The delayed reference mechanism is responsible for all updates to the
extent allocation trees, including those updates created while processing
the delayed references.
This commit tries to limit the amount of work that gets created during
the final run of delayed refs before a commit. It avoids cowing new blocks
unless it is required to finish the commit, and so it avoids new allocations
that were not really required.
The goal is to avoid infinite loops where we are always making more work
on the final run of delayed refs. Over the long term we'll make a
special log for the last delayed ref updates as well.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This reads in blocks in the checksum btree before starting the
transaction in btrfs_finish_ordered_io. It makes it much more likely
we'll be able to do operations inside the transaction without
needing any btree reads, which limits transaction latencies overall.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_mark_buffer dirty would set dirty bits in the extent_io tree
for the buffers it was dirtying. This may require a kmalloc and it
was not atomic. So, anyone who called btrfs_mark_buffer_dirty had to
set any btree locks they were holding to blocking first.
This commit changes dirty tracking for extent buffers to just use a flag
in the extent buffer. Now that we have one and only one extent buffer
per page, this can be safely done without losing dirty bits along the way.
This also introduces a path->leave_spinning flag that callers of
btrfs_search_slot can use to indicate they will properly deal with a
path returned where all the locks are spinning instead of blocking.
Many of the btree search callers now expect spinning paths,
resulting in better btree concurrency overall.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Commits are fairly expensive, and so btrfs has code to sit around for a while
during the commit and let new writers come in.
But, while we're sitting there, new delayed refs might be added, and those
can be expensive to process as well. Unless the transaction is very very
young, it makes sense to go ahead and let the commit finish without hanging
around.
The commit grow loop isn't as important as it used to be, the fsync logging
code handles most performance critical syncs now.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This reduces contention on the extent buffer spin locks by testing for a
blocking lock before trying to take the spinlock.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The fs/btrfs/inode.c code to run delayed allocation during writout
needed some stack usage optimization. This is the first pass, it does
the check for compression earlier on, which allows us to do the common
(no compression) case higher up in the call chain.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
To avoid deadlocks and reduce latencies during some critical operations, some
transaction writers are allowed to jump into the running transaction and make
it run a little longer, while others sit around and wait for the commit to
finish.
This is a bit unfair, especially when the callers that jump in do a bunch
of IO that makes all the others procs on the box wait. This commit
reduces the stalls this produces by pre-reading file extent pointers
during btrfs_finish_ordered_io before the transaction is joined.
It also tunes the drop_snapshot code to politely wait for transactions
that have started writing out their delayed refs to finish. This avoids
new delayed refs being flooded into the queue while we're trying to
close off the transaction.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The delayed reference queue maintains pending operations that need to
be done to the extent allocation tree. These are processed by
finding records in the tree that are not currently being processed one at
a time.
This is slow because it uses lots of time searching through the rbtree
and because it creates lock contention on the extent allocation tree
when lots of different procs are running delayed refs at the same time.
This commit changes things to grab a cluster of refs for processing,
using a cursor into the rbtree as the starting point of the next search.
This way we walk smoothly through the rbtree.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When extents are freed, it is likely that we've removed the last
delayed reference update for the extent. This checks the delayed
ref tree when things are freed, and if no ref updates area left it
immediately processes the delayed ref.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Many of the tree balancing functions follow the same pattern.
1) cow a block
2) do something to the result
This commit breaks them up into two functions so the variables and
code required for part two don't suck down stack during part one.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The extent allocation tree maintains a reference count and full
back reference information for every extent allocated in the
filesystem. For subvolume and snapshot trees, every time
a block goes through COW, the new copy of the block adds a reference
on every block it points to.
If a btree node points to 150 leaves, then the COW code needs to go
and add backrefs on 150 different extents, which might be spread all
over the extent allocation tree.
These updates currently happen during btrfs_cow_block, and most COWs
happen during btrfs_search_slot. btrfs_search_slot has locks held
on both the parent and the node we are COWing, and so we really want
to avoid IO during the COW if we can.
This commit adds an rbtree of pending reference count updates and extent
allocations. The tree is ordered by byte number of the extent and byte number
of the parent for the back reference. The tree allows us to:
1) Modify back references in something close to disk order, reducing seeks
2) Significantly reduce the number of modifications made as block pointers
are balanced around
3) Do all of the extent insertion and back reference modifications outside
of the performance critical btrfs_search_slot code.
#3 has the added benefit of greatly reducing the btrfs stack footprint.
The extent allocation tree modifications are done without the deep
(and somewhat recursive) call chains used in the past.
These delayed back reference updates must be done before the transaction
commits, and so the rbtree is tied to the transaction. Throttling is
implemented to help keep the queue of backrefs at a reasonable size.
Since there was a similar mechanism in place for the extent tree
extents, that is removed and replaced by the delayed reference tree.
Yan Zheng <yan.zheng@oracle.com> helped review and fixup this code.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
In order to avoid doing expensive extent management with tree locks held,
btrfs_search_slot will preallocate tree blocks for use by COW without
any tree locks held.
A later commit moves all of the extent allocation work for COW into
a delayed update mechanism, and this preallocation will no longer be
required.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
With a sufficiently new compiler and binutils, code which wasn't
previously generating .eh_frame sections has begun to. Certain
architectures (powerpc, in this case) may generate unexpected relocation
formats in response to this, preventing modules from loading.
While the new relocation types should probably be handled, revert to the
previous behaviour with regards to generation of .eh_frame sections.
(This was reported against Fedora, which appears to be the only distro
doing any building against gcc-4.4 at present: RH bz#486545.)
Signed-off-by: Kyle McMartin <kyle@redhat.com>
Acked-by: Roland McGrath <roland@redhat.com>
Cc: Alexandre Oliva <aoliva@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Revert the change to the orphan dates of Windows 95, DOS, compression.
Add a new orphan date for OS/2.
Signed-off-by: Jody McIntyre <scjody@sun.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (32 commits)
ucc_geth: Fix oops when using fixed-link support
dm9000: locking bugfix
net: update dnet.c for bus_id removal
dnet: DNET should depend on HAS_IOMEM
dca: add missing copyright/license headers
nl80211: Check that function pointer != NULL before using it
sungem: missing net_device_ops
be2net: fix to restore vlan ids into BE2 during a IF DOWN->UP cycle
be2net: replenish when posting to rx-queue is starved in out of mem conditions
bas_gigaset: correctly allocate USB interrupt transfer buffer
smsc911x: reset last known duplex and carrier on open
sh_eth: Fix mistake of the address of SH7763
sh_eth: Change handling of IRQ
netns: oops in ip[6]_frag_reasm incrementing stats
net: kfree(napi->skb) => kfree_skb
net: fix sctp breakage
ipv6: fix display of local and remote sit endpoints
net: Document /proc/sys/net/core/netdev_budget
tulip: fix crash on iface up with shirq debug
virtio_net: Make virtio_net support carrier detection
...
This patch fixes bug #12208:
Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=12208
Subject : uml is very slow on 2.6.28 host
This turned out to be not a scheduler regression, but an already
existing problem in ptrace being triggered by subtle scheduler
changes.
The problem is this:
- task A is ptracing task B
- task B stops on a trace event
- task A is woken up and preempts task B
- task A calls ptrace on task B, which does ptrace_check_attach()
- this calls wait_task_inactive(), which sees that task B is still on the runq
- task A goes to sleep for a jiffy
- ...
Since UML does lots of the above sequences, those jiffies quickly add
up to make it slow as hell.
This patch solves this by not rescheduling in read_unlock() after
ptrace_stop() has woken up the tracer.
Thanks to Oleg Nesterov and Ingo Molnar for the feedback.
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
CC: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Grant picked up the wrong version of "Respect _PAGE_COHERENT on classic
ppc32 SW" (commit a4bd6a93c3)
It was missing the code to actually deal with the fixup of
_PAGE_COHERENT based on the CPU feature.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
commit b1c4a9dddf ("ucc_geth: Change
uec phy id to the same format as gianfar's") introduced a regression
in the ucc_geth driver that causes this oops when fixed-link is used:
Unable to handle kernel paging request for data at address 0x00000000
Faulting instruction address: 0xc0151270
Oops: Kernel access of bad area, sig: 11 [#1]
TMCUTU
NIP: c0151270 LR: c0151270 CTR: c0017760
REGS: cf81fa60 TRAP: 0300 Not tainted (2.6.29-rc8)
MSR: 00009032 <EE,ME,IR,DR> CR: 24024042 XER: 20000000
DAR: 00000000, DSISR: 20000000
TASK = cf81cba0[1] 'swapper' THREAD: cf81e000
GPR00: c0151270 cf81fb10 cf81cba0 00000000 c0272e20 c025f354 00001e80
cf86b08c
GPR08: d1068200 cffffb74 06000000 d106c200 42024042 10085148 0fffd000
0ffc81a0
GPR16: 00000001 00000001 00000000 007ffeb0 00000000 0000c000 cf83f36c
cf83f000
GPR24: 00000030 cf83f360 cf81fb20 00000000 d106c200 20000000 00001e80
cf83f360
NIP [c0151270] ucc_geth_open+0x330/0x1efc
LR [c0151270] ucc_geth_open+0x330/0x1efc
Call Trace:
[cf81fb10] [c0151270] ucc_geth_open+0x330/0x1efc (unreliable)
[cf81fba0] [c0187638] dev_open+0xbc/0x12c
[cf81fbc0] [c0187e38] dev_change_flags+0x8c/0x1b0
This patch fixes the issue by removing offending (and somewhat
duplicate) code from init_phy() routine, and changes _probe()
function to use uec_mdio_bus_name().
Also, since we fully construct phy_bus_id in the _probe() routine,
we no longer need ->phy_address and ->mdio_bus fields in
ucc_geth_info structure.
I wish the patch would be a bit shorter, but it seems like the only
way to fix the issue in a sane way. Luckily, the patch has been
tested with real PHYs and fixed-link, so no further regressions
expected.
Reported-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Tested-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes a locking bug in the dm9000 driver. It calls
request_irq() without setting IRQF_DISABLED ... which is
correct for handlers that support IRQ sharing, since that
behavior is not guaranteed for shared IRQs. However, its
IRQ handler then wrongly assumes that IRQs are blocked.
So the fix just uses the right spinlock primitives in the
IRQ handler.
NOTE: this is a classic example of the type of bug which
lockdep currently masks by forcibly setting IRQF_DISABLED
on IRQ handlers that did not request that flag.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
* 'fix-includes' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
m68k: merge the non-MMU and MMU versions of siginfo.h
m68k: use the MMU version of unistd.h for all m68k platforms
m68k: merge the non-MMU and MMU versions of signal.h
m68k: merge the non-MMU and MMU versions of ptrace.h
m68k: use MMU version of setup.h for both MMU and non-MMU
m68k: merge the non-MMU and MMU versions of sigcontext.h
m68k: merge the non-MMU and MMU versions of swab.h
m68k: merge the non-MMU and MMU versions of param.h
Update all previous incarnations of my email address to the correct one.
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If ecryptfs_encrypted_view or ecryptfs_xattr_metadata were being
specified as mount options, a NULL pointer dereference of crypt_stat
was possible during lookup.
This patch moves the crypt_stat assignment into
ecryptfs_lookup_and_interpose_lower(), ensuring that crypt_stat
will not be NULL before we attempt to dereference it.
Thanks to Dan Carpenter and his static analysis tool, smatch, for
finding this bug.
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Acked-by: Dustin Kirkland <kirkland@canonical.com>
Cc: Dan Carpenter <error27@gmail.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When allocating the memory used to store the eCryptfs header contents, a
single, zeroed page was being allocated with get_zeroed_page().
However, the size of an eCryptfs header is either PAGE_CACHE_SIZE or
ECRYPTFS_MINIMUM_HEADER_EXTENT_SIZE (8192), whichever is larger, and is
stored in the file's private_data->crypt_stat->num_header_bytes_at_front
field.
ecryptfs_write_metadata_to_contents() was using
num_header_bytes_at_front to decide how many bytes should be written to
the lower filesystem for the file header. Unfortunately, at least 8K
was being written from the page, despite the chance of the single,
zeroed page being smaller than 8K. This resulted in random areas of
kernel memory being written between the 0x1000 and 0x1FFF bytes offsets
in the eCryptfs file headers if PAGE_SIZE was 4K.
This patch allocates a variable number of pages, calculated with
num_header_bytes_at_front, and passes the number of allocated pages
along to ecryptfs_write_metadata_to_contents().
Thanks to Florian Streibelt for reporting the data leak and working with
me to find the problem. 2.6.28 is the only kernel release with this
vulnerability. Corresponds to CVE-2009-0787
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Acked-by: Dustin Kirkland <kirkland@canonical.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Eugene Teo <eugeneteo@kernel.sg>
Cc: Greg KH <greg@kroah.com>
Cc: dann frazier <dannf@dannf.org>
Cc: Serge E. Hallyn <serue@us.ibm.com>
Cc: Florian Streibelt <florian@f-streibelt.de>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This fixes a regression introduced when we switched to using the core
pci_set_power_state(). The chip seems to need the state to be written
over and over again until it sticks, so we do that.
Note that the code is a bit blunt, without timeout, etc... but that's
pretty much because I put back in there the code exactly as it used to
be before the regression. I still add a call to pci_set_power_state()
at the end so that ACPI gets called appropriately on x86.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Tested-by: Raymond Wooninck <tittiatcoke@gmail.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In two dca files copyright and license headers are missing.
This patch adds them there.
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
NL80211_CMD_GET_MESH_PARAMS and NL80211_CMD_SET_MESH_PARAMS handlers
did not verify whether a function pointer is NULL (not supported by
the driver) before trying to call the function. The former nl80211
command is available for unprivileged users, too, so this can
potentially allow normal users to kill networking (or worse..) if
mac80211 is built without CONFIG_MAC80211_MESH=y.
Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Sungem driver only got partially converted to net_device_ops.
Since this could cause bugs, please push this to 2.6.29
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a patch to reconfigure vlan-ids during an i/f down/up cycle
Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a patch to replenish the rx-queue when it is in a starved
state (due to out-of-mem conditions)
Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The libaio test harness turned up a problem whereby lookup_ioctx on a
bogus io context was returning the 1 valid io context from the list
(harness/cases/3.p).
Because of that, an extra put_iocontext was done, and when the process
exited, it hit a BUG_ON in the put_iocontext macro called from exit_aio
(since we expect a users count of 1 and instead get 0).
The problem was introduced by "aio: make the lookup_ioctx() lockless"
(commit abf137dd77).
Thanks to Zach for pointing out that hlist_for_each_entry_rcu will not
return with a NULL tpos at the end of the loop, even if the entry was
not found.
Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Acked-by: Zach Brown <zach.brown@oracle.com>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Remove a source of fput() call from inside IRQ context. Myself, like Eric,
wasn't able to reproduce an fput() call from IRQ context, but Jeff said he was
able to, with the attached test program. Independently from this, the bug is
conceptually there, so we might be better off fixing it. This patch adds an
optimization similar to the one we already do on ->ki_filp, on ->ki_eventfd.
Playing with ->f_count directly is not pretty in general, but the alternative
here would be to add a brand new delayed fput() infrastructure, that I'm not
sure is worth it.
Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Cc: Zach Brown <zach.brown@oracle.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Sam Ravnborg says:
"We have several architectures that plays strange games with $(CC) and
$(CROSS_COMPILE).
So we need to postpone any use of $(call cc-option..) until we have
included the arch specific Makefile so we try with the correct $(CC)
version."
Requested-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
[S390] make page table upgrade work again
[S390] make page table walking more robust
[S390] Dont check for pfn_valid() in uaccess_pt.c
[S390] ftrace/mcount: fix kernel stack backchain
[S390] topology: define SD_MC_INIT to fix performance regression
[S390] __div64_31 broken for CONFIG_MARCH_G5
* git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
Btrfs: Clear space_info full when adding new devices
Btrfs: Fix locking around adding new space_info
Nick Piggin noticed this (very unlikely) race between setting a page
dirty and creating the buffers for it - we need to hold the mapping
private_lock until we've set the page dirty bit in order to make sure
that create_empty_buffers() might not build up a set of buffers without
the dirty bits set when the page is dirty.
I doubt anybody has ever hit this race (and it didn't solve the issue
Nick was looking at), but as Nick says: "Still, it does appear to solve
a real race, which we should close."
Acked-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This makes sure that gcc doesn't try to optimize away wrapping
arithmetic, which the kernel occasionally uses for overflow testing, ie
things like
if (ptr + offset < ptr)
which technically is undefined for non-unsigned types. See
http://bugzilla.kernel.org/show_bug.cgi?id=12597
for details.
Not all versions of gcc support it, so we need to make it conditional
(it looks like it was introduced in gcc-3.4).
Reminded-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When you compile kernel on Sparc64 with heap memory checking and type
"cat /proc/iomem", you get a crash, because pointers in struct
resource are uninitialized.
Most code fills struct resource with zeros, so I assume that it is
responsibility of the caller of request_resource to initialized it,
not the responsibility of request_resource functuion.
After 2.6.29 is out, there could be a check for uninitialized fields
added to request_resource to avoid crashes like this.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Otherwise it might interrupt switch_to() midstream and use
half-cooked register window state.
Reported-by: Chris Torek <chris.torek@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Every USB transfer buffer has to be allocated individually by kmalloc.
Impact: bugfix, no functional change
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Tested-by: Kolja Waschk <kawk@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
smsc911x_phy_adjust_link is called periodically by the phy layer (as
it's run in polling mode), and it only updates the hardware when it sees
a change in duplex or carrier. This patch clears the last known values
every time the interface is brought up, instead of only when the module
is loaded.
Without this patch the adjust_link function never updates the hardware
after an ifconfig down; ifconfig up. On a full duplex link this causes
the tx error counter to increment, even though packets are correctly
transmitted, as the default MAC_CR register setting is for half duplex.
The tx errors are "no carrier" errors, which should be ignored in
full-duplex mode. When MAC_CR is set to "full duplex" mode they are
correctly ignored by the hardware.
Note that even with this patch the tx error counter can increment if
packets are transmitted between "ifconfig up" and the first phy poll
interval. An improved solution would use the phy interrupt with phylib,
but I haven't managed to make this work 100% robustly yet.
Signed-off-by: Steve Glendinning <steve.glendinning@smsc.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Address of SH_TSU_ADDR and ARSTR of SH7763 was wrong.
This revise it.
Signed-off-by: Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
