Andrew Morton bd4c8ce41a [PATCH] invalidate_inode_pages2(): ignore page refcounts ...

The recent fix to invalidate_inode_pages() (git commit 016eb4a) managed to
unfix invalidate_inode_pages2().

The problem is that various bits of code in the kernel can take transient refs
on pages: the page scanner will do this when inspecting a batch of pages, and
the lru_cache_add() batching pagevecs also hold a ref.

Net result is transient failures in invalidate_inode_pages2().  This affects
NFS directory invalidation (observed) and presumably also block-backed
direct-io (not yet reported).

Fix it by reverting invalidate_inode_pages2() back to the old version which
ignores the page refcounts.

We may come up with something more clever later, but for now we need a 2.6.18
fix for NFS.

Cc: Chuck Lever <cel@citi.umich.edu>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

2006-10-01 00:39:33 -07:00

..

allocpercpu.c

[PATCH] Extract the allocpercpu functions from the slab allocator

2006-09-26 08:48:51 -07:00

bootmem.c

[PATCH] bootmem: use MAX_DMA_ADDRESS instead of LOW32LIMIT

2006-09-26 08:48:49 -07:00

bounce.c

[PATCH] BLOCK: Separate the bounce buffering code from the highmem code [try #6]

2006-09-30 20:32:11 +02:00

fadvise.c

…

filemap_xip.c

…

filemap.c

[PATCH] Streamline generic_file_* interfaces and filemap cleanups

2006-10-01 00:39:28 -07:00

filemap.h

…

fremap.c

…

highmem.c

[PATCH] BLOCK: Separate the bounce buffering code from the highmem code [try #6]

2006-09-30 20:32:11 +02:00

hugetlb.c

[PATCH] NUMA: Add zone_to_nid function

2006-09-26 08:48:52 -07:00

internal.h

…

Kconfig

[PATCH] hot-add-mem x86_64: Kconfig changes

2006-10-01 00:39:18 -07:00

madvise.c

…

Makefile

[PATCH] BLOCK: Make it possible to disable the block layer [try #6]

2006-09-30 20:52:31 +02:00

memory_hotplug.c

[PATCH] hot-add-mem x86_64: use CONFIG_MEMORY_HOTPLUG_RESERVE

2006-10-01 00:39:18 -07:00

memory.c

[PATCH] mm: fix a race condition under SMC + COW

2006-09-29 09:18:03 -07:00

mempolicy.c

[PATCH] kmemdup: some users

2006-10-01 00:39:19 -07:00

mempool.c

…

migrate.c

[PATCH] BLOCK: Make it possible to disable the block layer [try #6]

2006-09-30 20:52:31 +02:00

mincore.c

…

mlock.c

…

mmap.c

[PATCH] ZVC: Support NR_SLAB_RECLAIMABLE / NR_SLAB_UNRECLAIMABLE

2006-09-26 08:48:51 -07:00

mmzone.c

…

mprotect.c

…

mremap.c

…

msync.c

…

nommu.c

[PATCH] NOMMU: don't try and give NULL to fput()

2006-10-01 00:39:17 -07:00

oom_kill.c

[PATCH] oom: don't kill current when another OOM in progress

2006-09-29 09:18:21 -07:00

page_alloc.c

[PATCH] mm/page_alloc: use NULL instead of 0 for ptr

2006-09-27 08:26:13 -07:00

page_io.c

[PATCH] swsusp: read speedup

2006-09-26 08:48:58 -07:00

page-writeback.c

[PATCH] BLOCK: Make it possible to disable the block layer [try #6]

2006-09-30 20:52:31 +02:00

pdflush.c

…

prio_tree.c

…

readahead.c

…

rmap.c

…

shmem_acl.c

[PATCH] Access Control Lists for tmpfs

2006-09-29 09:18:24 -07:00

shmem.c

[PATCH] r/o bind mount prepwork: inc_nlink() helper

2006-10-01 00:39:30 -07:00

slab.c

[PATCH] single bit flip detector

2006-09-29 09:18:10 -07:00

slob.c

[PATCH] Make kmem_cache_destroy() return void

2006-09-27 08:26:11 -07:00

sparse.c

…

swap_state.c

…

swap.c

…

swapfile.c

[PATCH] valid_swaphandles() fix

2006-09-29 09:18:23 -07:00

thrash.c

…

tiny-shmem.c

…

truncate.c

[PATCH] invalidate_inode_pages2(): ignore page refcounts

2006-10-01 00:39:33 -07:00

util.c

[PATCH] kmemdup: introduce

2006-10-01 00:39:19 -07:00

vmalloc.c

[PATCH] Mark __remove_vm_area() static

2006-09-27 08:26:13 -07:00

vmscan.c

[PATCH] page invalidation cleanup

2006-09-27 08:26:12 -07:00

vmstat.c

[PATCH] zone_statistics: Use hot node instead of cold zone_pgdat

2006-09-27 08:26:13 -07:00