9 years agomemory-hotplug: create /sys/firmware/memmap entry for new memory
akpm@linux-foundation.org [Fri, 5 Mar 2010 21:41:58 +0000]
memory-hotplug: create /sys/firmware/memmap entry for new memory

A memmap is a directory in sysfs which includes 3 text files: start, end
and type.  For example:

start:  0x100000
end: 0x7e7b1cff
type: System RAM

Interface firmware_map_add was not called explicitly.  Remove it and add
function firmware_map_add_hotplug as hotplug interface of memmap.

Each memory entry has a memmap in sysfs, When we hot-add new memory, sysfs
does not export memmap entry for it.  We add a call in function add_memory
to function firmware_map_add_hotplug.

Add a new function add_sysfs_fw_map_entry() to create memmap entry, it
will be called when initialize memmap and hot-add memory.

[akpm@linux-foundation.org: un-kernedoc a no longer kerneldoc comment]
Signed-off-by: Shaohui Zheng <shaohui.zheng@intel.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: fix mbind vma merge problem
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:57 +0000]
mm: fix mbind vma merge problem

Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...

This patch fixes it.

    reproduced program
    ----------------------------------------------------------------
     #include <numaif.h>
     #include <numa.h>
     #include <sys/mman.h>
     #include <stdio.h>
     #include <unistd.h>
     #include <stdlib.h>
     #include <string.h>

    static unsigned long pagesize;

    int main(int argc, char** argv)
    {
     void* addr;
     int ch;
     int node;
     struct bitmask *nmask = numa_allocate_nodemask();
     int err;
     int node_set = 0;
     char buf[128];

     while ((ch = getopt(argc, argv, "n:")) != -1){
     switch (ch){
     case 'n':
     node = strtol(optarg, NULL, 0);
     numa_bitmask_setbit(nmask, node);
     node_set = 1;
     break;
     default:
     ;
     }
     }
     argc -= optind;
     argv += optind;

     if (!node_set)
     numa_bitmask_setbit(nmask, 0);

     pagesize = getpagesize();

     addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
         MAP_ANON|MAP_PRIVATE, 0, 0);
     if (addr == MAP_FAILED)
     perror("mmap "), exit(1);

     fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);

     /* make page populate */
     memset(addr, 0, pagesize*3);

     /* first mbind */
     err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
         nmask->size, MPOL_MF_MOVE_ALL);
     if (err)
     error("mbind1 ");

     /* second mbind */
     err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
     if (err)
     error("mbind2 ");

     sprintf(buf, "cat /proc/%d/maps", getpid());
     system(buf);

     return 0;
    }
    ----------------------------------------------------------------

result without this patch

addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0

=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.

result with this patch

addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]

=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.

[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: restore zone->all_unreclaimable to independence word
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:55 +0000]
mm: restore zone->all_unreclaimable to independence word

commit e815af95 ("change all_unreclaimable zone member to flags") changed
all_unreclaimable member to bit flag.  But it had an undesireble side
effect.  free_one_page() is one of most hot path in linux kernel and
increasing atomic ops in it can reduce kernel performance a bit.

Thus, this patch revert such commit partially. at least
all_unreclaimable shouldn't share memory word with other zone flags.

[akpm@linux-foundation.org: fix patch interaction]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Huang Shijie <shijie8@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: remove free_hot_page()
Li Hong [Fri, 5 Mar 2010 21:41:54 +0000]
mm: remove free_hot_page()

free_hot_page() is just a wrapper around free_hot_cold_page() with
parameter 'cold = 0'.  After adding a clear comment for
free_hot_cold_page(), it is reasonable to remove a level of call.

[akpm@linux-foundation.org: fix build]
Signed-off-by: Li Hong <lihong.hi@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Li Ming Chun <macli@brc.ubc.ca>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Americo Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm/page_alloc.c: adjust a call site to trace_mm_page_free_direct
Li Hong [Fri, 5 Mar 2010 21:41:53 +0000]
mm/page_alloc.c: adjust a call site to trace_mm_page_free_direct

Move a call of trace_mm_page_free_direct() from free_hot_page() to
free_hot_cold_page().  It is clearer and close to kmemcheck_free_shadow(),
as it is done in function __free_pages_ok().

Signed-off-by: Li Hong <lihong.hi@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Li Ming Chun <macli@brc.ubc.ca>
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>

9 years agomm/page_alloc.c: remove duplicate call to trace_mm_page_free_direct
Li Hong [Fri, 5 Mar 2010 21:41:52 +0000]
mm/page_alloc.c: remove duplicate call to trace_mm_page_free_direct

trace_mm_page_free_direct() is called in function __free_pages().  But it
is called again in free_hot_page() if order == 0 and produce duplicate
records in trace file for mm_page_free_direct event.  As below:

K-PID    CPU#    TIMESTAMP  FUNCTION
  gnome-terminal-1567  [000]  4415.246466: mm_page_free_direct: page=ffffea0003db9f40 pfn=1155800 order=0
  gnome-terminal-1567  [000]  4415.246468: mm_page_free_direct: page=ffffea0003db9f40 pfn=1155800 order=0
  gnome-terminal-1567  [000]  4415.246506: mm_page_alloc: page=ffffea0003db9f40 pfn=1155800 order=0 migratetype=0 gfp_flags=GFP_KERNEL
  gnome-terminal-1567  [000]  4415.255557: mm_page_free_direct: page=ffffea0003db9f40 pfn=1155800 order=0
  gnome-terminal-1567  [000]  4415.255557: mm_page_free_direct: page=ffffea0003db9f40 pfn=1155800 order=0

This patch removes the first call and adds a call to
trace_mm_page_free_direct() in __free_pages_ok().

Signed-off-by: Li Hong <lihong.hi@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Li Ming Chun <macli@brc.ubc.ca>
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>

9 years agomm, lockdep: annotate reclaim context to zone reclaim too
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:47 +0000]
mm, lockdep: annotate reclaim context to zone reclaim too

Commit cf40bd16fd ("lockdep: annotate reclaim context") introduced reclaim
context annotation.  But it didn't annotate zone reclaim.  This patch do
it.

The point is, commit cf40bd16fd annotate __alloc_pages_direct_reclaim but
zone-reclaim doesn't use __alloc_pages_direct_reclaim.

current call graph is

__alloc_pages_nodemask
   get_page_from_freelist
       zone_reclaim()
   __alloc_pages_slowpath
       __alloc_pages_direct_reclaim
           try_to_free_pages

Actually, if zone_reclaim_mode=1, VM never call
__alloc_pages_direct_reclaim in usual VM pressure.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Nick Piggin <npiggin@suse.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agovmscan: get_scan_ratio() cleanup
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:47 +0000]
vmscan: get_scan_ratio() cleanup

The get_scan_ratio() should have all scan-ratio related calculations.
Thus, this patch move some calculation into get_scan_ratio.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agovmscan: check high watermark after shrink zone
Minchan Kim [Fri, 5 Mar 2010 21:41:45 +0000]
vmscan: check high watermark after shrink zone

Kswapd checks that zone has sufficient pages free via zone_watermark_ok().

If any zone doesn't have enough pages, we set all_zones_ok to zero.
!all_zone_ok makes kswapd retry rather than sleeping.

I think the watermark check before shrink_zone() is pointless.  Only after
kswapd has tried to shrink the zone is the check meaningful.

Move the check to after the call to shrink_zone().

[akpm@linux-foundation.org: fix comment, layout]
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: use rlimit helpers
Jiri Slaby [Fri, 5 Mar 2010 21:41:44 +0000]
mm: use rlimit helpers

Make sure compiler won't do weird things with limits.  E.g.  fetching them
twice may return 2 different values after writable limits are implemented.

I.e.  either use rlimit helpers added in
3e10e716abf3c71bdb5d86b8f507f9e72236c9cd ("resource: add helpers for
fetching rlimits") or ACCESS_ONCE if not applicable.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: mlock_vma_pages_range() only return success or failure
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:43 +0000]
mm: mlock_vma_pages_range() only return success or failure

Currently, mlock_vma_pages_range() only return len or 0.  then current
error handling of mmap_region() is meaningless complex.

This patch makes simplify and makes consist with brk() code.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: KAMEZAWA Hiroyuki <kamewzawa.hiroyu@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: mlock_vma_pages_range() never return negative value
KOSAKI Motohiro [Fri, 5 Mar 2010 21:41:43 +0000]
mm: mlock_vma_pages_range() never return negative value

Currently, mlock_vma_pages_range() never return negative value.  Then, we
can remove some worthless error check.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: KAMEZAWA Hiroyuki <kamewzawa.hiroyu@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: count swap usage
KAMEZAWA Hiroyuki [Fri, 5 Mar 2010 21:41:42 +0000]
mm: count swap usage

A frequent questions from users about memory management is what numbers of
swap ents are user for processes.  And this information will give some
hints to oom-killer.

Besides we can count the number of swapents per a process by scanning
/proc/<pid>/smaps, this is very slow and not good for usual process
information handler which works like 'ps' or 'top'.  (ps or top is now
enough slow..)

This patch adds a counter of swapents to mm_counter and update is at each
swap events.  Information is exported via /proc/<pid>/status file as

[kamezawa@bluextal memory]$ cat /proc/self/status
Name:   cat
State:  R (running)
Tgid:   2910
Pid:    2910
PPid:   2823
TracerPid:      0
Uid:    500     500     500     500
Gid:    500     500     500     500
FDSize: 256
Groups: 500
VmPeak:    82696 kB
VmSize:    82696 kB
VmLck:         0 kB
VmHWM:       432 kB
VmRSS:       432 kB
VmData:      172 kB
VmStk:        84 kB
VmExe:        48 kB
VmLib:      1568 kB
VmPTE:        40 kB
VmSwap:        0 kB <=============== this.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: avoid false sharing of mm_counter
KAMEZAWA Hiroyuki [Fri, 5 Mar 2010 21:41:40 +0000]
mm: avoid false sharing of mm_counter

Considering the nature of per mm stats, it's the shared object among
threads and can be a cache-miss point in the page fault path.

This patch adds per-thread cache for mm_counter.  RSS value will be
counted into a struct in task_struct and synchronized with mm's one at
events.

Now, in this patch, the event is the number of calls to handle_mm_fault.
Per-thread value is added to mm at each 64 calls.

 rough estimation with small benchmark on parallel thread (2threads) shows
 [before]
     4.5 cache-miss/faults
 [after]
     4.0 cache-miss/faults
 Anyway, the most contended object is mmap_sem if the number of threads grows.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: clean up mm_counter
KAMEZAWA Hiroyuki [Fri, 5 Mar 2010 21:41:39 +0000]
mm: clean up mm_counter

Presently, per-mm statistics counter is defined by macro in sched.h

This patch modifies it to
  - defined in mm.h as inlinf functions
  - use array instead of macro's name creation.

This patch is for reducing patch size in future patch to modify
implementation of per-mm counter.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoinfiniband: use for_each_set_bit()
Akinobu Mita [Fri, 5 Mar 2010 21:41:38 +0000]
infiniband: use for_each_set_bit()

Replace open-coded loop with for_each_set_bit().

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Acked-by: Roland Dreier <rolandd@cisco.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agobitops: rename for_each_bit() to for_each_set_bit()
Akinobu Mita [Fri, 5 Mar 2010 21:41:37 +0000]
bitops: rename for_each_bit() to for_each_set_bit()

Rename for_each_bit to for_each_set_bit in the kernel source tree.  To
permit for_each_clear_bit(), should that ever be added.

The patch includes a macro to map the old for_each_bit() onto the new
for_each_set_bit().  This is a (very) temporary thing to ease the migration.

[akpm@linux-foundation.org: add temporary for_each_bit()]
Suggested-by: Alexey Dobriyan <adobriyan@gmail.com>
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Artem Bityutskiy <dedekind@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agotimbgpio: fix build
David Miller [Fri, 5 Mar 2010 21:41:36 +0000]
timbgpio: fix build

Use of get_irq_chip_data() et al.  requires including linux/irq.h

Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: Richard Röjfors <richard.rojfors@pelagicore.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoFix a dumb typo - use of & instead of &&
Al Viro [Sat, 6 Mar 2010 18:41:07 +0000]
Fix a dumb typo - use of & instead of &&

We managed to lose O_DIRECTORY testing due to a stupid typo in commit
1f36f774b2 ("Switch !O_CREAT case to use of do_last()")

Reported-by: Walter Sheets <w41ter@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'slab-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/penber...
Linus Torvalds [Fri, 5 Mar 2010 22:35:40 +0000]
Merge branch 'slab-for-linus' of git://git./linux/kernel/git/penberg/slab-2.6

* 'slab-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/slab-2.6:
  SLUB: Fix per-cpu merge conflict
  failslab: add ability to filter slab caches
  slab: fix regression in touched logic
  dma kmalloc handling fixes
  slub: remove impossible condition
  slab: initialize unused alien cache entry as NULL at alloc_alien_cache().
  SLUB: Make slub statistics use this_cpu_inc
  SLUB: this_cpu: Remove slub kmem_cache fields
  SLUB: Get rid of dynamic DMA kmalloc cache allocation
  SLUB: Use this_cpu operations in slub

9 years agoMerge branch 'nfs-for-2.6.34' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Fri, 5 Mar 2010 21:25:45 +0000]
Merge branch 'nfs-for-2.6.34' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'nfs-for-2.6.34' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6: (44 commits)
  NFS: Remove requirement for inode->i_mutex from nfs_invalidate_mapping
  NFS: Clean up nfs_sync_mapping
  NFS: Simplify nfs_wb_page()
  NFS: Replace __nfs_write_mapping with sync_inode()
  NFS: Simplify nfs_wb_page_cancel()
  NFS: Ensure inode is always marked I_DIRTY_DATASYNC, if it has unstable pages
  NFS: Run COMMIT as an asynchronous RPC call when wbc->for_background is set
  NFS: Reduce the number of unnecessary COMMIT calls
  NFS: Add a count of the number of unstable writes carried by an inode
  NFS: Cleanup - move nfs_write_inode() into fs/nfs/write.c
  nfs41 fix NFS4ERR_CLID_INUSE for exchange id
  NFS: Fix an allocation-under-spinlock bug
  SUNRPC: Handle EINVAL error returns from the TCP connect operation
  NFSv4.1: Various fixes to the sequence flag error handling
  nfs4: renewd renew operations should take/put a client reference
  nfs41: renewd sequence operations should take/put client reference
  nfs: prevent backlogging of renewd requests
  nfs: kill renewd before clearing client minor version
  NFS: Make close(2) asynchronous when closing NFS O_DIRECT files
  NFS: Improve NFS iostat byte count accuracy for writes
  ...

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh...
Linus Torvalds [Fri, 5 Mar 2010 21:25:24 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/ericvh/v9fs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  fs/9p: Add hardlink support to .u extension
  9P2010.L handshake: .L protocol negotiation
  9P2010.L handshake: Remove "dotu" variable
  9P2010.L handshake: Add mount option
  9P2010.L handshake: Add VFS flags
  net/9p: Handle mount errors correctly.
  net/9p: Remove MAX_9P_CHAN limit
  net/9p: Add multi channel support.

9 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Fri, 5 Mar 2010 21:20:53 +0000]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6: (33 commits)
  quota: stop using QUOTA_OK / NO_QUOTA
  dquot: cleanup dquot initialize routine
  dquot: move dquot initialization responsibility into the filesystem
  dquot: cleanup dquot drop routine
  dquot: move dquot drop responsibility into the filesystem
  dquot: cleanup dquot transfer routine
  dquot: move dquot transfer responsibility into the filesystem
  dquot: cleanup inode allocation / freeing routines
  dquot: cleanup space allocation / freeing routines
  ext3: add writepage sanity checks
  ext3: Truncate allocated blocks if direct IO write fails to update i_size
  quota: Properly invalidate caches even for filesystems with blocksize < pagesize
  quota: generalize quota transfer interface
  quota: sb_quota state flags cleanup
  jbd: Delay discarding buffers in journal_unmap_buffer
  ext3: quota_write cross block boundary behaviour
  quota: drop permission checks from xfs_fs_set_xstate/xfs_fs_set_xquota
  quota: split out compat_sys_quotactl support from quota.c
  quota: split out netlink notification support from quota.c
  quota: remove invalid optimization from quota_sync_all
  ...

Fixed trivial conflicts in fs/namei.c and fs/ufs/inode.c

9 years agoMerge branch 'kvm-updates/2.6.34' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Fri, 5 Mar 2010 21:12:34 +0000]
Merge branch 'kvm-updates/2.6.34' of git://git./virt/kvm/kvm

* 'kvm-updates/2.6.34' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (145 commits)
  KVM: x86: Add KVM_CAP_X86_ROBUST_SINGLESTEP
  KVM: VMX: Update instruction length on intercepted BP
  KVM: Fix emulate_sys[call, enter, exit]()'s fault handling
  KVM: Fix segment descriptor loading
  KVM: Fix load_guest_segment_descriptor() to inject page fault
  KVM: x86 emulator: Forbid modifying CS segment register by mov instruction
  KVM: Convert kvm->requests_lock to raw_spinlock_t
  KVM: Convert i8254/i8259 locks to raw_spinlocks
  KVM: x86 emulator: disallow opcode 82 in 64-bit mode
  KVM: x86 emulator: code style cleanup
  KVM: Plan obsolescence of kernel allocated slots, paravirt mmu
  KVM: x86 emulator: Add LOCK prefix validity checking
  KVM: x86 emulator: Check CPL level during privilege instruction emulation
  KVM: x86 emulator: Fix popf emulation
  KVM: x86 emulator: Check IOPL level during io instruction emulation
  KVM: x86 emulator: fix memory access during x86 emulation
  KVM: x86 emulator: Add Virtual-8086 mode of emulation
  KVM: x86 emulator: Add group9 instruction decoding
  KVM: x86 emulator: Add group8 instruction decoding
  KVM: do not store wqh in irqfd
  ...

Trivial conflicts in Documentation/feature-removal-schedule.txt

9 years agofs/9p: Add hardlink support to .u extension
Aneesh Kumar K.V [Fri, 5 Mar 2010 20:43:43 +0000]
fs/9p: Add hardlink support to .u extension

For regular file and directories we put the link
count in th extension field in a tagged string format.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years ago9P2010.L handshake: .L protocol negotiation
Sripathi Kodi [Fri, 5 Mar 2010 18:51:04 +0000]
9P2010.L handshake: .L protocol negotiation

This patch adds 9P2010.L protocol negotiation with the server

Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years ago9P2010.L handshake: Remove "dotu" variable
Sripathi Kodi [Fri, 5 Mar 2010 18:50:14 +0000]
9P2010.L handshake: Remove "dotu" variable

Removes 'dotu' variable and make everything dependent
on 'proto_version' field.

Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years ago9P2010.L handshake: Add mount option
Sripathi Kodi [Fri, 5 Mar 2010 18:49:11 +0000]
9P2010.L handshake: Add mount option

Add new mount V9FS mount option to specify protocol version

This patch adds a new mount option to specify protocol version.
With this option it is possible to use "-o version=" switch to
specify 9P protocol version to use. Valid options for version
are:
9p2000
9p2000.u
9p2010.L

Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years ago9P2010.L handshake: Add VFS flags
Sripathi Kodi [Fri, 5 Mar 2010 18:48:00 +0000]
9P2010.L handshake: Add VFS flags

Add 9P2000.u and 9P2010.L protocol flags to V9FS VFS

This patch adds 9P2000.u and 9P2010.L protocol flags into V9FS VFS side code
and removes the single flag used for 'extended'.

Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years agonet/9p: Handle mount errors correctly.
Aneesh Kumar K.V [Mon, 15 Feb 2010 17:27:02 +0000]
net/9p: Handle mount errors correctly.

With this patch we have

# mount -t 9p -o trans=virtio virtio2 /mnt/
# mount -t 9p -o trans=virtio virtio2 /mnt/
mount: virtio2 already mounted or /mnt/ busy
mount: according to mtab, virtio2 is already mounted on /mnt
# mount -t 9p -o trans=virtio virtio3 /mnt/ -o debug=0xfff
mount: special device virtio3 does not exist

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years agonet/9p: Remove MAX_9P_CHAN limit
Aneesh Kumar K.V [Mon, 15 Feb 2010 17:27:01 +0000]
net/9p: Remove MAX_9P_CHAN limit

Use a list to track the channel instead of statically
allocated array

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years agonet/9p: Add multi channel support.
Aneesh Kumar K.V [Mon, 15 Feb 2010 17:27:00 +0000]
net/9p: Add multi channel support.

This is needed for supporting multiple mount points.

We can find out the device names to be used with mount by checking

/sys/devices/virtio-pci/virtio*/device file

if the device file have value 9 then the specific virtio device can
be used for mounting.

ex:
 #cat /sys/devices/virtio-pci/virtio1/device
 9

now we can mount using
# mount -t 9p -o trans=virtio virtio1  /mnt/

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>

9 years agoMerge branch 'writeback-for-2.6.34' into nfs-for-2.6.34
Trond Myklebust [Fri, 5 Mar 2010 20:46:18 +0000]
Merge branch 'writeback-for-2.6.34' into nfs-for-2.6.34

9 years agoNFS: Remove requirement for inode->i_mutex from nfs_invalidate_mapping
Trond Myklebust [Sat, 20 Feb 2010 01:03:30 +0000]
NFS: Remove requirement for inode->i_mutex from nfs_invalidate_mapping

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Clean up nfs_sync_mapping
Trond Myklebust [Sat, 20 Feb 2010 01:03:29 +0000]
NFS: Clean up nfs_sync_mapping

Remove the redundant call to filemap_write_and_wait().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Simplify nfs_wb_page()
Trond Myklebust [Sat, 20 Feb 2010 01:03:28 +0000]
NFS: Simplify nfs_wb_page()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Replace __nfs_write_mapping with sync_inode()
Trond Myklebust [Sat, 20 Feb 2010 01:03:26 +0000]
NFS: Replace __nfs_write_mapping with sync_inode()

Now that we have correct COMMIT semantics in writeback_single_inode, we can
reduce and simplify nfs_wb_all(). Also replace nfs_wb_nocommit() with a
call to filemap_write_and_wait(), which doesn't need to hold the
inode->i_mutex.

With that done, we can eliminate nfs_write_mapping() altogether.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Simplify nfs_wb_page_cancel()
Trond Myklebust [Sat, 20 Feb 2010 01:03:21 +0000]
NFS: Simplify nfs_wb_page_cancel()

In all cases we should be able to just remove the request and call
cancel_dirty_page().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Ensure inode is always marked I_DIRTY_DATASYNC, if it has unstable pages
Trond Myklebust [Sat, 20 Feb 2010 01:03:18 +0000]
NFS: Ensure inode is always marked I_DIRTY_DATASYNC, if it has unstable pages

Since nfs_scan_list() doesn't wait for locked pages, we have a race in
which it is possible to end up with an inode that needs to send a COMMIT,
but which does not have the I_DIRTY_DATASYNC flag set.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Run COMMIT as an asynchronous RPC call when wbc->for_background is set
Trond Myklebust [Sat, 20 Feb 2010 01:02:24 +0000]
NFS: Run COMMIT as an asynchronous RPC call when wbc->for_background is set

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Wu Fengguang <fengguang.wu@intel.com>

9 years agoNFS: Reduce the number of unnecessary COMMIT calls
Trond Myklebust [Sat, 20 Feb 2010 01:00:02 +0000]
NFS: Reduce the number of unnecessary COMMIT calls

If the caller is doing a non-blocking flush, and there are still writebacks
pending on the wire, we can usually defer the COMMIT call until those
writes are done.

Also ensure that we honour the wbc->nonblocking flag.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Add a count of the number of unstable writes carried by an inode
Trond Myklebust [Sat, 20 Feb 2010 00:53:39 +0000]
NFS: Add a count of the number of unstable writes carried by an inode

In order to know when we should do opportunistic commits of the unstable
writes, when the VM is doing a background flush, we add a field to count
the number of unstable writes.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Cleanup - move nfs_write_inode() into fs/nfs/write.c
Trond Myklebust [Sat, 20 Feb 2010 00:46:56 +0000]
NFS: Cleanup - move nfs_write_inode() into fs/nfs/write.c

The sole purpose of nfs_write_inode is to commit unstable writes, so
move it into fs/nfs/write.c, and make nfs_commit_inode static.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoMerge branch 'write_inode2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro...
Linus Torvalds [Fri, 5 Mar 2010 19:53:53 +0000]
Merge branch 'write_inode2' of git://git./linux/kernel/git/viro/vfs-2.6

* 'write_inode2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  pass writeback_control to ->write_inode
  make sure data is on disk before calling ->write_inode

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
Linus Torvalds [Fri, 5 Mar 2010 19:46:31 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  Switch !O_CREAT case to use of do_last()
  Get rid of symlink body copying
  Finish pulling of -ESTALE handling to upper level in do_filp_open()
  Turn do_link spaghetty into a normal loop
  Unify exits in O_CREAT handling
  Kill is_link argument of do_last()
  Pull handling of LAST_BIND into do_last(), clean up ok: part in do_filp_open()
  Leave mangled flag only for setting nd.intent.open.flag
  Get rid of passing mangled flag to do_last()
  Don't pass mangled open_flag to finish_open()
  pull more into do_last()
  bail out with ELOOP earlier in do_link loop
  pull the common predecessors into do_last()
  postpone __putname() until after do_last()
  unroll do_last: loop in do_filp_open()
  Shift releasing nd->root from do_last() to its caller
  gut do_filp_open() a bit more (do_last separation)
  beginning to untangle do_filp_open()

9 years agox86: fix mtrr missing kernel-doc
Randy Dunlap [Fri, 5 Mar 2010 17:52:52 +0000]
x86: fix mtrr missing kernel-doc

Fix missing kernel-doc notation in mtrr/main.c:

Warning(arch/x86/kernel/cpu/mtrr/main.c:152): No description found for parameter 'info'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'perf-probes-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 5 Mar 2010 18:50:22 +0000]
Merge branch 'perf-probes-for-linus-2' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-probes-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: Issue at least one memory barrier in stop_machine_text_poke()
  perf probe: Correct probe syntax on command line help
  perf probe: Add lazy line matching support
  perf probe: Show more lines after last line
  perf probe: Check function address range strictly in line finder
  perf probe: Use libdw callback routines
  perf probe: Use elfutils-libdw for analyzing debuginfo
  perf probe: Rename probe finder functions
  perf probe: Fix bugs in line range finder
  perf probe: Update perf probe document
  perf probe: Do not show --line option without dwarf support
  kprobes: Add documents of jump optimization
  kprobes/x86: Support kprobes jump optimization on x86
  x86: Add text_poke_smp for SMP cross modifying code
  kprobes/x86: Cleanup save/restore registers
  kprobes/x86: Boost probes when reentering
  kprobes: Jump optimization sysctl interface
  kprobes: Introduce kprobes jump optimization
  kprobes: Introduce generic insn_slot framework
  kprobes/x86: Cleanup RELATIVEJUMP_INSTRUCTION to RELATIVEJUMP_OPCODE

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Fri, 5 Mar 2010 18:47:57 +0000]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  padata: Allocate the cpumask for the padata instance
  crypto: authenc - Move saved IV in front of the ablkcipher request
  crypto: hash - Fix handling of unaligned buffers
  crypto: authenc - Use correct ahash complete functions
  crypto: md5 - Set statesize

9 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Fri, 5 Mar 2010 18:47:00 +0000]
Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (36 commits)
  ext4: fix up rb_root initializations to use RB_ROOT
  ext4: Code cleanup for EXT4_IOC_MOVE_EXT ioctl
  ext4: Fix the NULL reference in double_down_write_data_sem()
  ext4: Fix insertion point of extent in mext_insert_across_blocks()
  ext4: consolidate in_range() definitions
  ext4: cleanup to use ext4_grp_offs_to_block()
  ext4: cleanup to use ext4_group_first_block_no()
  ext4: Release page references acquired in ext4_da_block_invalidatepages
  ext4: Fix ext4_quota_write cross block boundary behaviour
  ext4: Convert BUG_ON checks to use ext4_error() instead
  ext4: Use direct_IO_no_locking in ext4 dio read
  ext4: use ext4_get_block_write in buffer write
  ext4: mechanical rename some of the direct I/O get_block's identifiers
  ext4: make "offset" consistent in ext4_check_dir_entry()
  ext4: Handle non empty on-disk orphan link
  ext4: explicitly remove inode from orphan list after failed direct io
  ext4: fix error handling in migrate
  ext4: deprecate obsoleted mount options
  ext4: Fix fencepost error in chosing choosing group vs file preallocation.
  jbd2: clean up an assertion in jbd2_journal_commit_transaction()
  ...

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus
Linus Torvalds [Fri, 5 Mar 2010 18:46:04 +0000]
Merge git://git./linux/kernel/git/pkl/squashfs-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus:
  Squashfs: get rid of obsolete definition in header file
  Squashfs: get rid of obsolete variable in struct squashfs_sb_info
  Squashfs: add decompressor entries for lzma and lzo
  Squashfs: add a decompressor framework
  Squashfs: factor out remaining zlib dependencies into separate wrapper file
  Squashfs: move zlib decompression wrapper code into a separate file

9 years agopass writeback_control to ->write_inode
Christoph Hellwig [Fri, 5 Mar 2010 08:21:37 +0000]
pass writeback_control to ->write_inode

This gives the filesystem more information about the writeback that
is happening.  Trond requested this for the NFS unstable write handling,
and other filesystems might benefit from this too by beeing able to
distinguish between the different callers in more detail.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agomake sure data is on disk before calling ->write_inode
Christoph Hellwig [Fri, 5 Mar 2010 08:21:21 +0000]
make sure data is on disk before calling ->write_inode

Similar to the fsync issue fixed a while ago in commit
2daea67e966dc0c42067ebea015ddac6834cef88 we need to write for data to
actually hit the disk before writing out the metadata to guarantee
data integrity for filesystems that modify the inode in the data I/O
completion path.  Currently XFS and NFS handle this manually, and AFS
has a write_inode method that does nothing but waiting for data, while
others are possibly missing out on this.

Fortunately this change has a lot less impact than the fsync change
as none of the write_inode methods starts data writeout of any form
by itself.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoSquashfs: get rid of obsolete definition in header file
Phillip Lougher [Thu, 25 Feb 2010 01:31:13 +0000]
Squashfs: get rid of obsolete definition in header file

Signed-off-by: Phillip Lougher <phillip@lougher.demon.co.uk>

9 years agoSquashfs: get rid of obsolete variable in struct squashfs_sb_info
Phillip Lougher [Thu, 25 Feb 2010 00:54:48 +0000]
Squashfs: get rid of obsolete variable in struct squashfs_sb_info

Signed-off-by: Phillip Lougher <phillip@lougher.demon.co.uk>

9 years agoSwitch !O_CREAT case to use of do_last()
Al Viro [Sat, 26 Dec 2009 15:56:19 +0000]
Switch !O_CREAT case to use of do_last()

... and now we have all intents crap well localized

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoGet rid of symlink body copying
Al Viro [Sat, 26 Dec 2009 13:37:05 +0000]
Get rid of symlink body copying

Now that nd->last stays around until ->put_link() is called, we can
just postpone that ->put_link() in do_filp_open() a bit and don't
bother with copying.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoFinish pulling of -ESTALE handling to upper level in do_filp_open()
Al Viro [Sat, 26 Dec 2009 12:21:48 +0000]
Finish pulling of -ESTALE handling to upper level in do_filp_open()

Don't bother with path_walk() (and its retry loop); link_path_walk()
will do it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoTurn do_link spaghetty into a normal loop
Al Viro [Sat, 26 Dec 2009 12:16:40 +0000]
Turn do_link spaghetty into a normal loop

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoUnify exits in O_CREAT handling
Al Viro [Sat, 26 Dec 2009 12:09:49 +0000]
Unify exits in O_CREAT handling

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoKill is_link argument of do_last()
Al Viro [Sat, 26 Dec 2009 12:04:50 +0000]
Kill is_link argument of do_last()

We set it to 1 iff we return NULL

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoPull handling of LAST_BIND into do_last(), clean up ok: part in do_filp_open()
Al Viro [Sat, 26 Dec 2009 12:01:01 +0000]
Pull handling of LAST_BIND into do_last(), clean up ok: part in do_filp_open()

Note that in case of !O_CREAT we know that nd.root has already been given up

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoLeave mangled flag only for setting nd.intent.open.flag
Al Viro [Thu, 24 Dec 2009 12:15:41 +0000]
Leave mangled flag only for setting nd.intent.open.flag

Nothing else uses it anymore

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoGet rid of passing mangled flag to do_last()
Al Viro [Thu, 24 Dec 2009 11:51:13 +0000]
Get rid of passing mangled flag to do_last()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoDon't pass mangled open_flag to finish_open()
Al Viro [Thu, 24 Dec 2009 11:49:47 +0000]
Don't pass mangled open_flag to finish_open()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agopull more into do_last()
Al Viro [Thu, 24 Dec 2009 08:39:50 +0000]
pull more into do_last()

Handling of LAST_DOT/LAST_ROOT/LAST_DOTDOT/terminating slash
can be pulled in as well

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agobail out with ELOOP earlier in do_link loop
Al Viro [Thu, 24 Dec 2009 07:27:30 +0000]
bail out with ELOOP earlier in do_link loop

If we'd passed through 32 trailing symlinks already, there's
no sense following the 33rd - we'll bail out anyway.  Better
bugger off earlier.

It *does* change behaviour, after a fashion - if the 33rd happens
to be a procfs-style symlink, original code *would* allow it.
This one will not.  Cry me a river if that hurts you.  Please, do.
And post a video of that, while you are at it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agopull the common predecessors into do_last()
Al Viro [Thu, 24 Dec 2009 07:12:06 +0000]
pull the common predecessors into do_last()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agopostpone __putname() until after do_last()
Al Viro [Thu, 24 Dec 2009 07:08:19 +0000]
postpone __putname() until after do_last()

Since do_last() doesn't mangle nd->last_name, we can safely postpone
__putname() done in handling of trailing symlinks until after the
call of do_last()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agounroll do_last: loop in do_filp_open()
Al Viro [Thu, 24 Dec 2009 07:05:43 +0000]
unroll do_last: loop in do_filp_open()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoShift releasing nd->root from do_last() to its caller
Al Viro [Thu, 24 Dec 2009 07:02:38 +0000]
Shift releasing nd->root from do_last() to its caller

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agogut do_filp_open() a bit more (do_last separation)
Al Viro [Thu, 24 Dec 2009 06:58:28 +0000]
gut do_filp_open() a bit more (do_last separation)

Brute-force separation of stuff reachable from do_last: with
the exception of do_link:; just take all that crap to a helper
function as-is and have it tell the caller if it has to go
to do_link.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agobeginning to untangle do_filp_open()
Al Viro [Thu, 24 Dec 2009 06:26:48 +0000]
beginning to untangle do_filp_open()

That's going to be a long and painful series.  The first step:
take the stuff reachable from 'ok' label in do_filp_open() into
a new helper (finish_open()).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

9 years agoext4: fix up rb_root initializations to use RB_ROOT
Venkatesh Pallipadi [Fri, 5 Mar 2010 03:25:21 +0000]
ext4: fix up rb_root initializations to use RB_ROOT

ext4 uses rb_node = NULL; to zero rb_root at few places.  Using
RB_ROOT as the initializer is more portable in case the underlying
implementation of rbtrees changes in the future.

Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Eric Paris <eparis@redhat.com>

9 years agoquota: stop using QUOTA_OK / NO_QUOTA
Christoph Hellwig [Wed, 3 Mar 2010 14:05:08 +0000]
quota: stop using QUOTA_OK / NO_QUOTA

Just use 0 / -EDQUOT directly - that's what it translates to anyway.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: cleanup dquot initialize routine
Christoph Hellwig [Wed, 3 Mar 2010 14:05:07 +0000]
dquot: cleanup dquot initialize routine

Get rid of the initialize dquot operation - it is now always called from
the filesystem and if a filesystem really needs it's own (which none
currently does) it can just call into it's own routine directly.

Rename the now static low-level dquot_initialize helper to __dquot_initialize
and vfs_dq_init to dquot_initialize to have a consistent namespace.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: move dquot initialization responsibility into the filesystem
Christoph Hellwig [Wed, 3 Mar 2010 14:05:06 +0000]
dquot: move dquot initialization responsibility into the filesystem

Currently various places in the VFS call vfs_dq_init directly.  This means
we tie the quota code into the VFS.  Get rid of that and make the
filesystem responsible for the initialization.   For most metadata operations
this is a straight forward move into the methods, but for truncate and
open it's a bit more complicated.

For truncate we currently only call vfs_dq_init for the sys_truncate case
because open already takes care of it for ftruncate and open(O_TRUNC) - the
new code causes an additional vfs_dq_init for those which is harmless.

For open the initialization is moved from do_filp_open into the open method,
which means it happens slightly earlier now, and only for regular files.
The latter is fine because we don't need to initialize it for operations
on special files, and we already do it as part of the namespace operations
for directories.

Add a dquot_file_open helper that filesystems that support generic quotas
can use to fill in ->open.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: cleanup dquot drop routine
Christoph Hellwig [Wed, 3 Mar 2010 14:05:05 +0000]
dquot: cleanup dquot drop routine

Get rid of the drop dquot operation - it is now always called from
the filesystem and if a filesystem really needs it's own (which none
currently does) it can just call into it's own routine directly.

Rename the now static low-level dquot_drop helper to __dquot_drop
and vfs_dq_drop to dquot_drop to have a consistent namespace.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: move dquot drop responsibility into the filesystem
Christoph Hellwig [Wed, 3 Mar 2010 14:05:04 +0000]
dquot: move dquot drop responsibility into the filesystem

Currently clear_inode calls vfs_dq_drop directly.  This means
we tie the quota code into the VFS.  Get rid of that and make the
filesystem responsible for the drop inside the ->clear_inode
superblock operation.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: cleanup dquot transfer routine
Christoph Hellwig [Wed, 3 Mar 2010 14:05:03 +0000]
dquot: cleanup dquot transfer routine

Get rid of the transfer dquot operation - it is now always called from
the filesystem and if a filesystem really needs it's own (which none
currently does) it can just call into it's own routine directly.

Rename the now static low-level dquot_transfer helper to __dquot_transfer
and vfs_dq_transfer to dquot_transfer to have a consistent namespace,
and make the new dquot_transfer return a normal negative errno value
which all callers expect.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: move dquot transfer responsibility into the filesystem
Christoph Hellwig [Wed, 3 Mar 2010 14:05:02 +0000]
dquot: move dquot transfer responsibility into the filesystem

Currently notify_change calls vfs_dq_transfer directly.  This means
we tie the quota code into the VFS.  Get rid of that and make the
filesystem responsible for the transfer.  Most filesystems already
do this, only ufs and udf need the code added, and for jfs it needs to
be enabled unconditionally instead of only when ACLs are enabled.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: cleanup inode allocation / freeing routines
Christoph Hellwig [Wed, 3 Mar 2010 14:05:01 +0000]
dquot: cleanup inode allocation / freeing routines

Get rid of the alloc_inode and free_inode dquot operations - they are
always called from the filesystem and if a filesystem really needs
their own (which none currently does) it can just call into it's
own routine directly.

Also get rid of the vfs_dq_alloc/vfs_dq_free wrappers and always
call the lowlevel dquot_alloc_inode / dqout_free_inode routines
directly, which now lose the number argument which is always 1.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agodquot: cleanup space allocation / freeing routines
Christoph Hellwig [Wed, 3 Mar 2010 14:05:00 +0000]
dquot: cleanup space allocation / freeing routines

Get rid of the alloc_space, free_space, reserve_space, claim_space and
release_rsv dquot operations - they are always called from the filesystem
and if a filesystem really needs their own (which none currently does)
it can just call into it's own routine directly.

Move shared logic into the common __dquot_alloc_space,
dquot_claim_space_nodirty and __dquot_free_space low-level methods,
and rationalize the wrappers around it to move as much as possible
code into the common block for CONFIG_QUOTA vs not.  Also rename
all these helpers to be named dquot_* instead of vfs_dq_*.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoext3: add writepage sanity checks
Dmitry Monakhov [Tue, 2 Mar 2010 12:51:02 +0000]
ext3: add writepage sanity checks

- There is theoretical possibility to perform writepage on
   RO superblock. Add explicit check for what case.
- Page must being locked before writepage.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoext3: Truncate allocated blocks if direct IO write fails to update i_size
Jan Kara [Mon, 1 Mar 2010 13:02:37 +0000]
ext3: Truncate allocated blocks if direct IO write fails to update i_size

We have to truncate blocks allocated to file during direct IO when we
fail to update i_size properly.

Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: Properly invalidate caches even for filesystems with blocksize < pagesize
Jan Kara [Mon, 22 Feb 2010 20:07:17 +0000]
quota: Properly invalidate caches even for filesystems with blocksize < pagesize

Sometimes invalidate_bdev() can fail to invalidate a part of block
device cache because of dirty data. If the filesystem has blocksize
smaller than page size, this can happen even for pages containing
quota files and thus kernel would operate on stale data. Fix the
issue by syncing the filesystem before invalidating the cache.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: generalize quota transfer interface
Dmitry Monakhov [Tue, 16 Feb 2010 05:31:50 +0000]
quota: generalize quota transfer interface

Current quota transfer interface support only uid/gid.
This patch extend interface in order to support various quotas types
The goal is accomplished without changes in most frequently used
vfs_dq_transfer() func.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: sb_quota state flags cleanup
Dmitry Monakhov [Tue, 16 Feb 2010 05:31:49 +0000]
quota: sb_quota state flags cleanup

- remove hardcoded USRQUOTA/GRPQUOTA flags
- convert int to bool for appropriate functions

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agojbd: Delay discarding buffers in journal_unmap_buffer
Jan Kara [Tue, 16 Feb 2010 19:37:12 +0000]
jbd: Delay discarding buffers in journal_unmap_buffer

Delay discarding buffers in journal_unmap_buffer until
we know that "add to orphan" operation has definitely been
committed, otherwise the log space of committing transation
may be freed and reused before truncate get committed, updates
may get lost if crash happens.

This patch is a backport of JBD2 fix by dingdinghua <dingdinghua@nrchpc.ac.cn>.

Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoext3: quota_write cross block boundary behaviour
Dmitry Monakhov [Tue, 16 Feb 2010 16:33:42 +0000]
ext3: quota_write cross block boundary behaviour

We always assume what dquot update result in changes in one data block
But ext3_quota_write() function may handle cross block boundary writes
In fact if this ever happen it will result in incorrect journal credits
reservation. And later bug_on triggering. As soon this never happen the
boundary cross loop is NOOP. In order to make things straight
let's remove this loop and assert cross boundary condition.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: drop permission checks from xfs_fs_set_xstate/xfs_fs_set_xquota
Christoph Hellwig [Tue, 16 Feb 2010 08:44:56 +0000]
quota: drop permission checks from xfs_fs_set_xstate/xfs_fs_set_xquota

We already do these checks in the generic code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: split out compat_sys_quotactl support from quota.c
Christoph Hellwig [Tue, 16 Feb 2010 08:44:55 +0000]
quota: split out compat_sys_quotactl support from quota.c

Instead of adding ifdefs just split it into a new file.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: split out netlink notification support from quota.c
Christoph Hellwig [Tue, 16 Feb 2010 08:44:54 +0000]
quota: split out netlink notification support from quota.c

Instead of adding ifdefs just split it into a new file.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: remove invalid optimization from quota_sync_all
Christoph Hellwig [Tue, 16 Feb 2010 08:44:53 +0000]
quota: remove invalid optimization from quota_sync_all

Checking the "VFS" quota enabled and dirty bits from generic code means
this code will never get called for other implementations, e.g. XFS and
GFS2.  Grabbing the reference on the superblock really isn't much overhead
for a global Q_SYNC call, so just drop this optimization.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: move code from sync_quota_sb into vfs_quota_sync
Christoph Hellwig [Tue, 16 Feb 2010 08:44:52 +0000]
quota: move code from sync_quota_sb into vfs_quota_sync

Currenly sync_quota_sb does a lot of sync and truncate action that only
applies to "VFS" style quotas and is actively harmful for the sync
performance in XFS.  Move it into vfs_quota_sync and add a wait parameter
to ->quota_sync to tell if we need it or not.

My audit of the GFS2 code says it's also not needed given the way GFS2
implements quotas, but I'd be happy if this can get a detailed review.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: clean up Q_XQUOTASYNC
Christoph Hellwig [Tue, 16 Feb 2010 08:44:51 +0000]
quota: clean up Q_XQUOTASYNC

Currently Q_XQUOTASYNC calls into the quota_sync method, but XFS does something
entirely different in it than the rest of the filesystems.  xfs_quota which
calls Q_XQUOTASYNC expects an asynchronous data writeout to flush delayed
allocations, while the "VFS" quota support wants to flush changes to the quota
file.

So make Q_XQUOTASYNC call into the writeback code directly and make the
quota_sync method optional as XFS doesn't need in the sense expected by the
rest of the quota code.

GFS2 was using limited XFS-style quota and has a quota_sync method fitting
neither the style used by vfs_quota_sync nor xfs_fs_quota_sync.  I left it
in for now as per discussion with Steve it expects to be called from the
sync path this way.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: simplify permission checking
Christoph Hellwig [Tue, 16 Feb 2010 08:44:50 +0000]
quota: simplify permission checking

Stop having complicated different routines for checking permissions for
XQM vs "VFS" quotas.  Instead do the checks for having sb->s_qcop and
a valid type directly in do_quotactl, and munge the *quotactl_valid functions
into a check_quotactl_permission helper that only checks for permissions.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: special case Q_SYNC without device name
Christoph Hellwig [Tue, 16 Feb 2010 08:44:49 +0000]
quota: special case Q_SYNC without device name

The Q_SYNC command can be called without the path to a device, in which case
it iterates over all superblocks.  Special case this variant directly in
sys_quotactl so that the other code always gets a superblock and doesn't
need to deal with this case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: clean up checks for supported quota methods
Christoph Hellwig [Tue, 16 Feb 2010 08:44:48 +0000]
quota: clean up checks for supported quota methods

Move the checks for sb->s_qcop->foo next to the actual calls for them, same
for sb_has_quota_active checks where applicable.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: split do_quotactl
Christoph Hellwig [Tue, 16 Feb 2010 08:44:47 +0000]
quota: split do_quotactl

Split out a helper for each non-trivial command from do_quotactl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

9 years agoquota: Fix warning when a delayed write happens before quota is enabled
Jan Kara [Tue, 9 Feb 2010 17:20:39 +0000]
quota: Fix warning when a delayed write happens before quota is enabled

If a delayed-allocation write happens before quota is enabled, the
kernel spits out a warning:
WARNING: at fs/quota/dquot.c:988 dquot_claim_space+0x77/0x112()

because the fact that user has some delayed allocation is not recorded
in quota structure.

Make dquot_initialize() update amount of reserved space for user if it sees
inode has some space reserved. Also make sure that reserved quota space does
not go negative and we warn about the filesystem bug just once.

Signed-off-by: Jan Kara <jack@suse.cz>