Vlad Apostolov [Thu, 11 Oct 2007 07:44:18 +0000 (17:44 +1000)]
[XFS] get_bulkall() could return incorrect inode state
In the following scenario xfs_bulkstat() returns incorrect stale inode
state:
1. File_A is created and its inode synced to disk. 2. File_A is unlinked
and doesn't exist anymore. 3. Filesystem sync is invoked. 4. File_B is
created. File_B happens to reclaim File_A's inode. 5. xfs_bulkstat() is
called and detects File_B but reports the
incorrect File_A inode state.
Explanation for the incorrect inode state is that inodes are not
immediately synced on file create for performance reasons. This leaves the
on-disk inode buffer uninitialized (or with old state from a previous
generation inode) and this is what xfs_bulkstat() would report.
The patch marks the on-disk inode buffer "dirty" on unlink. When the inode
is reclaimed (by a new file create), xfs_bulkstat() would filter this
inode by the "dirty" mark. Once the inode is flushed to disk, the on-disk
buffer "dirty" mark is automatically removed and a following
xfs_bulkstat() would return the correct inode state.
Marking the on-disk inode buffer "dirty" on unlink is achieved by setting
the on-disk di_nlink field to 0. Note that the in-core di_nlink has
already been set to 0 and a corresponding transaction logged by
xfs_droplink(). This is an exception from the rule that any on-disk inode
buffer changes has to be followed by a disk write (inode flush).
Synchronizing the in-core to on-disk di_nlink values in advance (before
the actual inode flush to disk) should be fine in this case because the
inode is already unlinked and it would never change its di_nlink again for
this inode generation.
Signed-off-by: Vlad Apostolov <vapo@sgi.com> Signed-off-by: Alex Elder <aelder@sgi.com> Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Mark Goodwin <markgw@sgi.com> Signed-off-by: Tim Shimmin <tes@sgi.com>
[XFS] fix when DMAPI mount option processing happens
Fix for a regression caused by a recent patch
that moved the DMAPI mount option processing inside xfs_parseargs(). The
DMAPI mount option used to be processed in the DMAPI module loaded before
xfs_parseargs() was invoked.
[XFS] ensure file size is logged on synchronous writes
Synchronous writes currently log inode changes before syncing pages to
disk. Since the file size is updated on I/O completion we wont be writing
out the updated file size and if we crash the file will have the wrong
size. This change moves the logging after the syncing of the pages to
ensure we log the correct file size.
All flags are added to xfs_mount's m_flag instead. Note that the 32bit
inode flag was duplicated in both of them, but only cleared in the mount
when it was not nessecary due to the filesystem beeing small enough. Two
flags are still required here - one to indicate the mount option setting,
and one to indicate if it applies or not.
[XFS] kill the vfs_fsid and vfs_altfsid members in struct bhv_vfs
vfs_altfsid was just a pointer to mp->m_fixedfsid so we can trivially
replace it with the latter. vfs_fsid also was identical to m_fixedfsid
through rather obfuscated ways so we can kill it as well and simply its
only user.
[XFS] remove dependency of the quota module on behaviors
Mount options are now parsed by the main XFS module and rejected if quota
support is not available, and there are some new quota operation for the
quotactl syscall and calls to quote in the mount, unmount and sync
callchains.
[XFS] remove dependency of the dmapi module on behaviors
Mount options are now parsed by the main XFS module and rejected if dmapi
support is not available, and there is a new dm operation to send the
mount event.
[XFS] move freeing the mount structure from xfs_mount_free into the callers
In the next patch we need to look at the mount structure until just before
it's freed, so we need to be able to free it as the very last thing in
xfs_unmount.
Now that struct bhv_vnode is empty we can just kill it. Retain bhv_vnode_t
as a typedef for struct inode for the time being until all the fallout is
cleaned up.
struct bhv_vnode is on it's way out, so move the trace buffer to the XFS
inode. Note that this makes the tracing macros rather misnamed, but this
kind of fallout will be fixed up incrementally later on.
All flags previously handled at the vnode level are not in the xfs_inode
where we already have a flags mechanisms and free bits for flags
previously in the vnode.
[XFS] decontaminate vnode operations from behavior details
All vnode ops now take struct xfs_inode pointers and the behaviour related
glue is split out into methods of it's own. This required fixing
xfs_create/mkdir/symlink to not mess with the inode pointer but rather use
a separate boolean for error handling. Thanks to Dave Chinner for that
fix.
Vlad Apostolov [Tue, 28 Aug 2007 04:00:28 +0000 (14:00 +1000)]
[XFS] do not have XFSMNT_IDELETE as default when mounted with XFSMNT_DMAPI
XFS inodes are dynamically allocated on demand, rather than being
allocated at mkfs time. Chunks of 64 inodes are allocated at once, but
they are never freed. Over time, this can lead to filesystem
fragmentation, clusters of inodes and the btrees which point at them can
be scattered around the system.
By freeing clusters as they are emptied, we will reduce fragmentation of
the free space after removing files. This in turn will allow us to make
better placement decisions when repopulating a filesystem. The
XFSMNT_IDELETE mount option enables freeing clusters when they get empty.
Unfortunately a side effect of freeing inode clusters is that the inode
generation numbers of such inodes would be reset to zero when the cluster
is reclaimed. This is a problem in particular for a DMAPI enabled
filesystem as the the DMAPI handles need to be unique and persistent in
time. An unique DMAPI handle is built with the help of the inode
generation number. When the last one is prematurely reset by an inode
cluster reclaim, there is a high probability of different generation
inodes to end up having identical DMAPI handles.
To avoid the problem with identical DMAPI handles, the XFSMNT_IDELETE
mount option should be set as default, only if the filesystem is not
mounted with XFSMNT_DMAPI.
David Chinner [Tue, 28 Aug 2007 04:00:13 +0000 (14:00 +1000)]
[XFS] Radix tree based inode caching
One of the perpetual scaling problems XFS has is indexing it's incore
inodes. We currently uses hashes and the default hash sizes chosen can
only ever be a tradeoff between memory consumption and the maximum
realistic size of the cache.
As a result, anyone who has millions of inodes cached on a filesystem
needs to tunes the size of the cache via the ihashsize mount option to
allow decent scalability with inode cache operations.
A further problem is the separate inode cluster hash, whose size is based
on the ihashsize but is smaller, and so under certain conditions (sparse
cluster cache population) this can become a limitation long before the
inode hash is causing issues.
The following patchset removes the inode hash and cluster hash and
replaces them with radix trees to avoid the scalability limitations of the
hashes. It also reduces the size of the inodes by 3 pointers....
Currently xfs has a rather complicated internal scheme to allow for
different directory formats in IRIX. This patch rips all code related to
this out and pushes useage of the Linux filldir callback into the lowlevel
directory code. This does not make the code any less portable because
filldir can be used to create dirents of all possible variations
(including the IRIX ones as proved by the IRIX binary emulation code under
arch/mips/).
This patch get rid of an unessecary copy in the readdir path, about 400
lines of code and one of the last two users of the uio structure.
This version is updated to deal with dmapi aswell which greatly simplifies
the get_dirattrs code. The dmapi part has been tested using the
get_dirattrs tools from the xfstest dmapi suite1 with various small and
large directories.
Creates a new xfs_dsb_t that is __be annotated and keeps xfs_sb_t for the
incore one. xfs_xlatesb is renamed to xfs_sb_to_disk and only handles the
incore -> disk conversion. A new helper xfs_sb_from_disk handles the other
direction and doesn't need the slightly hacky table-driven approach
because we only ever read the full sb from disk.
The handling of shared r/o filesystems has been buggy on little endian
system and fixing this required shuffling around of some code in that
area.
Biggest bit is duplicating the dinode structure so we have one annotated for
native endianess and one for disk endianess. The other significant change
is that xfs_xlate_dinode_core is split into one helper per direction to
allow for proper annotations, everything else is trivial.
As a sidenode splitting out the incore dinode means we can move it into
xfs_inode.h in a later patch and severely improving on the include hell in
xfs.
Jesper Juhl [Thu, 16 Aug 2007 06:25:42 +0000 (16:25 +1000)]
[XFS] Fix a potential NULL pointer deref in XFS on failed mount.
If we fail to open the the log device buftarg, we can fall through to
error handling code that fails to check for a NULL log device buftarg
before calling xfs_free_buftarg().
This patch fixes the issue by checking mp->m_logdev_targp against NULL in
xfs_unmountfs_close() and doing the proper xfs_blkdev_put(logdev); and
xfs_blkdev_put(rtdev); on (!mp->m_rtdev_targp) in xfs_mount().
Eric Sandeen [Thu, 16 Aug 2007 06:24:43 +0000 (16:24 +1000)]
[XFS] choose single default logbuf count & size
Remove sizing of logbuf size & count based on physical memory; this was
never a very good gauge as it's looking at global memory, but deciding on
sizing per-filesystem; no account is made of the total number of
filesystems, for example.
For now just take the largest "default" case, as was set for machines with
>400MB - 8 x 32k buffers. This can always be tuned higher or lower with
mount options if necessary. Removes one more user of xfs_physmem.
Eric Sandeen [Thu, 16 Aug 2007 06:24:31 +0000 (16:24 +1000)]
[XFS] Remove m_nreadaheads
m_nreadaheads in the mount struct is never used; remove it and the various
macros assigned to it. Also remove a couple other unused macros in the
same areas.
The BMBT_*BITLEN are currently defined in a complicated way depending on
XFS_NATIVE_HOST. But if all the macros are expanded they (obviously)
expand to the same value for both cases.
This patch defines the macros in the most simple way and updates the
comment describing them to remove outdated bits.
[XFS] clean up xfs_bmbt_set_all/xfs_bmbt_disk_set_all
xfs_bmbt_set_all/xfs_bmbt_disk_set_all are identical to
xfs_bmbt_set_allf/xfs_bmbt_disk_set_allf except that the former take a
xfs_bmbt_irec_t and the latter take the individual extent fields as scalar
values.
This patch reimplements xfs_bmbt_set_all/xfs_bmbt_disk_set_all as trivial
wrappers around xfs_bmbt_set_allf/xfs_bmbt_disk_set_allf and cleans up the
variable naming in xfs_bmbt_set_allf/xfs_bmbt_disk_set_allf to have some
meaning instead of one char variable names.
[XFS] split ondisk vs incore versions of xfs_bmbt_rec_t
currently xfs_bmbt_rec_t is used both for ondisk extents as well as
host-endian ones. This patch adds a new xfs_bmbt_rec_host_t for the native
endian ones and cleans up the fallout. There have been various endianess
issues in the tracing / debug printf code that are fixed by this patch.
David Chinner [Mon, 14 May 2007 08:24:23 +0000 (18:24 +1000)]
[XFS] Barriers need to be dynamically checked and switched off
If the underlying block device suddenly stops supporting barriers, we need
to handle the -EOPNOTSUPP error in a sane manner rather than shutting
down the filesystem. If we get this error, clear the barrier flag, reissue
the I/O, and tell the world bad things are occurring.
Linus Torvalds [Tue, 9 Oct 2007 19:38:44 +0000 (12:38 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
[MIPS] Au1000: set the PCI controller IO base
[MIPS] Alchemy: Fix USB initialization.
[MIPS] IP32: Fix fatal typo in address computation.
Trond Myklebust [Tue, 9 Oct 2007 15:04:57 +0000 (11:04 -0400)]
NLM: Fix a memory leak in nlmsvc_testlock
The recent fix for a circular lock dependency unfortunately introduced a
potential memory leak in the event where the call to nlmsvc_lookup_host
fails for some reason.
Jeff Garzik [Tue, 9 Oct 2007 17:51:57 +0000 (13:51 -0400)]
sata_mv: correct S/G table limits
The recent mv_fill_sg() rewrite, to fix a data corruption problem
related to IOMMU virtual merging, forgot to account for the
potentially-increased size of the scatter/gather table after its run.
Additionally, the DMA boundary is reduced from 0xffffffff to 0xffff
to more closely match the needs of mv_fill_sg().
Signed-off-by: Jeff Garzik <jgarzik@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch fixes a wrong ifdef in the board setup code, leading to the GPIO
pin not being pulled high, and thus the USB switch not being powered at all.
This finishes the rename of CONFIG_USB_OHCI to CONFIG_USB_OHCI_HCD, which
started in 2005 (before 2.6.12-rc2), then probably because things were
working anyway for most people got forgotten.
[Ralf: Paolo's original patch didn't fix the module case, Florian's patch
only fixed MTX1 etc. so this is a combined patch plus some cleanups.]
Cc: Giuseppe Patanè <giuseppe.patane@tvblob.com> Cc: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John Crispin <blogic@openwrt.org> Signed-off-by: Florian Fainelli <florian.fainelli@telecomint.eu> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Correct Makefile rule for generating custom keymap
When building a custom keymap, after setting GENERATE_KEYMAP := 1 in
drivers/char/Makefile, the kernel build fails like this:
CC drivers/char/vt.o
make[2]: *** No rule to make target `drivers/char/%.map', needed by `drivers/char/defkeymap.c'. Stop.
make[1]: *** [drivers/char] Error 2
make: *** [drivers] Error 2
This was caused by commit af8b128719f5248e542036ea994610a29d0642a6, which
deleted a necessary colon from the Makefile rule that generates the keymap,
since that rule contains both a target and a target-pattern. The following
patch puts the colon back:
Yan Zheng [Mon, 8 Oct 2007 19:16:20 +0000 (12:16 -0700)]
AIO: fix cleanup in io_submit_one(...)
When IOCB_FLAG_RESFD flag is set and iocb->aio_resfd is incorrect,
statement 'goto out_put_req' is executed. At label 'out_put_req',
aio_put_req(..) is called, which requires 'req->ki_filp' set.
Signed-off-by: Yan Zheng<yanzheng@21cn.com> Cc: Zach Brown <zach.brown@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Yan Zheng [Mon, 8 Oct 2007 17:08:37 +0000 (10:08 -0700)]
fix page release issue in filemap_fault
find_lock_page increases page's usage count, we should decrease it
before return VM_FAULT_SIGBUS
Signed-off-by: Yan Zheng<yanzheng@21cn.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Peter Zijlstra [Mon, 8 Oct 2007 16:54:37 +0000 (18:54 +0200)]
mm: set_page_dirty_balance() vs ->page_mkwrite()
All the current page_mkwrite() implementations also set the page dirty. Which
results in the set_page_dirty_balance() call to _not_ call balance, because the
page is already found dirty.
This allows us to dirty a _lot_ of pages without ever hitting
balance_dirty_pages(). Not good (tm).
Force a balance call if ->page_mkwrite() was successful.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Brian Haley [Mon, 8 Oct 2007 07:12:05 +0000 (00:12 -0700)]
[IPv6]: Fix ICMPv6 redirect handling with target multicast address
When the ICMPv6 Target address is multicast, Linux processes the
redirect instead of dropping it. The problem is in this code in
ndisc_redirect_rcv():
if (ipv6_addr_equal(dest, target)) {
on_link = 1;
} else if (!(ipv6_addr_type(target) & IPV6_ADDR_LINKLOCAL)) {
ND_PRINTK2(KERN_WARNING
"ICMPv6 Redirect: target address is not
link-local.\n");
return;
}
This second check will succeed if the Target address is, for example,
FF02::1 because it has link-local scope. Instead, it should be checking
if it's a unicast link-local address, as stated in RFC 2461/4861 Section
8.1:
- The ICMP Target Address is either a link-local address (when
redirected to a router) or the same as the ICMP Destination
Address (when redirected to the on-link destination).
I know this doesn't explicitly say unicast link-local address, but it's
implied.
This bug is preventing Linux kernels from achieving IPv6 Logo Phase II
certification because of a recent error that was found in the TAHI test
suite - Neighbor Disovery suite test 206 (v6LC.2.3.6_G) had the
multicast address in the Destination field instead of Target field, so
we were passing the test. This won't be the case anymore.
The patch below fixes this problem, and also fixes ndisc_send_redirect()
to not send an invalid redirect with a multicast address in the Target
field. I re-ran the TAHI Neighbor Discovery section to make sure Linux
passes all 245 tests now.
Signed-off-by: Brian Haley <brian.haley@hp.com> Acked-by: David L Stevens <dlstevens@us.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Alexey Dobriyan [Mon, 8 Oct 2007 06:44:17 +0000 (23:44 -0700)]
[ROSE]: Fix rose.ko oops on unload
Commit a3d384029aa304f8f3f5355d35f0ae274454f7cd aka
"[AX.25]: Fix unchecked rose_add_loopback_neigh uses"
transformed rose_loopback_neigh var into statically allocated one.
However, on unload it will be kfree's which can't work.
Ilpo Järvinen [Mon, 8 Oct 2007 06:43:10 +0000 (23:43 -0700)]
[TCP]: Fix fastpath_cnt_hint when GSO skb is partially ACKed
When only GSO skb was partially ACKed, no hints are reset,
therefore fastpath_cnt_hint must be tweaked too or else it can
corrupt fackets_out. The corruption to occur, one must have
non-trivial ACK/SACK sequence, so this bug is not very often
that harmful. There's a fackets_out state reset in TCP because
fackets_out is known to be inaccurate and that fixes the issue
eventually anyway.
In case there was also at least one skb that got fully ACKed,
the fastpath_skb_hint is set to NULL which causes a recount for
fastpath_cnt_hint (the old value won't be accessed anymore),
thus it can safely be decremented without additional checking.
Reported by Cedric Le Goater <clg@fr.ibm.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
Dmitry Torokhov [Sun, 7 Oct 2007 16:22:21 +0000 (12:22 -0400)]
Driver core: fix SYSF_DEPRECATED breakage for nested classdevs
We should only reparent to a class former class devices that
form the base of class hierarchy. Nested devices should still
grow from their real parents.
Al Viro [Sun, 7 Oct 2007 07:24:36 +0000 (00:24 -0700)]
fix bogus reporting of signals by audit
Async signals should not be reported as sent by current in audit log. As
it is, we call audit_signal_info() too early in check_kill_permission().
Note that check_kill_permission() has that test already - it needs to know
if it should apply current-based permission checks. So the solution is to
move the call of audit_signal_info() between those.
Bogosity in question is easily reproduced - add a rule watching for e.g.
kill(2) from specific process (so that audit_signal_info() would not
short-circuit to nothing), say load_policy, watch the bogus OBJ_PID entry
in audit logs claiming that write(2) on selinuxfs file issued by
load_policy(8) had somehow managed to send a signal to syslogd...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Acked-by: Steve Grubb <sgrubb@redhat.com> Acked-by: Eric Paris <eparis@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Rafal Bilski [Sun, 7 Oct 2007 07:24:32 +0000 (00:24 -0700)]
Longhaul: add auto enabled "revid_errata" option
VIA C3 Ezra-T has RevisionID equal to 1, but it needs RevisionKey to be 0
or CPU will ignore new frequency and will continue to work at old
frequency. New "revid_errata" option will force RevisionKey to be set to
0, whatever RevisionID is.
Additionaly "Longhaul" will not silently ignore unsuccessful transition.
It will try to check if "revid_errata" or "disable_acpi_c3" options need to
be enabled for this processor/system.
Same for Longhaul ver. 2 support. It will be disabled if none of above
options will work.
Best case scenario (with patch apllied and v2 enabled):
longhaul: VIA C3 'Ezra' [C5C] CPU detected. Longhaul v2 supported.
longhaul: Using northbridge support.
longhaul: VRM 8.5
longhaul: Max VID=1.350 Min VID=1.050, 13 possible voltage scales
longhaul: f: 300000 kHz, index: 0, vid: 1050 mV
[...]
longhaul: Voltage scaling enabled.
Worst case scenario:
longhaul: VIA C3 'Ezra-T' [C5M] CPU detected. Powersaver supported.
longhaul: Using northbridge support.
longhaul: Using ACPI support.
longhaul: VRM 8.5
longhaul: Claims to support voltage scaling but min & max are both 1.250. Voltage scaling disabled
longhaul: Failed to set requested frequency!
longhaul: Enabling "Ignore Revision ID" option.
longhaul: Failed to set requested frequency!
longhaul: Disabling ACPI C3 support.
longhaul: Disabling "Ignore Revision ID" option.
longhaul: Failed to set requested frequency!
longhaul: Enabling "Ignore Revision ID" option.
[akpm@linux-foundation.org: coding-style cleanups] Signed-off-by: Rafal Bilski <rafalbilski@interia.pl> Signed-off-by: Dave Jones <davej@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Anton Blanchard [Sun, 7 Oct 2007 07:24:31 +0000 (00:24 -0700)]
Fix timer_stats printout of events/sec
When using /proc/timer_stats on ppc64 I noticed the events/sec field wasnt
accurate. Sometimes the integer part was incorrect due to rounding (we
werent taking the fractional seconds into consideration).
The fraction part is also wrong, we need to pad the printf statement and
take the bottom three digits of 1000 times the value.
Signed-off-by: Anton Blanchard <anton@samba.org> Acked-by: Ingo Molnar <mingo@elte.hu> Cc: <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Sun, 7 Oct 2007 23:17:38 +0000 (16:17 -0700)]
Don't do load-average calculations at even 5-second intervals
It turns out that there are a few other five-second timers in the
kernel, and if the timers get in sync, the load-average can get
artificially inflated by events that just happen to coincide.
So just offset the load average calculation it by a timer tick.
Noticed by Anders Boström, for whom the coincidence started triggering
on one of his machines with the JBD jiffies rounding code (JBD is one of
the subsystems that also end up using a 5-second timer by default).
Tested-by: Anders Boström <anders@bostrom.dyndns.org> Cc: Chuck Ebbert <cebbert@redhat.com> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Sun, 7 Oct 2007 23:02:55 +0000 (16:02 -0700)]
VT_WAITACTIVE: Avoid returning EINTR when not necessary
We should generally prefer to return ERESTARTNOHAND rather than EINTR,
so that processes with unhandled signals that get ignored don't return
EINTR.
This can help with X startup issues:
Fatal server error:
xf86OpenConsole: VT_WAITACTIVE failed: Interrupted system call
although the real fix is having the X server always retry EINTR
regardless (since EINTR does happen for signals that have handlers
installed). Keithp has a patch for that.
Regardless, ERESTARTNOHAND is the correct thing to use.
Fix non-terminated PCI match table in PowerMac IDE
The PCI device table in the powermac IDE driver isn't properly
terminated. Depending on how your kernel is linked and other random
factors, you can end up with this driver matched against any other PCI
device in your system, possibly crashing at boot.
Thanks to Heikki for tracking this down with me, the bug have been there
for some time, though it rarely hurts due to luck. In this case, the
switch from .22 to .23-rc9 is causing it to show up due to differences
in the resulting layout of .data I suppose.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <pmac@au1.ibm.com> Cc: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl> Cc: Heikki Lindholm <holindho@cs.helsinki.fi> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When pinning and unpinning pagetables, we must protect them against
being used by other CPUs, lest they see the pagetable in an
intermediate read-only-but-not-pinned state.
When using split pte locks, doing this properly would require taking
all the pte locks for the pagetable while pinning, but this may overflow
the PREEMPT_BITS part of the preempt counter if the process has mapped
more than about 512M of memory.
However, failing to take the pte locks causes write-protect faults when
the pageout code is trying to clear the Access bit on a pte which is part
of a freshy created and still being pinned process after fork.
This is a short-term fix until the problem is solved properly.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Hugh Dickins <hugh@veritas.com> Cc: David Rientjes <rientjes@google.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andi Kleen <ak@suse.de> Cc: Keir Fraser <keir@xensource.com> Cc: Jan Beulich <jbeulich@novell.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Fri, 5 Oct 2007 21:09:10 +0000 (14:09 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm
* master.kernel.org:/home/rmk/linux-2.6-arm:
[ARM] 4598/2: OSIRIS: Ensure we do not get nRSTOUT during suspend
[ARM] 4597/2: OSIRIS: ensure CPLD0 is preserved after suspend
Ben Dooks [Thu, 4 Oct 2007 22:16:42 +0000 (23:16 +0100)]
[ARM] 4597/2: OSIRIS: ensure CPLD0 is preserved after suspend
Ensure that CPLD is restored to the original state
on resume, and that before going into suspend we
select the NAND bank we booted from for restarting.
Signed-off-by: Ben Dooks <ben-linux@fluff.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The 8169/8110SC currently announces itself as:
[...]
eth0: RTL8169sc/8110sc at 0x........, ..:..:..:..:..:.., XID 18000000 IRQ ..
^^^^^^^^
It uses RTL_GIGA_MAC_VER_05 and this part of the changeset can cut
its performance by a factor of 2~2.5 as reported by Timo.
(the driver includes code just before the hunk to write the ChipCmd
register when mac_version == RTL_GIGA_MAC_VER_0[1-4])
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com> Cc: Timo Jantunen <jeti@welho.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Linus Torvalds [Fri, 5 Oct 2007 15:08:08 +0000 (08:08 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
[SPARC64]: Fix 'niu' complex IRQ probing.
[SPARC64]: check fork_idle() error
[SPARC64]: Temporary workaround for PCI-E slot on T1000.
[SPARC64]: VIO device addition log message level is too high.
[SPARC64]: Fix domain-services port probing.
[SPARC64]: Don't use in/local regs for ldx/stx data in N1 memcpy.
Serge Belyshev [Thu, 4 Oct 2007 21:10:04 +0000 (23:10 +0200)]
Remove unnecessary cast in prefetch()
It is ok to call prefetch() function with NULL argument, as specifically
commented in include/linux/prefetch.h. But in standard C, it is invalid
to dereference NULL pointer (see C99 standard 6.5.3.2 paragraph 4 and
note #84).
prefetch() has a memory reference for its argument.
Newer gcc versions (4.3 and above) will use that to conclude that "x"
argument is non-null and thus wreaking havok everywhere prefetch() was
inlined.
Fixed by removing cast and changing asm constraint.
[ It seems in theory gcc 4.2 could miscompile this too; although no
cases known. In 2.6.24 we should probably switch to
__builtin_prefetch() instead, but this is a simpler fix for now.
-- AK ]
Giuseppe Sacco [Thu, 4 Oct 2007 21:09:12 +0000 (23:09 +0200)]
[MIPS] IP32: Enable PCI bridges
Fixe MACE PCI addressing by adding the bus number parameter.
Remove check of the used slot since every slot should be valid.
Converted mkaddr from #define to inline function.
Signed-off-by: Giuseppe Sacco <eppesuig@debian.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Hugh Dickins [Thu, 4 Oct 2007 15:56:06 +0000 (16:56 +0100)]
Fix sys_remap_file_pages BUG at highmem.c:15!
Gurudas Pai reports kernel BUG at arch/i386/mm/highmem.c:15! below
sys_remap_file_pages, while running Oracle database test on x86 in 6GB
RAM: kunmap thinks we're in_interrupt because the preempt count has
wrapped.
That's because __do_fault expected to unmap page_table, but one of its
two callers do_nonlinear_fault already unmapped it: let do_linear_fault
unmap it first too, and then there's no need to pass the page_table arg
down.
Why have we been so slow to notice this? Probably through forgetting
that the mapping_cap_account_dirty test means that sys_remap_file_pages
nowadays only goes the full nonlinear vma route on a few memory-backed
filesystems like ramfs, tmpfs and hugetlbfs.
[ It also depends on CONFIG_HIGHPTE, so it becomes even harder to
trigger in practice. Many who have need of large memory have probably
migrated to x86-64..
FUJITA Tomonori [Wed, 3 Oct 2007 00:00:58 +0000 (09:00 +0900)]
[SCSI] megaraid_old: fix READ_CAPACITY
The bulk transfer mode got eleminated by 3f6270ef76f2ce5c134615a470685d6c2a66c07e. Unfortunately, this mode is
required for READ_CAPACITY commands on certain cards, so put it back
again. This fixes a boot failure regression reported by Burton
Windle.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
David S. Miller [Thu, 4 Oct 2007 04:37:57 +0000 (21:37 -0700)]
[SPARC64]: Temporary workaround for PCI-E slot on T1000.
The PCI-E slot on T1000 connects directly to the Fire PCI chip with no
intervening bridges visible in the OBP tree.
Unfortunately the bus numbering of the device in that slot is
different (2) from the PCI host controller (0), and thus the
pci_bus_{read,write}_config_*() calls don't work out.
Complicating things further the Fire PCI controller has no config
space it responds to either.
For now treat this case specially so that devices in the slot work.
Longer term we need to perhaps cons up a dummy bridge between the Fire
and the PCI-E slot so that the bus hierarchy is complete inside of the
kernel and thus the bus numbering all works out right.
Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-beck.git / log