When the SMP kernel decides to crash_kexec() the local APICs may have
pending interrupts in their vector tables.
The setup routine for the local APIC has a deficient mechanism for
clearing these interrupts, it only handles interrupts that has already
been dispatched to the local core for servicing (the ISR register) safely,
it doesn't consider lower prioritized queued interrupts stored in the IRR
register.
If you have more than one pending interrupt within the same 32 bit word in
the LAPIC vector table registers you may find yourself entering the IO
APIC setup with pending interrupts left in the LAPIC. This is a situation
for wich the IO APIC setup is not prepared. Depending of what/which
interrupt vector/vectors are stuck in the APIC tables your system may show
various degrees of malfunctioning. That was the reason why the
check_timer() failed in our system, the timer interrupts was blocked by
pending interrupts from the old kernel when routed trough the IO APIC.
Additional comment from Jiri Bohac:
==============
If this should go into stable release,
I'd add some kind of limit on the number of iterations, just to be safe from
hard to debug lock-ups:
with MAX_LOOPS something like 1E9 this would leave plenty of time for the
pending IRQs to be cleared and would and still cause at most a second of delay
if the loop were to lock-up for whatever reason.
[trenn@suse.de:
V2: Use tsc if avail to bail out after 1 sec due to possible virtual
apic_read calls which may take rather long (suggested by: Avi Kivity
<avi@redhat.com>) If no tsc is available bail out quickly after
cpu_khz, if we broke out too early and still have irqs pending (which
should never happen?) we still get a WARN_ON...
V3: - Fixed indentation -> checkpatch clean
- max_loops must be signed
V4: - Fix typo, mixed up tsc and ntsc in first rdtscll() call
V5: Adjust WARN_ON() condition to also catch error in cpu_has_tsc case]
Cc: <jbohac@novell.com> Cc: Yinghai Lu <yinghai@kernel.org> Cc: Kerstin Jonsson <kerstin.jonsson@ericsson.com> Cc: Avi Kivity <avi@redhat.com> Cc: Suresh Siddha <suresh.b.siddha@intel.com> Tested-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Thomas Renninger <trenn@suse.de>
LKML-Reference: <201005241913.o4OJDGWM010865@imap1.linux-foundation.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Thomas Renninger <trenn@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add ftdi product ID for Lenz LI-USB, a model train interface. This
was NOT tested against 2.6.35, but a similar patch was tested with the
CentOS 2.6.18-194.11.1.el5 kernel. It wasn't clear to me what
ordering is being used in ftdi_sio.c, so I inserted the ID after another
model train entry(SPROG_II).
The code to increment the TRB pointer has a slight ambiguity that could
lead to a bug on different compilers. The ANSI C specification does not
specify the precedence of the assignment operator over the postfix
operator. gcc 4.4 produced the correct code (increment the pointer and
assign the value), but a MIPS compiler that one of John's clients used
assigned the old (unincremented) value.
Remove the unnecessary assignment to make all compilers produce the
correct assembly.
Signed-off-by: John Youn <johnyoun@synopsys.com> Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If we can't read the firmware for a device from the disk, and yet the
device already has a valid firmware image in it, we don't want to
replace the firmware with something invalid. So check the version
number to be less than the current one to verify this is the correct
thing to do.
Reported-by: Chris Beauchamp <chris@chillibean.tv> Tested-by: Chris Beauchamp <chris@chillibean.tv> Cc: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add support for the Zeagle N2iTiON3 dive computer interface. Since
Zeagle devices are actually manufactured by Seiko, this patch will
support other Seiko based models as well.
>Try the navman driver instead. You can either add the device id to the
> driver and rebuild it, or do this before you plug the device in:
> modprobe navman
> echo -n "0x0df7 0x0900" > /sys/bus/usb-serial/drivers/navman/new_id
>
> and then plug your device in and see if that works.
I can confirm that the navman driver works with the right device IDs on
my i-gotU GT-600, which has the same device IDs. Attached is a patch
adding the IDs.
From: Ross Burton <ross@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Userspace controls the amount of memory to be allocate, so it can
get the ioctl to allocate more memory than the kernel uses, and get
access to kernel stack. This can only be done for processes authenticated
to the X server for DRI access, and if the user has DRI access.
Fix is to just memset the data to 0 if the user doesn't copy into
it in the first place.
net/compat/wext: send different messages to compat tasks
we had a race condition when setting and then
restoring frag_list. Eric attempted to fix it,
but the fix created even worse problems.
However, the original motivation I had when I
added the code that turned out to be racy is
no longer clear to me, since we only copy up
to skb->len to userspace, which doesn't include
the frag_list length. As a result, not doing
any frag_list clearing and restoring avoids
the race condition, while not introducing any
other problems.
Additionally, while preparing this patch I found
that since none of the remaining netlink code is
really aware of the frag_list, we need to use the
original skb's information for packet information
and credentials. This fixes, for example, the
group information received by compat tasks.
Cc: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
BugLink: https://bugs.launchpad.net/bugs/619439
This ThinkPad model needs External Amplifier muted for audible playback,
so set the inv_eapd quirk for it.
Reported-and-tested-by: Dennis Bell <dennis.bell@parkerg.co.uk> Signed-off-by: Daniel T Chen <crimsun@ubuntu.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It doesn't like pattern and explicit rules to be on the same line,
and it seems to be more picky when matching file (or really directory)
names with different numbers of trailing slashes.
Signed-off-by: Jan Beulich <jbeulich@novell.com> Acked-by: Sam Ravnborg <sam@ravnborg.org>
Andrew Benton <b3nton@gmail.com> Signed-off-by: Michal Marek <mmarek@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Newer Intel processors identifying themselves as model 30 are not recognized by
oprofile.
<cpuinfo snippet>
model : 30
model name : Intel(R) Xeon(R) CPU X3470 @ 2.93GHz
</cpuinfo snippet>
Running oprofile on these machines gives the following:
+ opcontrol --init
+ opcontrol --list-events
oprofile: available events for CPU type "Intel Architectural Perfmon"
See Intel 64 and IA-32 Architectures Software Developer's Manual
Volume 3B (Document 253669) Chapter 18 for architectural perfmon events
This is a limited set of fallback events because oprofile doesn't know your CPU
CPU_CLK_UNHALTED: (counter: all)
Clock cycles when not halted (min count: 6000)
INST_RETIRED: (counter: all)
number of instructions retired (min count: 6000)
LLC_MISSES: (counter: all)
Last level cache demand requests from this core that missed the LLC
(min count: 6000)
Unit masks (default 0x41)
----------
0x41: No unit mask
LLC_REFS: (counter: all)
Last level cache demand requests from this core (min count: 6000)
Unit masks (default 0x4f)
----------
0x4f: No unit mask
BR_MISS_PRED_RETIRED: (counter: all)
number of mispredicted branches retired (precise) (min count: 500)
+ opcontrol --shutdown
Back when the patch was submitted for "Add Xeon 7500 series support to
oprofile", Robert Richter had asked for a followon patch that
converted all the CPU ID values to hex.
I have done that here for the "i386/core_i7" and "i386/atom" class
processors in the ppro_init() function and also added some comments on
where to find documentation on the Intel processors.
Signed-off-by: John L. Villalovos <john.l.villalovos@intel.com> Signed-off-by: Robert Richter <robert.richter@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
There are duplicate macro definitions of in_range() in mballoc.h and
balloc.c. This consolidates these two definitions into ext4.h, and
changes extents.c to use in_range() as well.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger@sun.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
NR_IRQS may be as low as 16, causing a (harmless?) buffer overflow in
pcmcia_setup_isa_irq():
static u8 pcmcia_used_irq[NR_IRQS];
...
if ((try < 32) && pcmcia_used_irq[irq])
continue;
This is read-only, so if this address would be non-zero, it would just
mean we would not attempt an IRQ >= NR_IRQS -- which would fail anyway!
And as request_irq() fails for an irq >= NR_IRQS, the setting code path:
pcmcia_used_irq[irq]++;
is never reached as well.
Reported-by: Christoph Fritz <chf.fritz@googlemail.com> Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net> Signed-off-by: Christoph Fritz <chf.fritz@googlemail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix "system goes unresponsive under memory pressure and lots of
dirty/writeback pages" bug.
http://lkml.org/lkml/2010/4/4/86
In the above thread, Andreas Mohr described that
Invoking any command locked up for minutes (note that I'm
talking about attempted additional I/O to the _other_,
_unaffected_ main system HDD - such as loading some shell
binaries -, NOT the external SSD18M!!).
This happens when the two conditions are both meet:
- under memory pressure
- writing heavily to a slow device
OOM also happens in Andreas' system. The OOM trace shows that 3 processes
are stuck in wait_on_page_writeback() in the direct reclaim path. One in
do_fork() and the other two in unix_stream_sendmsg(). They are blocked on
this condition:
(sc->order && priority < DEF_PRIORITY - 2)
which was introduced in commit 78dc583d (vmscan: low order lumpy reclaim
also should use PAGEOUT_IO_SYNC) one year ago. That condition may be too
permissive. In Andreas' case, 512MB/1024 = 512KB. If the direct reclaim
for the order-1 fork() allocation runs into a range of 512KB
hard-to-reclaim LRU pages, it will be stalled.
It's a severe problem in three ways.
Firstly, it can easily happen in daily desktop usage. vmscan priority can
easily go below (DEF_PRIORITY - 2) on _local_ memory pressure. Even if
the system has 50% globally reclaimable pages, it still has good
opportunity to have 0.1% sized hard-to-reclaim ranges. For example, a
simple dd can easily create a big range (up to 20%) of dirty pages in the
LRU lists. And order-1 to order-3 allocations are more than common with
SLUB. Try "grep -v '1 :' /proc/slabinfo" to get the list of high order
slab caches. For example, the order-1 radix_tree_node slab cache may
stall applications at swap-in time; the order-3 inode cache on most
filesystems may stall applications when trying to read some file; the
order-2 proc_inode_cache may stall applications when trying to open a
/proc file.
Secondly, once triggered, it will stall unrelated processes (not doing IO
at all) in the system. This "one slow USB device stalls the whole system"
avalanching effect is very bad.
Thirdly, once stalled, the stall time could be intolerable long for the
users. When there are 20MB queued writeback pages and USB 1.1 is writing
them in 1MB/s, wait_on_page_writeback() will stuck for up to 20 seconds.
Not to mention it may be called multiple times.
So raise the bar to only enable PAGEOUT_IO_SYNC when priority goes below
DEF_PRIORITY/3, or 6.25% LRU size. As the default dirty throttle ratio is
20%, it will hardly be triggered by pure dirty pages. We'd better treat
PAGEOUT_IO_SYNC as some last resort workaround -- its stall time is so
uncomfortably long (easily goes beyond 1s).
The bar is only raised for (order < PAGE_ALLOC_COSTLY_ORDER) allocations,
which are easy to satisfy in 1TB memory boxes. So, although 6.25% of
memory could be an awful lot of pages to scan on a system with 1TB of
memory, it won't really have to busy scan that much.
Andreas tested an older version of this patch and reported that it mostly
fixed his problem. Mel Gorman helped improve it and KOSAKI Motohiro will
fix it further in the next patch.
after updating the value of the ICMP payload, inet_proto_csum_replace4() should
be called with zero pseudohdr.
Signed-off-by: Changli Gao <xiaosuo@gmail.com> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The main motivation of this patch changing strcpy() to strlcpy().
We strcpy() to copy a 48 byte buffers into a 49 byte buffers. So at
best the last byte has leaked information, or maybe there is an
overflow? Anyway, this patch closes the information leaks by zeroing
the memory and the calls to strlcpy() prevent overflows.
Signed-off-by: Dan Carpenter <error27@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch adds a limit for nframes as the number of frames in TX_SETUP and
RX_SETUP are derived from a single byte multiplex value by default.
Use-cases that would require to send/filter more than 256 CAN frames should
be implemented in userspace for complexity reasons anyway.
Additionally the assignments of unsigned values from userspace to signed
values in kernelspace and vice versa are fixed by using unsigned values in
kernelspace consistently.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Reported-by: Ben Hawkes <hawkes@google.com> Acked-by: Urs Thuermann <urs.thuermann@volkswagen.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
>Xin Xiaohui wrote:
> I looked into the code dev_gro_receive(), found the code here:
> if the frags[0] is pulled to 0, then the page will be released,
> and memmove() frags left.
> Is that right? I'm not sure if memmove do right or not, but
> frags[0].size is never set after memove at least. what I think
> a simple way is not to do anything if we found frags[0].size == 0.
> The patch is as followed.
...
This version of the patch fixes the bug directly in memmove.
Reported-by: "Xin, Xiaohui" <xiaohui.xin@intel.com> Signed-off-by: Jarek Poplawski <jarkao2@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
SunBlade-2500 has 'parallel' device node with compatible
property "pnpALI,1533,3" so add that to the ID table.
Reported-by: Mikael Pettersson <mikpe@it.uu.se> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
These just represent the secondary and further heads attached to the
card, and they have different sets of PCI bar registers to map.
So don't try to drive them in the main driver.
Reported-by: Frans van Berckel <fberckel@xs4all.nl> Tested-by: Frans van Berckel <fberckel@xs4all.nl> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes alignment of slab objects in case CONFIG_DEBUG_PAGEALLOC is
active.
Before this spot in kmem_cache_create, we have this situation:
- align contains the required alignment of the object
- cachep->obj_offset is 0 or equals align in case of CONFIG_DEBUG_SLAB
- size equals the size of the object, or object plus trailing redzone in case
of CONFIG_DEBUG_SLAB
This spot tries to fill one page per object if the object is in certain size
limits, however setting obj_offset to PAGE_SIZE - size does break the object
alignment since size may not be aligned with the required alignment.
This patch simply adds an ALIGN(size, align) to the equation and fixes the
object size detection accordingly.
This code in drivers/s390/cio/qdio_setup_init has lead to incorrectly aligned
slab objects (sizeof(struct qdio_q) equals 1792):
qdio_q_cache = kmem_cache_create("qdio_q", sizeof(struct qdio_q),
256, 0, NULL);
Acked-by: Christoph Lameter <cl@linux.com> Signed-off-by: Carsten Otte <cotte@de.ibm.com> Signed-off-by: Pekka Enberg <penberg@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The hibernate issues that got fixed in commit 985b823b9192 ("drm/i915:
fix hibernation since i915 self-reclaim fixes") turn out to have been
incomplete. Vefa Bicakci tested lots of hibernate cycles, and without
the __GFP_RECLAIMABLE flag the system eventually fails to resume.
With the flag added, Vefa can apparently hibernate forever (or until he
gets bored running his automated scripts, whichever comes first).
The reclaimable flag was there originally, and was one of the flags that
were dropped (unintentionally) by commit 4bdadb978569 ("drm/i915:
Selectively enable self-reclaim") that introduced all these problems,
but I didn't want to just blindly add back all the flags in commit 985b823b9192, and it looked like __GFP_RECLAIM wasn't necessary. It
clearly was.
I still suspect that there is some subtle reason we're missing that
causes the problems, but __GFP_RECLAIMABLE is certainly not wrong to use
in this context, and is what the code historically used. And we have no
idea what the causes the corruption without it.
Reported-and-tested-by: M. Vefa Bicakci <bicave@superonline.com> Cc: Dave Airlie <airlied@gmail.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Since commit 4bdadb9785696439c6e2b3efe34aa76df1149c83 ("drm/i915:
Selectively enable self-reclaim"), we've been passing GFP_MOVABLE to the
i915 page allocator where we weren't before due to some over-eager
removal of the page mapping gfp_flags games the code used to play.
This caused hibernate on Intel hardware to result in a lot of memory
corruptions on resume. See for example
http://bugzilla.kernel.org/show_bug.cgi?id=13811
Reported-by: Evengi Golov (in bugzilla) Signed-off-by: Dave Airlie <airlied@redhat.com> Tested-by: M. Vefa Bicakci <bicave@superonline.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Like the mlock() change previously, this makes the stack guard check
code use vma->vm_prev to see what the mapping below the current stack
is, rather than have to look it up with find_vma().
Also, accept an abutting stack segment, since that happens naturally if
you split the stack with mlock or mprotect.
It's a really simple list, and several of the users want to go backwards
in it to find the previous vma. So rather than have to look up the
previous entry with 'find_vma_prev()' or something similar, just make it
doubly linked instead.
This patch changes dm_hash_remove_all() to release _hash_lock when
removing a device. After removing the device, dm_hash_remove_all()
takes _hash_lock and searches the hash from scratch again.
This patch is a preparation for the next patch, which changes device
deletion code to wait for md reference to be 0. Without this patch,
the wait in the next patch may cause AB-BA deadlock:
CPU0 CPU1
-----------------------------------------------------------------------
dm_hash_remove_all()
down_write(_hash_lock)
table_status()
md = find_device()
dm_get(md)
<increment md->holders>
dm_get_live_or_inactive_table()
dm_get_inactive_table()
down_write(_hash_lock)
<in the md deletion code>
<wait for md->holders to be 0>
Test on a PXA310 platform with Samsung K9F2G08X0B NAND flash,
with tCH=5 and clk is 156MHz, ns2cycle(5, 156000000) returns -1.
ns2cycle returns negtive value will break NDTR0_tXX macros.
After checking the commit log, I found the problem is introduced by
commit 5b0d4d7c8a67c5ba3d35e6ceb0c5530cc6846db7
"[MTD] [NAND] pxa3xx: convert from ns to clock ticks more accurately"
To get num of clock cycles, we use below equation:
num of clock cycles = time (ns) / one clock cycle (ns) + 1
We need to add 1 cycle here because integer division will truncate the result.
It is possible the developers set the Min values in SPEC for timing settings.
Thus the truncate may cause problem, and it is safe to add an extra cycle here.
The various fields in NDTR{01} are in units of clock ticks minus one,
thus we should subtract 1 cycle then.
Thus the correct equation should be:
num of clock cycles = time (ns) / one clock cycle (ns) + 1 - 1
= time (ns) / one clock cycle (ns)
Signed-off-by: Axel Lin <axel.lin@gmail.com> Signed-off-by: Lei Wen <leiwen@marvell.com> Acked-by: Eric Miao <eric.y.miao@gmail.com> Signed-off-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Atheros PCIe wireless cards handled by ath5k do require L0s disabled.
For distributions shipping with CONFIG_PCIEASPM (this will be enabled
by default in the future in 2.6.36) this will also mean both L1 and L0s
will be disabled when a pre 1.1 PCIe device is detected. We do know L1
works correctly even for all ath5k pre 1.1 PCIe devices though but cannot
currently undue the effect of a blacklist, for details you can read
pcie_aspm_sanity_check() and see how it adjusts the device link
capability.
It may be possible in the future to implement some PCI API to allow
drivers to override blacklists for pre 1.1 PCIe but for now it is
best to accept that both L0s and L1 will be disabled completely for
distributions shipping with CONFIG_PCIEASPM rather than having this
issue present. Motivation for adding this new API will be to help
with power consumption for some of these devices.
Example of issues you'd see:
- On the Acer Aspire One (AOA150, Atheros Communications Inc. AR5001
Wireless Network Adapter [168c:001c] (rev 01)) doesn't work well
with ASPM enabled, the card will eventually stall on heavy traffic
with often 'unsupported jumbo' warnings appearing. Disabling
ASPM L0s in ath5k fixes these problems.
- On the same card you would see a storm of RXORN interrupts
even though medium is idle.
Credit for root causing and fixing the bug goes to Jussi Kivilinna.
Cc: David Quan <David.Quan@atheros.com> Cc: Matthew Garrett <mjg59@srcf.ucam.org> Cc: Tim Gardner <tim.gardner@canonical.com> Cc: Jussi Kivilinna <jussi.kivilinna@mbnet.fi> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It's not OK to call platform_device_add_resources() multiple times
in a row. Despite its name, this functions sets the resources, it
doesn't add them. So we have to prepare an array with all the
resources, and then call platform_device_add_resources() once.
Before this fix, only the last I/O resource would be actually
registered. The other I/O resources were leaked.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Cc: Jim Cromie <jim.cromie@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Thanks a lot for all the review and comments so far;) I'd like to send
the improved (V4) version of this patch.
This patch fixes a deadlock in OCFS2 ACL. We found this bug in OCFS2
and Samba integration using scenario, the symptom is several smbd
processes will be hung under heavy workload. Finally we found out it
is the nested PR lock calling that leads to this deadlock:
node1 node2
gr PR
|
V
PR(EX)---> BAST:OCFS2_LOCK_BLOCKED
|
V
rq PR
|
V
wait=1
After requesting the 2nd PR lock, the process "smbd" went into D
state. It can only be woken up when the 1st PR lock's RO holder equals
zero. There should be an ocfs2_inode_unlock in the calling path later
on, which can decrement the RO holder. But since it has been in
uninterruptible sleep, the unlock function has no chance to be called.
When testing cpu hotplug code on 32-bit we kept hitting the "CPU%d:
Stuck ??" message due to multiple cores concurrently accessing the
cpu_callin_mask, among others.
Since these codepaths are not protected from concurrent access due to
the fact that there's no sane reason for making an already complex
code unnecessarily more complex - we hit the issue only when insanely
switching cores off- and online - serialize hotplugging cores on the
sysfs level and be done with it.
[ v2.1: fix !HOTPLUG_CPU build ]
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <20100819181029.GC17171@aftab> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When we need to take both dlm_domain_lock and dlm->spinlock, we should take
them in order of: dlm_domain_lock then dlm->spinlock.
There is pathes disobey this order. That is calling dlm_lockres_put() with
dlm->spinlock held in dlm_run_purge_list. dlm_lockres_put() calls dlm_put() at
the ref and dlm_put() locks on dlm_domain_lock.
Fix:
Don't grab/put the dlm when the initialising/releasing lockres.
That grab is not required because we don't call dlm_unregister_domain()
based on refcount.
Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In the following situation, there remains an incorrect bit in refmap on the
recovery master. Finally the recovery master will fail at purging the lockres
due to the incorrect bit in refmap.
1) node A has no interest on lockres A any longer, so it is purging it.
2) the owner of lockres A is node B, so node A is sending de-ref message
to node B.
3) at this time, node B crashed. node C becomes the recovery master. it recovers
lockres A(because the master is the dead node B).
4) node A migrated lockres A to node C with a refbit there.
5) node A failed to send de-ref message to node B because it crashed. The failure
is ignored. no other action is done for lockres A any more.
For mormal, re-send the deref message to it to recovery master can fix it. Well,
ignoring the failure of deref to the original master and not recovering the lockres
to recovery master has the same effect. And the later is simpler.
Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Srinivas Eeda <srinivas.eeda@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The refcount record calculation in ocfs2_calc_refcount_meta_credits
is too optimistic that we can always allocate contiguous clusters
and handle an already existed refcount rec as a whole. Actually
because of file system fragmentation, we may have the chance to split
a refcount record into 3 parts during the transaction. So consider
the worst case in record calculation.
Signed-off-by: Tao Ma <tao.ma@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes two problems in dlm_run_purgelist
1. If a lockres is found to be in use, dlm_run_purgelist keeps trying to purge
the same lockres instead of trying the next lockres.
2. When a lockres is found unused, dlm_run_purgelist releases lockres spinlock
before setting DLM_LOCK_RES_DROPPING_REF and calls dlm_purge_lockres.
spinlock is reacquired but in this window lockres can get reused. This leads
to BUG.
This patch modifies dlm_run_purgelist to skip lockres if it's in use and purge
next lockres. It also sets DLM_LOCK_RES_DROPPING_REF before releasing the
lockres spinlock protecting it from getting reused.
When we have to take both dlm->master_lock and lockres->spinlock,
take them in order
lockres->spinlock and then dlm->master_lock.
The patch fixes a violation of the rule.
We can simply move taking dlm->master_lock to where we have dropped res->spinlock
since when we access res->state and free mle memory we don't need master_lock's
protection.
Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Setting the acl while creating a new inode depends on
the error codes of posix_acl_create_masq. This patch fix
a issue of overwriting the error codes of it.
Reported-by: Pawel Zawora <pzawora@gmail.com> Signed-off-by: Tiger Yang <tiger.yang@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I discovered tonight that ALSA no longer sets up a stream for the second ADC
provided by the Realtek ALC260 HDA codec. At some point alc_build_pcms()
started using stream_analog_alt_capture when constructing the second ADC
stream, but patch_alc260() was never updated accordingly. I have no idea
when this regression occurred. The trivial patch to patch_alc260() given
below fixes the problem as far as I can tell. The patch is against 2.6.35.
With some hardware combinations, the PCM interrupts are acknowledged
before the period boundary from the emu10k1 chip. The midlevel PCM code
gets confused and the playback stream is interrupted.
It seems that the interrupt processing shift by 2 samples is enough
to fix this issue. This default value does not harm other,
non-affected hardware.
The detection and loading of firmeware on riptide driver has been broken
due to rewrite of some codes, checking the presense wrongly.
This patch fixes the logic again.
mspro_block_remove() is called from detect thread that first calls the
mspro_block_stop(), which stops the request queue. If we call
del_gendisk() with the queue stopped we get a deadlock.
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com> Cc: Alex Dubov <oakad@yahoo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This commit makes the stack guard page somewhat less visible to user
space. It does this by:
- not showing the guard page in /proc/<pid>/maps
It looks like lvm-tools will actually read /proc/self/maps to figure
out where all its mappings are, and effectively do a specialized
"mlockall()" in user space. By not showing the guard page as part of
the mapping (by just adding PAGE_SIZE to the start for grows-up
pages), lvm-tools ends up not being aware of it.
- by also teaching the _real_ mlock() functionality not to try to lock
the guard page.
That would just expand the mapping down to create a new guard page,
so there really is no point in trying to lock it in place.
It would perhaps be nice to show the guard page specially in
/proc/<pid>/maps (or at least mark grow-down segments some way), but
let's not open ourselves up to more breakage by user space from programs
that depends on the exact deails of the 'maps' file.
Special thanks to Henrique de Moraes Holschuh for diving into lvm-tools
source code to see what was going on with the whole new warning.
We do in fact need to unmap the page table _before_ doing the whole
stack guard page logic, because if it is needed (mainly 32-bit x86 with
PAE and CONFIG_HIGHPTE, but other architectures may use it too) then it
will do a kmap_atomic/kunmap_atomic.
And those kmaps will create an atomic region that we cannot do
allocations in. However, the whole stack expand code will need to do
anon_vma_prepare() and vma_lock_anon_vma() and they cannot do that in an
atomic region.
Now, a better model might actually be to do the anon_vma_prepare() when
_creating_ a VM_GROWSDOWN segment, and not have to worry about any of
this at page fault time. But in the meantime, this is the
straightforward fix for the issue.
See https://bugzilla.kernel.org/show_bug.cgi?id=16588 for details.
It's wrong for several reasons, but the most direct one is that the
fault may be for the stack accesses to set up a previous SIGBUS. When
we have a kernel exception, the kernel exception handler does all the
fixups, not some user-level signal handler.
Even apart from the nested SIGBUS issue, it's also wrong to give out
kernel fault addresses in the signal handler info block, or to send a
SIGBUS when a system call already returns EFAULT.
.. which didn't show up in my tests because it's a no-op on x86-64 and
most other architectures. But we enter the function with the last-level
page table mapped, and should unmap it at exit.
This is a rather minimally invasive patch to solve the problem of the
user stack growing into a memory mapped area below it. Whenever we fill
the first page of the stack segment, expand the segment down by one
page.
Now, admittedly some odd application might _want_ the stack to grow down
into the preceding memory mapping, and so we may at some point need to
make this a process tunable (some people might also want to have more
than a single page of guarding), but let's try the minimal approach
first.
Tested with trivial application that maps a single page just below the
stack, and then starts recursing. Without this, we will get a SIGSEGV
_after_ the stack has smashed the mapping. With this patch, we'll get a
nice SIGBUS just as the stack touches the page just above the mapping.
Since 2.6.31, swap_map[]'s refcounting was changed to show that a used
swap entry is just for swap-cache, can be reused. Then, while scanning
free entry in swap_map[], a swap entry may be able to be reclaimed and
reused. It was caused by commit c9e444103b5e7a5 ("mm: reuse unused swap
entry if necessary").
But this caused deta corruption at resume. The scenario is
- Assume a clean-swap cache, but mapped.
- at hibernation_snapshot[], clean-swap-cache is saved as
clean-swap-cache and swap_map[] is marked as SWAP_HAS_CACHE.
- then, save_image() is called. And reuse SWAP_HAS_CACHE entry to save
image, and break the contents.
After resume:
- the memory reclaim runs and finds clean-not-referenced-swap-cache and
discards it because it's marked as clean. But here, the contents on
disk and swap-cache is inconsistent.
Hance memory is corrupted.
This patch avoids the bug by not reclaiming swap-entry during hibernation.
This is a quick fix for backporting.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Reported-by: Ondreg Zary <linux@rainbow-software.org> Tested-by: Ondreg Zary <linux@rainbow-software.org> Tested-by: Andrea Gelmini <andrea.gelmini@gmail.com> Signed-off-by: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When a raid1 array is configured to support write-behind
on some devices, it normally only reads from other devices.
If all devices are write-behind (because the rest have failed)
it is possible for a read request to be serviced before a
behind-write request, which would appear as data corruption.
So when forced to read from a WriteMostly device, wait for any
write-behind to complete, and don't start any more behind-writes.
If a command times out resulting in EH getting invoked, we wait for the
aborted commands to come back after sending the abort. Shorten
the amount of time we wait for these responses, to ensure we don't
get stuck in EH for several minutes.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Commands which are completed by the VIOS are placed on a CRQ
in kernel memory for the ibmvfc driver to process. Each CRQ
entry is 16 bytes. The ibmvfc driver reads the first 8 bytes
to check if the entry is valid, then reads the next 8 bytes to get
the handle, which is a pointer the completed command. This fixes
an issue seen on Power 7 where the processor reordered the
loads from memory, resulting in processing command completion
with a stale handle. This could result in command timeouts,
and also early completion of commands.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When removing several devices aic79xx will occasionally Oops
in ahd_handle_nonpkt_busfree during rescan. Looking at the
code I found that we're indeed not checking if the scb in
question is NULL. So check for it before accessing it.
Signed-off-by: Hannes Reinecke <hare@suse.de> Signed-off-by: James Bottomley <James.Bottomley@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
ocfs2_lock() will skip locks on file which has mode set to 02666. This
is a problem in cases where the mode of the file is changed after a
process has obtained a lock on the file.
ocfs2_lock() should skip the check for mandatory locks when unlocking a
file.
Signed-off-by: Sachin Prabhu <sprabhu@redhat.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We have to set MS_POSIXACL on remount as well. Otherwise VFS
would not know we started supporting ACLs after remount and
thus ACLs would not work.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
ocfs2 refcount tree is stored as an extent tree while
the leaf ocfs2_refcount_rec points to a refcount block.
The following step can trip a kernel panic.
mkfs.ocfs2 -b 512 -C 1M --fs-features=refcount $DEVICE
mount -t ocfs2 $DEVICE $MNT_DIR
FILE_NAME=$RANDOM
FILE_NAME_1=$RANDOM
FILE_REF="${FILE_NAME}_ref"
FILE_REF_1="${FILE_NAME}_ref_1"
for((i=0;i<305;i++))
do
# /mnt/1048576 is a file with 1048576 sizes.
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME_1
done
for((i=0;i<3;i++))
do
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
done
for((i=0;i<11;i++))
do
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME_1
done
reflink $MNT_DIR/$FILE_NAME $MNT_DIR/$FILE_REF
# write_f is a program which will write some bytes to a file at offset.
# write_f -f file_name -l offset -w write_bytes.
./write_f -f $MNT_DIR/$FILE_REF -l $[310*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_REF -l $[306*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_REF -l $[311*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_NAME -l $[310*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_NAME -l $[311*1048576] -w 4096
reflink $MNT_DIR/$FILE_NAME $MNT_DIR/$FILE_REF_1
./write_f -f $MNT_DIR/$FILE_NAME -l $[311*1048576] -w 4096
#kernel panic here.
The reason is that if the ocfs2_extent_rec is the last record
in a leaf extent block, the old solution fails to find the
suitable end cpos. So this patch try to walk through the b-tree,
find the next sub root and get the c_pos the next sub-tree starts
from.
btw, I have runned tristan's test case against the patched kernel
for several days and this type of kernel panic never happens again.
Signed-off-by: Tao Ma <tao.ma@oracle.com> Signed-off-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the lock master processes a successful operation (request,
convert, cancel, or unlock), it will process the effects of the
change before sending the reply for the operation. The "effects"
of the operation are:
- blocking callbacks (basts) for any newly granted locks
- waiting or converting locks that can now be granted
The cast is queued on the local node when the reply from the lock
master is received. This means that a lock holder can receive a
bast for a lock mode that is doesn't yet know has been granted.
Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When both blocking and completion callbacks are queued for lock,
the dlm would always deliver the completion callback (cast) first.
In some cases the blocking callback (bast) is queued before the
cast, though, and should be delivered first. This patch keeps
track of the order in which they were queued and delivers them
in that order.
This patch also keeps track of the granted mode in the last cast
and eliminates the following bast if the bast mode is compatible
with the preceding cast mode. This happens when a remotely mastered
lock is demoted, e.g. EX->NL, in which case the local node queues
a cast immediately after sending the demote message. In this way
a cast can be queued for a mode, e.g. NL, that makes an in-transit
bast extraneous.
Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Replace all GFP_KERNEL and ls_allocation with GFP_NOFS.
ls_allocation would be GFP_KERNEL for userland lockspaces
and GFP_NOFS for file system lockspaces.
It was discovered that any lockspaces on the system can
affect all others by triggering memory reclaim in the
file system which could in turn call back into the dlm
to acquire locks, deadlocking dlm threads that were
shared by all lockspaces, like dlm_recv.
Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Commit 57fe60df ("reiserfs: add atomic addition of selinux attributes
during inode creation") contains a bug that will cause it to oops when
mounting a file system that didn't previously contain extended attributes
on a system using security.* xattrs.
The issue is that while creating the privroot during mount
reiserfs_security_init calls reiserfs_xattr_jcreate_nblocks which
dereferences the xattr root. The xattr root doesn't exist, so we get an
oops.
The reiserfs journal behaves inconsistently when determining whether to
allow a mount of a read-only device.
This is due to the use of the continue_replay variable to short circuit
the journal scanning. If it's set, it's assumed that there are
transactions to replay, but there may not be. If it's unset, it's assumed
that there aren't any, and that may not be the case either.
I've observed two failure cases:
1) Where a clean file system on a read-only device refuses to mount
2) Where a clean file system on a read-only device passes the
optimization and then tries writing the journal header to update
the latest mount id.
The former is easily observable by using a freshly created file system on
a read-only loopback device.
This patch moves the check into journal_read_transaction, where it can
bail out before it's about to replay a transaction. That way it can go
through and skip transactions where appropriate, yet still refuse to mount
a file system with outstanding transactions.
Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We have 2 mount options, "barrier" and "auto_da_alloc" which may or
may not take a 1/0 argument. This causes the ext4 superblock mount
code to subtract uninitialized pointers and pass the result to
kmalloc, which results in very noisy failures.
Per Ted's suggestion, initialize the args struct so that
we know whether match_token() found an argument for the
option, and skip match_int() if not.
Also, return error (0) from parse_options if we thought
we found an argument, but match_int() Fails.
Reported-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Dan Roseberg has reported a problem with the MOVE_EXT ioctl. If the
donor file is an append-only file, we should not allow the operation
to proceed, lest we end up overwriting the contents of an append-only
file.
Earlier, Ingo Molnar posted a patch to make it so that the kernel would avoid
reading _PPC on his broken T60. Unfortunately, it seems that with Thomas
Renninger's patch last July to eliminate _PPC evaluations when the processor
driver loads, the kernel never actually reads _PPC at all! This is problematic
if you happen to boot your non-T60 computer in a state where the BIOS _wants_
_PPC to be something other than zero.
So, put the _PPC evaluation back into acpi_processor_get_performance_info if
ignore_ppc isn't 1.
Signed-off-by: Darrick J. Wong <djwong@us.ibm.com> Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
During a EEH recover, the pci_dev structure can be null, mainly if an
eeh event is detected during cpi config operation. In this case, the
pci_dev will not be known (and will be null) the kernel will crash
with the following message:
Unable to handle kernel paging request for data at address 0x000000a0
Faulting instruction address: 0xc00000000006b8b4
Oops: Kernel access of bad area, sig: 11 [#1]
It turns out that the system has three io apic controllers, but
boot ioapic routing is in the second one, and that gsi_base is
not 0 - it is using a bunch of INT_SRC_OVR...
So these recent changes:
1. one set routing for first io apic controller
2. assume irq = gsi
... will break that system.
So try to remap those gsis, need to seperate boot_ioapic_idx
detection out of enable_IO_APIC() and call them early.
So introduce boot_ioapic_idx, and remap_ioapic_gsi()...
-v2: shift gsi with delta instead of gsi_base of boot_ioapic_idx
-v3: double check with find_isa_irq_apic(0, mp_INT) to get right
boot_ioapic_idx
-v4: nr_legacy_irqs
-v5: add print out for boot_ioapic_idx, and also make it could be
applied for current kernel and previous kernel
-v6: add bus_irq, in acpi_sci_ioapic_setup, so can get overwride
for sci right mapping...
-v7: looks like pnpacpi get irq instead of gsi, so need to revert
them back...
-v8: split into two patches
-v9: according to Eric, use fixed 16 for shifting instead of remap
-v10: still need to touch rsparser.c
-v11: just revert back to way Eric suggest...
anyway the ioapic in first ioapic is blocked by second...
-v12: two patches, this one will add more loop but check apic_id and irq > 16
Reported-by: Iranna D Ankad <iranna.ankad@in.ibm.com> Bisected-by: Iranna D Ankad <iranna.ankad@in.ibm.com> Tested-by: Gary Hade <garyhade@us.ibm.com> Signed-off-by: Yinghai Lu <yinghai@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Thomas Renninger <trenn@suse.de> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Suresh Siddha <suresh.b.siddha@intel.com> Cc: len.brown@intel.com
LKML-Reference: <4B8A321A.1000008@kernel.org> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When load aesni-intel and ghash_clmulni-intel driver,kernel will complain no
test for some internal used algorithm.
The strange information as following:
alg: No test for __aes-aesni (__driver-aes-aesni)
alg: No test for __ecb-aes-aesni (__driver-ecb-aes-aesni)
alg: No test for __cbc-aes-aesni (__driver-cbc-aes-aesni)
alg: No test for __ecb-aes-aesni (cryptd(__driver-ecb-aes-aesni)
alg: No test for __ghash (__ghash-pclmulqdqni)
alg: No test for __ghash (cryptd(__ghash-pclmulqdqni))
This patch add NULL test entries for these algorithm and driver.
Signed-off-by: Song Youquan <youquan.song@intel.com> Signed-off-by: Hang Ying <ying.huang@intel.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It's possible that SBA IOMMU might fail to find I/O space under heavy
I/Os. SBA IOMMU panics on allocation failure but it shouldn't; drivers
can handle the failure. The majority of other IOMMU drivers don't panic
on allocation failure.
This patch fixes SBA IOMMU path to handle allocation failure properly.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Tony Luck <tony.luck@intel.com> Acked-by: Leonardo Chiquitto <lchiquitto@novell.com> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Due to recent load-balancer changes that delay the task migration to
the next wakeup, the adaptive mutex spinning ends up in a live lock
when the owner's CPU gets offlined because the cpu_online() check
lives before the owner running check.
This patch changes mutex_spin_on_owner() to return 0 (don't spin) in
any case where we aren't sure about the owner struct validity or CPU
number, and if the said CPU is offline. There is no point going back &
re-evaluate spinning in corner cases like that, let's just go to
sleep.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271212509.13059.135.camel@pasglop> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is a real fix for problem of utime/stime values decreasing
described in the thread:
http://lkml.org/lkml/2009/11/3/522
Now cputime is accounted in the following way:
- {u,s}time in task_struct are increased every time when the thread
is interrupted by a tick (timer interrupt).
- When a thread exits, its {u,s}time are added to signal->{u,s}time,
after adjusted by task_times().
- When all threads in a thread_group exits, accumulated {u,s}time
(and also c{u,s}time) in signal struct are added to c{u,s}time
in signal struct of the group's parent.
So {u,s}time in task struct are "raw" tick count, while
{u,s}time and c{u,s}time in signal struct are "adjusted" values.
And accounted values are used by:
- task_times(), to get cputime of a thread:
This function returns adjusted values that originates from raw
{u,s}time and scaled by sum_exec_runtime that accounted by CFS.
- thread_group_cputime(), to get cputime of a thread group:
This function returns sum of all {u,s}time of living threads in
the group, plus {u,s}time in the signal struct that is sum of
adjusted cputimes of all exited threads belonged to the group.
The problem is the return value of thread_group_cputime(),
because it is mixed sum of "raw" value and "adjusted" value:
This misbehavior can break {u,s}time monotonicity.
Assume that if there is a thread that have raw values greater
than adjusted values (e.g. interrupted by 1000Hz ticks 50 times
but only runs 45ms) and if it exits, cputime will decrease (e.g.
-5ms).
But task_times() contains hard divisions, so applying it for
every thread should be avoided.
This patch fixes the above problem in the following way:
- Modify thread's exit (= __exit_signal()) not to use task_times().
It means {u,s}time in signal struct accumulates raw values instead
of adjusted values. As the result it makes thread_group_cputime()
to return pure sum of "raw" values.
- Introduce a new function thread_group_times(*task, *utime, *stime)
that converts "raw" values of thread_group_cputime() to "adjusted"
values, in same calculation procedure as task_times().
- Modify group's exit (= wait_task_zombie()) to use this introduced
thread_group_times(). It make c{u,s}time in signal struct to
have adjusted values like before this patch.
- Replace some thread_group_cputime() by thread_group_times().
This replacements are only applied where conveys the "adjusted"
cputime to users, and where already uses task_times() near by it.
(i.e. sys_times(), getrusage(), and /proc/<PID>/stat.)
This patch have a positive side effect:
- Before this patch, if a group contains many short-life threads
(e.g. runs 0.9ms and not interrupted by ticks), the group's
cputime could be invisible since thread's cputime was accumulated
after adjusted: imagine adjustment function as adj(ticks, runtime),
{adj(0, 0.9) + adj(0, 0.9) + ....} = {0 + 0 + ....} = 0.
After this patch it will not happen because the adjustment is
applied after accumulated.
v2:
- remove if()s, put new variables into signal_struct.
It only changed the type of return value, but not the
implementation. As the result the granularity of task_s/utime()
is still that of clock_t, not that of cputime_t.
So using task_s/utime() in __exit_signal() makes values
accumulated to the signal struct to be rounded and coarse
grained.
This patch removes casts to clock_t in task_u/stime(), to keep
granularity of cputime_t over the calculation.
v2:
Use div_u64() to avoid error "undefined reference to `__udivdi3`"
on some 32bit systems.
Since commit 0a544198 "timekeeping: Move NTP adjusted clock multiplier
to struct timekeeper" the clock multiplier of vsyscall is updated with
the unmodified clock multiplier of the clock source and not with the
NTP adjusted multiplier of the timekeeper.
Add a new argument "mult" to update_vsyscall() and hand in the
timekeeping internal NTP adjusted multiplier.
Signed-off-by: Lin Ming <ming.m.lin@intel.com> Cc: "Zhang Yanmin" <yanmin_zhang@linux.intel.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Tony Luck <tony.luck@intel.com>
LKML-Reference: <1258436990.17765.83.camel@minggr.sh.intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Kurt Garloff <garloff@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
On a system with NOHZ=y tick_check_idle calls tick_nohz_stop_idle and
tick_nohz_update_jiffies. Given the right conditions (ts->idle_active
and/or ts->tick_stopped) both function get a time stamp with ktime_get.
The same time stamp can be reused if both function require one.
On s390 this change has the additional benefit that gcc inlines the
tick_nohz_stop_idle function into tick_check_idle. The number of
instructions to execute tick_check_idle drops from 225 to 144
(without the ktime_get optimization it is 367 vs 215 instructions).
before:
0) | tick_check_idle() {
0) | tick_nohz_stop_idle() {
0) | ktime_get() {
0) | read_tod_clock() {
0) 0.601 us | }
0) 1.765 us | }
0) 3.047 us | }
0) | ktime_get() {
0) | read_tod_clock() {
0) 0.570 us | }
0) 1.727 us | }
0) | tick_do_update_jiffies64() {
0) 0.609 us | }
0) 8.055 us | }
after:
0) | tick_check_idle() {
0) | ktime_get() {
0) | read_tod_clock() {
0) 0.617 us | }
0) 1.773 us | }
0) | tick_do_update_jiffies64() {
0) 0.593 us | }
0) 4.477 us | }
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: john stultz <johnstul@us.ibm.com>
LKML-Reference: <20090929122533.206589318@de.ibm.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: John Jolly <jjolly@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Allow the architecture to request a normal jiffy tick when the system
goes idle and tick_nohz_stop_sched_tick is called . On s390 the hook is
used to prevent the system going fully idle if there has been an
interrupt other than a clock comparator interrupt since the last wakeup.
On s390 the HiperSockets response time for 1 connection ping-pong goes
down from 42 to 34 microseconds. The CPU cost decreases by 27%.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
LKML-Reference: <20090929122533.402715150@de.ibm.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: John Jolly <jjolly@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
My test do: fallocate a big file and do write. The file is 512M, but
after file write is done btrfs-debug-tree shows:
item 6 key (257 EXTENT_DATA 0) itemoff 3516 itemsize 53
extent data disk byte 1103101952 nr 536870912
extent data offset 0 nr 399634432 ram 536870912
extent compression 0
Looks like a regression introducted by 6c7d54ac87f338c479d9729e8392eca3f76e11e1, where we set wrong slot.
Signed-off-by: Shaohua Li <shaohua.li@intel.com> Acked-by: Yan Zheng <zheng.yan@oracle.com> Signed-off-by: Chris Mason <chris.mason@oracle.com> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
you would see long stalls where no work was being done. That is because we were
doing all this extra work to read in the file extent outside of the transaction,
however in the random io case this ends up hurting us because the file extents
are not there to begin with. So axe this logic, since we end up reading in the
file extent when we go to update it anyway. This took the fio job from 11 mb/s
with several ~10 second stalls to 24 mb/s to a couple of 1-2 second stalls.
Signed-off-by: Josef Bacik <josef@redhat.com> Signed-off-by: Chris Mason <chris.mason@oracle.com> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
projects / karo-tx-linux.git / log