR_CPUS in arch/x86/kernel/cpu/mcheck/mce_64.c
  cpumask: Optimize cpumask_of_cpu in lib/smp_processor_id.c, fix
  cpumask: Use optimized CPUMASK_ALLOC macros in the centrino_target
  cpumask: Provide a generic set of CPUMASK_ALLOC macros
  cpumask: Optimize cpumask_of_cpu in lib/smp_processor_id.c
  cpumask: Optimize cpumask_of_cpu in kernel/time/tick-common.c
  cpumask: Optimize cpumask_of_cpu in drivers/misc/sgi-xp/xpc_main.c
  cpumask: Optimize cpumask_of_cpu in arch/x86/kernel/ldt.c
  cpumask: Optimize cpumask_of_cpu in arch/x86/kernel/io_apic_64.c
  cpumask: Replace cpumask_of_cpu with cpumask_of_cpu_ptr
  Revert "cpumask: introduce new APIs"
  cpumask: make for_each_cpu_mask a bit smaller
  net: Pass reference to cpumask variable in net/sunrpc/svc.c
  ...

Fix up trivial conflicts in drivers/cpufreq/cpufreq.c manually

16 years agoMerge branch 'core/softlockup-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 24 Jul 2008 01:34:13 +0000 (18:34 -0700)]
Merge branch 'core/softlockup-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core/softlockup-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  softlockup: fix invalid proc_handler for softlockup_panic
  softlockup: fix watchdog task wakeup frequency
  softlockup: fix watchdog task wakeup frequency
  softlockup: show irqtrace
  softlockup: print a module list on being stuck
  softlockup: fix NMI hangs due to lock race - 2.6.26-rc regression
  softlockup: fix false positives on nohz if CPU is 100% idle for more than 60 seconds
  softlockup: fix softlockup_thresh fix
  softlockup: fix softlockup_thresh unaligned access and disable detection at runtime
  softlockup: allow panic on lockup

16 years agoMerge branch 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Thu, 24 Jul 2008 01:24:08 +0000 (18:24 -0700)]
Merge branch 'devel' of /home/rmk/linux-2.6-arm

* 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm: (85 commits)
  [ARM] pxa: add base support for PXA930 Handheld Platform (aka SAAR)
  [ARM] pxa: add base support for PXA930 Evaluation Board (aka TavorEVB)
  [ARM] pxa: add base support for PXA930 (aka Tavor-P)
  [ARM] Update mach-types
  [ARM] pxa: make littleton to use the new smc91x platform data
  [ARM] pxa: make zylonite to use the new smc91x platform data
  [ARM] pxa: make mainstone to use the new smc91x platform data
  [ARM] pxa: make lubbock to use new smc91x platform data
  [NET] smc91x: prepare SMC_USE_PXA_DMA to be specified in platform data
  [NET] smc91x: prepare for SMC_IO_SHIFT to be a platform configurable variable
  [NET] smc91x: add SMC91X_NOWAIT flag to platform data
  [NET] smc91x: favor the use of SMC91X_USE_* instead of SMC_CAN_USE_*
  [NET] smc91x: remove "irq_flags" from "struct smc91x_platdata"
  [ARM] 5146/1: pxa2xx: convert all boards to call pxa2xx_transceiver_mode helper
  Support for LCD on e740 e750 e400 and e800 e-series PDAs
  E-series UDC support
  PXA UDC - allow use of inverted GPIO for pullup
  Add e350 support
  Fix broken e-series build
  E-series GPIO / IRQ definitions.
  ...

16 years agoi386 syscall audit fast-path
Roland McGrath [Tue, 24 Jun 2008 11:16:52 +0000 (04:16 -0700)]
i386 syscall audit fast-path

This adds fast paths for 32-bit syscall entry and exit when
TIF_SYSCALL_AUDIT is set, but no other kind of syscall tracing.
These paths does not need to save and restore all registers as
the general case of tracing does.  Avoiding the iret return path
when syscall audit is enabled helps performance a lot.

Signed-off-by: Roland McGrath <roland@redhat.com>
16 years agox86_64 ia32 syscall audit fast-path
Roland McGrath [Tue, 24 Jun 2008 08:13:31 +0000 (01:13 -0700)]
x86_64 ia32 syscall audit fast-path

This adds fast paths for 32-bit syscall entry and exit when
TIF_SYSCALL_AUDIT is set, but no other kind of syscall tracing.
These paths does not need to save and restore all registers as
the general case of tracing does.  Avoiding the iret return path
when syscall audit is enabled helps performance a lot.

Signed-off-by: Roland McGrath <roland@redhat.com>
16 years agox86_64 syscall audit fast-path
Roland McGrath [Mon, 23 Jun 2008 22:37:04 +0000 (15:37 -0700)]
x86_64 syscall audit fast-path

This adds a fast path for 64-bit syscall entry and exit when
TIF_SYSCALL_AUDIT is set, but no other kind of syscall tracing.
This path does not need to save and restore all registers as
the general case of tracing does.  Avoiding the iret return path
when syscall audit is enabled helps performance a lot.

Signed-off-by: Roland McGrath <roland@redhat.com>
16 years agox86_64: remove bogus optimization in sysret_signal
Roland McGrath [Tue, 24 Jun 2008 03:41:12 +0000 (20:41 -0700)]
x86_64: remove bogus optimization in sysret_signal

This short-circuit path in sysret_signal looks wrong to me.
AFAICT, in practice the branch is never taken--and if it were,
it would go wrong.  To wit, try loading a module whose init
function does set_thread_flag(TIF_IRET), and see insmod crash
(presumably with a wrong user stack pointer).

This is because the FIXUP_TOP_OF_STACK work hasn't been done yet
when we jump around the call to ptregscall_common and get to
int_with_check--where it expects the user RSP,SS,CS and EFLAGS to
have been stored by FIXUP_TOP_OF_STACK.

I don't think it's normally possible to get to sysret_signal with no
_TIF_DO_NOTIFY_MASK bits set anyway, so these two instructions are
already superfluous.  If it ever did happen, it is harmless to call
do_notify_resume with nothing for it to do.

Signed-off-by: Roland McGrath <roland@redhat.com>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc
Linus Torvalds [Wed, 23 Jul 2008 19:04:34 +0000 (12:04 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/drzeus/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc:
  sdhci: highmem capable PIO routines
  sg: reimplement sg mapping iterator
  mmc_test: print message when attaching to card
  mmc: Remove Russell as primecell mci maintainer
  mmc_block: bounce buffer highmem support
  sdhci: fix bad warning from commit c8b3e02
  sdhci: add warnings for bad buffers in ADMA path
  mmc_test: test oversized sg lists
  mmc_test: highmem tests
  s3cmci: ensure host stopped on machine shutdown
  au1xmmc: suspend/resume implementation
  s3cmci: fixes for section mismatch warnings
  pxamci: trivial fix of DMA alignment register bit clearing

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx
Linus Torvalds [Wed, 23 Jul 2008 19:03:18 +0000 (12:03 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/djbw/async_tx

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx: (24 commits)
  I/OAT: I/OAT version 3.0 support
  I/OAT: tcp_dma_copybreak default value dependent on I/OAT version
  I/OAT: Add watchdog/reset functionality to ioatdma
  iop_adma: cleanup iop_chan_xor_slot_count
  iop_adma: document how to calculate the minimum descriptor pool size
  iop_adma: directly reclaim descriptors on allocation failure
  async_tx: make async_tx_test_ack a boolean routine
  async_tx: remove depend_tx from async_tx_sync_epilog
  async_tx: export async_tx_quiesce
  async_tx: fix handling of the "out of descriptor" condition in async_xor
  async_tx: ensure the xor destination buffer remains dma-mapped
  async_tx: list_for_each_entry_rcu() cleanup
  dmaengine: Driver for the Synopsys DesignWare DMA controller
  dmaengine: Add slave DMA interface
  dmaengine: add DMA_COMPL_SKIP_{SRC,DEST}_UNMAP flags to control dma unmap
  dmaengine: Add dma_client parameter to device_alloc_chan_resources
  dmatest: Simple DMA memcpy test client
  dmaengine: DMA engine driver for Marvell XOR engine
  iop-adma: fix platform driver hotplug/coldplug
  dmaengine: track the number of clients using a channel
  ...

Fixed up conflict in drivers/dca/dca-sysfs.c manually

16 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel...
Linus Torvalds [Wed, 23 Jul 2008 18:59:37 +0000 (11:59 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jwessel/linux-2.6-kgdb

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/linux-2.6-kgdb:
  kgdb: kgdboc console poll hooks for mpsc uart
  kgdb: kgdboc console poll hooks for cpm uart
  kgdb, powerpc: arch specific powerpc kgdb support
  kgdb: support for ARCH=arm
  kgdb: remove unused HAVE_ARCH_KGDB_SHADOW_INFO config variable

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Wed, 23 Jul 2008 18:59:09 +0000 (11:59 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6: (60 commits)
  ide: small whitespace fixes
  ide: ide-cd_ioctl.c fix sparse integer as NULL pointer warnings
  ide: ide-cd.c fix sparse endianness warnings
  ide-cd: convert to using the new atapi_flags
  ide: remove unused PC_FLAG_DRQ_INTERRUPT
  ide-scsi: convert to using the new atapi_flags
  ide-tape: convert to using the new atapi_flags
  ide-floppy: convert to using the new atapi_flags (take 2)
  ide: add per-device flags
  ide: use rq->cmd instead of pc->c in atapi common code
  ide-scsi: pass packet command in rq->cmd
  ide-tape: pass packet command in rq->cmd
  ide-tape: make room for packet command ids in rq->cmd
  ide-floppy: pass packet command in rq->cmd
  ide: remove pc->callback member from ide_atapi_pc
  ide-scsi: use drive->pc_callback instead of pc->callback
  ide-tape: use drive->pc_callback instead of pc->callback
  ide-floppy: use drive->pc_callback instead of pc->callback
  ide: push pc callback pointer into the ide_drive_t structure
  drivers/ide/ide-tape.c: remove double kfree
  ...

16 years agoide: small whitespace fixes
Pavel Machek [Wed, 23 Jul 2008 17:56:02 +0000 (19:56 +0200)]
ide: small whitespace fixes

Small whitespace fixes for drivers/ide.

Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: ide-cd_ioctl.c fix sparse integer as NULL pointer warnings
Harvey Harrison [Wed, 23 Jul 2008 17:56:02 +0000 (19:56 +0200)]
ide: ide-cd_ioctl.c fix sparse integer as NULL pointer warnings

The function ide_cd_queue_pc should be checked as the bufflen arg is
dereferenced and lots of callers are passing in NULL.

drivers/ide/ide-cd_ioctl.c:124:46: warning: Using plain integer as NULL pointer
drivers/ide/ide-cd_ioctl.c:149:47: warning: Using plain integer as NULL pointer
drivers/ide/ide-cd_ioctl.c:231:46: warning: Using plain integer as NULL pointer
drivers/ide/ide-cd_ioctl.c:374:46: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: ide-cd.c fix sparse endianness warnings
Harvey Harrison [Wed, 23 Jul 2008 17:56:02 +0000 (19:56 +0200)]
ide: ide-cd.c fix sparse endianness warnings

drivers/ide/ide-cd.c:1276:46: warning: Using plain integer as NULL pointer
drivers/ide/ide-cd.c:1298:19: warning: cast to restricted __be32
drivers/ide/ide-cd.c:1300:4: warning: cast to restricted __be32
drivers/ide/ide-cd.c:1522:14: warning: cast to restricted __le16
drivers/ide/ide-cd.c:1523:14: warning: cast to restricted __le16
drivers/ide/ide-cd.c:1525:14: warning: cast to restricted __be16
drivers/ide/ide-cd.c:1526:14: warning: cast to restricted __be16
drivers/ide/ide-cd.c:1667:37: warning: cast to restricted __be16

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Acked-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-cd: convert to using the new atapi_flags
Borislav Petkov [Wed, 23 Jul 2008 17:56:02 +0000 (19:56 +0200)]
ide-cd: convert to using the new atapi_flags

There should be no functionality change resulting from this patch.

[bart: IDE_FLAG_* -> IDE_AFLAG_*, dev_flags -> atapi_flags]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove unused PC_FLAG_DRQ_INTERRUPT
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide: remove unused PC_FLAG_DRQ_INTERRUPT

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-scsi: convert to using the new atapi_flags
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide-scsi: convert to using the new atapi_flags

Remove unused IDESCSI_PC_RQ while at it.

There should be no functionality change resulting from this patch.

[bart: IDE_FLAG_* -> IDE_AFLAG_*, dev_flags -> atapi_flags]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-tape: convert to using the new atapi_flags
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide-tape: convert to using the new atapi_flags

There should be no functionality change resulting from this patch.

[bart: IDE_FLAG_* -> IDE_AFLAG_*, dev_flags -> atapi_flags]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-floppy: convert to using the new atapi_flags (take 2)
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide-floppy: convert to using the new atapi_flags (take 2)

while at it, remove PC_FLAG_ZIP_DRIVE from the packed command flags altogether
and query the drive type through drive->atapi_flags.

v2:
ide-floppy fix.

There should be no functionality change resulting from this patch.

[bart: IDE_FLAG_* -> IDE_AFLAG_*, dev_flags -> atapi_flags]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add per-device flags
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide: add per-device flags

Push device flags up into ide_drive_t.

There should be no functionality change resulting from this patch.

[bart: IDE_FLAG_* -> IDE_AFLAG_*, dev_flags -> atapi_flags]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use rq->cmd instead of pc->c in atapi common code
Borislav Petkov [Wed, 23 Jul 2008 17:56:01 +0000 (19:56 +0200)]
ide: use rq->cmd instead of pc->c in atapi common code

There should be no functionality change resulting from this patch.

[bart: s/HWGROUP(drive)/hwif->hwgroup/]

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-scsi: pass packet command in rq->cmd
Borislav Petkov [Wed, 23 Jul 2008 17:56:00 +0000 (19:56 +0200)]
ide-scsi: pass packet command in rq->cmd

Make a redundant copy of the packet command bits into rq->cmd.
Later, after all drivers have been converted, it'll be
switched to use that in the common code instead of pc->c.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-tape: pass packet command in rq->cmd
Borislav Petkov [Wed, 23 Jul 2008 17:56:00 +0000 (19:56 +0200)]
ide-tape: pass packet command in rq->cmd

Make a redundant copy of the packet command bits into rq->cmd. Later, after
all drivers have been converted, it'll be switched to use that in the
common code instead of pc->c. While at it, simplify ide_tape_create_rw_cmd.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-tape: make room for packet command ids in rq->cmd
Borislav Petkov [Wed, 23 Jul 2008 17:56:00 +0000 (19:56 +0200)]
ide-tape: make room for packet command ids in rq->cmd

ide-tape uses rq->cmd for special commands from the chrdev interface so move
those to byte 13 (BLK_MAX_CDB = 16) since a packet cmd is max 12 bytes.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-floppy: pass packet command in rq->cmd
Borislav Petkov [Wed, 23 Jul 2008 17:56:00 +0000 (19:56 +0200)]
ide-floppy: pass packet command in rq->cmd

Make a redundant copy of the packet command bits into rq->cmd.
Later, after all drivers have been converted, it'll be
switched to use that in the common code instead of pc->c.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove pc->callback member from ide_atapi_pc
Borislav Petkov [Wed, 23 Jul 2008 17:56:00 +0000 (19:56 +0200)]
ide: remove pc->callback member from ide_atapi_pc

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-scsi: use drive->pc_callback instead of pc->callback
Borislav Petkov [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
ide-scsi: use drive->pc_callback instead of pc->callback

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-tape: use drive->pc_callback instead of pc->callback
Borislav Petkov [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
ide-tape: use drive->pc_callback instead of pc->callback

The if-else block in the IRQ handler is only temporary so that bisect searches
don't break and it'll be removed after converting the remainder of the drivers.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-floppy: use drive->pc_callback instead of pc->callback
Borislav Petkov [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
ide-floppy: use drive->pc_callback instead of pc->callback

It is important that drive->pc_callback is set prior to enabling IRQs on the
device since this is called from the IRQ handler. Otherwise it hurts as I learnt
the hard way from the several "Kernel panic - not synching: Fatal exception in
interrupt" during the weekend :).

The if-else block in the IRQ handler is only temporary so that bisect searches
don't break and it'll be removed after converting the remainder of the drivers.

There should be no functionality change resulting from this patch.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: push pc callback pointer into the ide_drive_t structure
Borislav Petkov [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
ide: push pc callback pointer into the ide_drive_t structure

Refrain from carrying the callback ptr with every packet command since the
callback function is only one anyways. ide_drive_t is probably not the most
suitable place for it right now but is the more sane solution. Besides, these
structs are going to be reorganized anyways during the generic ide rewrite.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agodrivers/ide/ide-tape.c: remove double kfree
Darren Jenkins [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
drivers/ide/ide-tape.c: remove double kfree

Coverity CID: 2336 USE_AFTER_FREE

drivers/ide/ide-tape.c ide_tape_kfree_buffer() double free's a pointer
(hint freed by an alias in first outer loop).

This patch simply removes the superfluous kfree().

Signed-off-by: Darren Jenkins <darrenrjenkins@gmailcom>
Acked-by: Borislav Petkov <petkovbb@gmail.com>
Cc: gadio@netvision.net.il
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_host_free() helper (take 2)
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:59 +0000 (19:55 +0200)]
ide: add ide_host_free() helper (take 2)

* Add ide_host_free() helper and convert ide_host_remove() to use it.

* Fix handling of ide_host_register() failure in ide_host_add(),
  icside.c, ide-generic.c, falconide.c and sgiioc4.c.

While at it:

* Fix handling of ide_host_alloc_all() failure in ide-generic.c.

* Fix handling of ide_host_alloc() failure in falconide.c
  (also return the correct error value if no device is found).

v2:
* falconide build fix. (From Stephen Rothwell)

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: allocate ide_hwif_t instances dynamically
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:58 +0000 (19:55 +0200)]
ide: allocate ide_hwif_t instances dynamically

* Allocate ide_hwif_t instances dynamically and remove ide_hwifs[].

  This cuts almost ~14kB from ide-probe.o (x86-32, MAX_HWIFS == 10):

   text    data     bss     dec     hex filename
   9140      40   14084   23264    5ae0 drivers/ide/ide-probe.o.before
   9169      40      44    9253    2425 drivers/ide/ide-probe.o.after

* Remove no longer needed ide_init_port_data() call from ide_unregister().

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_ports[]
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:58 +0000 (19:55 +0200)]
ide: add ide_ports[]

* Add ide_ports[] table keeping pointers to all initialized ports.

* Set ide_ports[] entry for a given port in init_irq() and clear it
  ide_remove_port_from_hwgroup().

* Convert init_irq() to use ide_ports[] instead of ide_hwifs[] for
  grouping ports into hwgroup.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: move ide_remove_port_from_hwgroup() to ide-probe.c
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:58 +0000 (19:55 +0200)]
ide: move ide_remove_port_from_hwgroup() to ide-probe.c

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: fix IDE port slots reservation and freeing (take 2)
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:57 +0000 (19:55 +0200)]
ide: fix IDE port slots reservation and freeing (take 2)

* Make ide_find_port_slot() return port slot index.

* Add ide_free_port_slot() helper.

* Move ide_init_port_data() call and error printk() from
  ide_find_port_slot() to ide_host_alloc_all().

* Make ide_{find,free}_port_slot() take ide_cfg_mtx mutex
  and convert them use ide_indexes bitmap to keep track of
  reserved port slots.

* Don't set hwif->chipset in ide_host_alloc_all() as it is
  no longer necessary.

v2:
* Pass the correct slot index to ide_init_port_data().

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: fix ide_host_register() return value
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:57 +0000 (19:55 +0200)]
ide: fix ide_host_register() return value

Fix ide_host_register() to fail only if all ports cannot be registered.

While at it:

* Use host->ports[] instead of ide_hwifs[] and remove idx[].

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_host_add() helper
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:57 +0000 (19:55 +0200)]
ide: add ide_host_add() helper

Add ide_host_add() helper which does ide_host_alloc()+ide_host_register(),
then convert ide_setup_pci_device[s](), ide_legacy_device_add() and some
host drivers to use it.

While at it:

* Fix ide_setup_pci_device[s](), ide_arm.c, gayle.c, ide-4drives.c,
  macide.c, q40ide.c, cmd640.c and cs5520.c to return correct error value.

* -ENOENT -> -ENOMEM in rapide.c, ide-h8300.c, ide-generic.c, au1xxx-ide.c
  and pmac.c

* -ENODEV -> -ENOMEM in palm_bk3710.c, ide_platform.c and delkin_cb.c

* -1 -> -ENOMEM in ide-pnp.c

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add struct ide_host (take 3)
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:57 +0000 (19:55 +0200)]
ide: add struct ide_host (take 3)

* Add struct ide_host which keeps pointers to host's ports.

* Add ide_host_alloc[_all]() and ide_host_remove() helpers.

* Pass 'struct ide_host *host' instead of 'u8 *idx' to
  ide_device_add[_all]() and rename it to ide_host_register[_all]().

* Convert host drivers and core code to use struct ide_host.

* Remove no longer needed ide_find_port().

* Make ide_find_port_slot() static.

* Unexport ide_unregister().

v2:
* Add missing 'struct ide_host *host' to macide.c.

v3:
* Fix build problem in pmac.c (s/ide_alloc_host/ide_host_alloc/)
  (Noticed by Stephen Rothwell).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add struct ide_tp_ops (take 2)
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:56 +0000 (19:55 +0200)]
ide: add struct ide_tp_ops (take 2)

* Add struct ide_tp_ops for transport methods.

* Add 'const struct ide_tp_ops *tp_ops' to struct ide_port_info
  and ide_hwif_t.

* Set the default hwif->tp_ops in ide_init_port_data().

* Set host driver specific hwif->tp_ops in ide_init_port().

* Export ide_exec_command(), ide_read_status(), ide_read_altstatus(),
  ide_read_sff_dma_status(), ide_set_irq(), ide_tf_{load,read}()
  and ata_{in,out}put_data().

* Convert host drivers and core code to use struct ide_tp_ops.

* Remove no longer needed default_hwif_transport().

* Cleanup ide_hwif_t from methods that are now in struct ide_tp_ops.

While at it:

* Use struct ide_port_info in falconide.c and q40ide.c.

* Rename ata_{in,out}put_data() to ide_{in,out}put_data().

v2:

* Fix missing convertion in ns87415.c.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add 'config' field to hw_regs_t
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:56 +0000 (19:55 +0200)]
ide: add 'config' field to hw_regs_t

Add 'config' field to hw_regs_t and use it to set hwif->config_data in
ide_init_port_hw(), then convert ide_legacy_init_one() to use hw->config.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-generic: probing fix
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:56 +0000 (19:55 +0200)]
ide-generic: probing fix

* Don't skip probing IDE port if the corresponding ide_hwifs[] slot
  is already occupied.

* Remove duplicate idx[i] assignment.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agogayle: reserve memory resources at once
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:56 +0000 (19:55 +0200)]
gayle: reserve memory resources at once

* Reserve memory resources for all IDE ports at once by moving
  request_mem_region() out of 'for ()' loop and always defining
  GAYLE_IDEREG_SIZE to 0x2000.

* Keep memory resources even if no free IDE slots can be found
  (this driver is unloadable currently).

Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: filter out "default" transfer mode values in set_xfer_rate()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:56 +0000 (19:55 +0200)]
ide: filter out "default" transfer mode values in set_xfer_rate()

* Filter out "default" transfer mode values (0x00 - default PIO mode,
  0x01 - default PIO mode w/ IORDY disabled) in write handler for obsoleted
  /proc/ide/hd?/settings:current_speed setting.

  Allowing "default" transfer mode values is a dangerous thing to do as
  we don't support programming controller to the "default" transfer mode
  and devices often use different values for the default and maximum PIO
  mode (i.e. PIO2 default and PIO4 maximum) so the controller will stay
  programmed for higher PIO mode while device will use the lower PIO mode.

  There is no functionality loss as by using special IOCTLs device can
  still be programmed to "default" transfer modes (it is only useful for
  debugging/testing purposes anyway).

* Remove no longer needed IDE_HFLAG_ABUSE_SET_DMA_MODE host flag, it was
  previously used by few host drivers to program the controller to PIO0
  timings for "default" transfer mode == 0x01 (although some host drivers
  would program invalid PIO timings instead).

* Cleanup ide_set_xfer_rate() and add BUG_ON().

Suggested-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: filter out invalid transfer mode values in set_xfer_rate()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:55 +0000 (19:55 +0200)]
ide: filter out invalid transfer mode values in set_xfer_rate()

Filter out invalid transfer mode values (0x02-0x07) in write handler
for obsoleted /proc/ide/hd?/settings:current_speed setting.

While at it:

* Use XFER_UDMA_6 define.

Suggested-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove dead Virtual DMA support
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:55 +0000 (19:55 +0200)]
ide: remove dead Virtual DMA support

Lets remove dead Virtual DMA support for now so it doesn't clutter
core IDE code (it can be bring back when there is a need for it):

* Remove IDE_HFLAG_VDMA host flag.

* Remove ide_drive_t.vdma flag.

* cs5520.c: remove stale FIXMEs, cs5520_dma_host_set() and cs5520_dma_ops
  (also there is no longer a need to set IDE_HFLAG_NO_ATAPI_DMA).

There should be no functional changes caused by this patch.

Cc: TAKADA Yoshihito <takada@mbf.nifty.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: mark hpt34x host driver as BROKEN
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:55 +0000 (19:55 +0200)]
ide: mark hpt34x host driver as BROKEN

Alan Cox has noticed some time ago that hpt34x is broken
(http://lkml.org/lkml/2007/7/3/257):

"No the old driver doesn't even do that. It starts up. It spuriously
adjusts some non writable PCI BAR registers, prints a message about
resources being in use already then exits."

Reported-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: limit cy82c693 host driver to ALPHA
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:54 +0000 (19:55 +0200)]
ide: limit cy82c693 host driver to ALPHA

Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->INB, ->OUTB and ->OUTBSYNC methods
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:54 +0000 (19:55 +0200)]
ide: remove ->INB, ->OUTB and ->OUTBSYNC methods

* Remove no longer needed ->INB, ->OUTB and ->OUTBSYNC methods.

Then:

* Remove no longer used default_hwif_[mm]iops() and ide_[mm_]outbsync().

* Cleanup SuperIO handling in ns87415.c.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_read_bcount_and_ireason() helper
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:54 +0000 (19:55 +0200)]
ide: add ide_read_bcount_and_ireason() helper

Add ide_read_bcount_and_ireason() helper and use it instead of ->INB
in {cdrom_newpc,ide_pc}_intr().

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_read_ireason() helper
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:54 +0000 (19:55 +0200)]
ide: add ide_read_ireason() helper

Add ide_read_ireason() helper and use instead of ->INB for reading
ATAPI Interrupt Reason register.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_read_device() helper
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:54 +0000 (19:55 +0200)]
ide: add ide_read_device() helper

Add ide_read_device() helper and convert do_probe() to use it
instead of ->INB.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use ->tf_read in ide_read_error()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:53 +0000 (19:55 +0200)]
ide: use ->tf_read in ide_read_error()

* Add IDE_TFLAG_IN_FEATURE taskfile flag for reading Feature
  register and handle it in ->tf_read.

* Convert ide_read_error() to use ->tf_read instead of ->INB,
  then uninline and export it.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use ->tf_load in SELECT_DRIVE()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:53 +0000 (19:55 +0200)]
ide: use ->tf_load in SELECT_DRIVE()

Convert SELECT_DRIVE() to use ->tf_load instead of ->OUTB.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use ->tf_load in actual_try_to_identify()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:53 +0000 (19:55 +0200)]
ide: use ->tf_load in actual_try_to_identify()

Convert actual_try_to_identify() to use ->tf_load instead of ->OUTB.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use ->tf_load in ide_config_drive_speed()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:52 +0000 (19:55 +0200)]
ide: use ->tf_load in ide_config_drive_speed()

Convert ide_config_drive_speed() to use ->tf_load instead of ->OUTB.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: change order of register access in ide_config_drive_speed()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:52 +0000 (19:55 +0200)]
ide: change order of register access in ide_config_drive_speed()

Write ATA Feature register before ATA Sector Count register as
a preparation to use ->tf_load in ide_config_drive_speed().

This change shouldn't affect anything (just an usual paranoia).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ->set_irq method
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:52 +0000 (19:55 +0200)]
ide: add ->set_irq method

Add ->set_irq method for setting nIEN bit of ATA Device Control
register and use it instead of ide_set_irq().

While at it:

* Use ->set_irq in init_irq() and do_reset1().

* Don't use HWIF() macro in ide_check_pm_state().

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ->read_altstatus method
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:52 +0000 (19:55 +0200)]
ide: add ->read_altstatus method

* Remove ide_read_altstatus() inline helper.

* Add ->read_altstatus method for reading ATA Alternate Status
  register and use it instead of ->INB.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ->read_status method
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:52 +0000 (19:55 +0200)]
ide: add ->read_status method

* Remove ide_read_status() inline helper.

* Add ->read_status method for reading ATA Status register
  and use it instead of ->INB.

While at it:

* Don't use HWGROUP() macro.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ->exec_command method
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:51 +0000 (19:55 +0200)]
ide: add ->exec_command method

Add ->exec_command method for writing ATA Command register
and use it instead of ->OUTBSYNC.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: factor out simplex handling from ide_pci_dma_base()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:51 +0000 (19:55 +0200)]
ide: factor out simplex handling from ide_pci_dma_base()

* Factor out simplex handling from ide_pci_dma_base() to
  ide_pci_check_simplex().

* Set hwif->dma_base early in ->init_dma method / ide_hwif_setup_dma()
  and reset it in ide_init_port() if DMA initialization fails.

* Use ->read_sff_dma_status instead of ->INB in ide_pci_dma_base().

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ide_setup_dma()
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:51 +0000 (19:55 +0200)]
ide: remove ide_setup_dma()

Export sff_dma_ops and then remove ide_setup_dma().

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->dma_{status,command} fields from ide_hwif_t
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:51 +0000 (19:55 +0200)]
ide: remove ->dma_{status,command} fields from ide_hwif_t

* Use ->dma_base + offset instead of ->dma_{status,command}
  and remove no longer needed ->dma_{status,command}.

While at it:

* Use ATA_DMA_* defines.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use I/O ops directly in ide-dma.c
Bartlomiej Zolnierkiewicz [Wed, 23 Jul 2008 17:55:50 +0000 (19:55 +0200)]
ide: use I/O ops directly in ide-dma.c

Use I/O ops directly in ide_dma_host_set(), ide_dma_setup(),
ide_dma_start() and __ide_dma_end().

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>