Kdump kernel needs to use only those memory regions that it is allowed
to use (crashkernel, rtas, tce, etc.). Each of these regions have
their own sizes and are currently added under 'linux,usable-memory'
property under each memory@xxx node of the device tree.
The ibm,dynamic-memory property of ibm,dynamic-reconfiguration-memory
node (on POWER6) now stores in it the representation for most of the
logical memory blocks with the size of each memory block being a
constant (lmb_size). If one or more or part of the above mentioned
regions lie under one of the lmb from ibm,dynamic-memory property,
there is a need to identify those regions within the given lmb.
This makes the kernel recognize a new 'linux,drconf-usable-memory'
property added by kexec-tools. Each entry in this property is of the
form of a count followed by that many (base, size) pairs for the above
mentioned regions. The number of cells in the count value is given by
the #size-cells property of the root node.
Signed-off-by: Chandru Siddalingappa <chandru@in.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The return code from invocation of the notifier for
pSeries_reconfig_chain during update of the device tree is not
checked. This causes writes to /proc/ppc64/ofdt to update memory
properties (i.e. ibm,dyamic-reconfiguration-memory) to always
return success, instead of the result of the notifier chain.
This happens specifically when we remove/add memory from the
device tree on machines using memory specified in the
ibm,dynamic-reconfiguration-memory property of the device tree.
Signed-off-by: Nathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This new copy_4K_page() function was originally tuned for the best
performance on the Cell processor, but after testing on more 64bit
powerpc chips it was found that with a small modification it either
matched the performance offered by the current mainline version or
bettered it by a small amount.
It was found that on a Cell-based QS22 blade the amount of system
time measured when compiling a 2.6.26 pseries_defconfig decreased
by 4%. Using the same test, a 4-way 970MP machine saw a decrease of
2% in system time. No noticeable change was seen on Power4, Power5
or Power6.
The 4096 byte page is copied in thirty-two 128 byte strides. An
initial setup loop executes dcbt instructions for the whole source
page and dcbz instructions for the whole destination page. To do
this, the cache line size is retrieved from ppc64_caches.
A new CPU feature bit, CPU_FTR_CP_USE_DCBTZ, (introduced in the
previous patch) is used to make the modification to this new copy
routine - on Power4, 970 and Cell the feature bit is set so the
setup loop is executed, but on all other 64bit chips the setup
loop is nop'ed out.
Signed-off-by: Mark Nelson <markn@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Add a new CPU feature bit, CPU_FTR_CP_USE_DCBTZ, to be added to the
64bit powerpc chips that benefit from having dcbt and dcbz
instructions used in their memory copy routines.
This will be used in a subsequent patch that updates copy_4K_page().
The new bit is added to Cell, PPC970 and Power4 because they show
better performance with the new copy_4K_page() when dcbt and dcbz
instructions are used.
Signed-off-by: Mark Nelson <markn@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The gfs2 superblock pointer is NULL after a failed mount. When control
eventually goes to gfs2_kill_sb, we dereference this NULL pointer. This
patch ensures that the gfs2 superblock pointer is not NULL before being
dereferenced in gfs2_kill_sb.
Signed-off-by: Abhijith Das <adas@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This patch fixes a problem whereby a direct_io write doesn't fall
back to buffered write properly at end of file.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Initialized total objects atomic for the node in init_kmem_cache_node. The
uninitialized value was ruining the stats in /proc/slabinfo.
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Salman Qazi <sqazi@google.com>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
The SEC's h/w IV out implementation DMAs the trailing encrypted payload
block of the last encryption to ctx->iv. Since the last encryption may
still be pending completion, we can sufficiently prevent successive
packets from being transmitted with the same IV by xoring with sequence
number.
Also initialize alg_list earlier to prevent oopsing on a failed probe.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Prevent the HPET resources from appearing in PCI device 14.0 which
confuses the PCI resource engine.
Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
They were already called once in arch/x86/kernel/setup.c - we don't need to call them again.
fixes:
http://bugzilla.kernel.org/show_bug.cgi?id=11485
Signed-off-by: Alex Nixon <alex.nixon@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There's no good reason why a resource_size_t shouldn't just be a
physical address, so simply redefine it in terms of phys_addr_t.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
PFN_PHYS, as its name suggests, turns a pfn into a physical address.
However, it is a macro which just operates on its argument without
modifying its type. pfns are typed unsigned long, but an unsigned
long may not be long enough to hold a physical address (32-bit systems
with more than 32 bits of physcial address).
Make sure we cast to phys_addr_t to return a complete result.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add a kernel-wide "phys_addr_t" which is guaranteed to be able to hold
any physical address. By default it equals the word size of the
architecture, but a 32-bit architecture can set ARCH_PHYS_ADDR_T_64BIT
if it needs a 64-bit phys_addr_t.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
fix the UP build:
In file included from arch/x86/kernel/asm-offsets_32.c:9,
from arch/x86/kernel/asm-offsets.c:3:
include/linux/sched.h: In function ‘thread_group_cputime_clone_thread’:
include/linux/sched.h:2272: warning: no return statement in function returning non-void
include/linux/sched.h: In function ‘thread_group_cputime_account_user’:
include/linux/sched.h:2284: error: invalid type argument of ‘->’ (have ‘struct task_cputime’)
include/linux/sched.h:2284: error: invalid type argument of ‘->’ (have ‘struct task_cputime’)
include/linux/sched.h: In function ‘thread_group_cputime_account_system’:
include/linux/sched.h:2291: error: invalid type argument of ‘->’ (have ‘struct task_cputime’)
include/linux/sched.h:2291: error: invalid type argument of ‘->’ (have ‘struct task_cputime’)
include/linux/sched.h: In function ‘thread_group_cputime_account_exec_runtime’:
include/linux/sched.h:2298: error: invalid type argument of ‘->’ (have ‘struct task_cputime’)
distcc[14501] ERROR: compile arch/x86/kernel/asm-offsets.c on a/30 failed
make[1]: *** [arch/x86/kernel/asm-offsets.s] Error 1
Signed-off-by: Ingo Molnar <mingo@elte.hu>
PFN_PHYS() can truncate large addresses unless its passed a suitable
large type. This is fixed more generally in the patch series
introducing phys_addr_t, but we need a short-term fix to solve a
Xen regression reported by Roberto De Ioris.
Reported-by: Roberto De Ioris <roberto@unbit.it>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
gart alloc_coherent need to do virtual mapppings only when an
allocated buffer is not DMA-capable for a device.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
x86's common alloc_coherent (dma_alloc_coherent in dma-mapping.h) sets
up the gfp flag according to the device dma_mask but Calgary doesn't
need it because of virtual mappings. This patch avoids unnecessary low
zone allocation.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Muli Ben-Yehuda <muli@il.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently, GART IOMMU ingores device's dma_mask when it does virtual
mappings. So it could give a device a virtual address that the device
can't access to.
This patch fixes the above problem.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Several IOMMUs do the same thing to get the dma_mask of a device. This
adds a helper function to do the same thing to sweep them.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This function helps IOMMUs to know the highest address that a device
can access to.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
fix:
kernel/fork.c:843: error: ‘struct signal_struct’ has no member named ‘sum_sched_runtime’
kernel/irq/handle.c:117: warning: ‘sparse_irq_lock’ defined but not used
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Overview
This patch reworks the handling of POSIX CPU timers, including the
ITIMER_PROF, ITIMER_VIRT timers and rlimit handling. It was put together
with the help of Roland McGrath, the owner and original writer of this code.
The problem we ran into, and the reason for this rework, has to do with using
a profiling timer in a process with a large number of threads. It appears
that the performance of the old implementation of run_posix_cpu_timers() was
at least O(n*3) (where "n" is the number of threads in a process) or worse.
Everything is fine with an increasing number of threads until the time taken
for that routine to run becomes the same as or greater than the tick time, at
which point things degrade rather quickly.
This patch fixes bug 9906, "Weird hang with NPTL and SIGPROF."
Code Changes
This rework corrects the implementation of run_posix_cpu_timers() to make it
run in constant time for a particular machine. (Performance may vary between
one machine and another depending upon whether the kernel is built as single-
or multiprocessor and, in the latter case, depending upon the number of
running processors.) To do this, at each tick we now update fields in
signal_struct as well as task_struct. The run_posix_cpu_timers() function
uses those fields to make its decisions.
We define a new structure, "task_cputime," to contain user, system and
scheduler times and use these in appropriate places:
struct task_cputime {
cputime_t utime;
cputime_t stime;
unsigned long long sum_exec_runtime;
};
This is included in the structure "thread_group_cputime," which is a new
substructure of signal_struct and which varies for uniprocessor versus
multiprocessor kernels. For uniprocessor kernels, it uses "task_cputime" as
a simple substructure, while for multiprocessor kernels it is a pointer:
struct thread_group_cputime {
struct task_cputime totals;
};
struct thread_group_cputime {
struct task_cputime *totals;
};
We also add a new task_cputime substructure directly to signal_struct, to
cache the earliest expiration of process-wide timers, and task_cputime also
replaces the it_*_expires fields of task_struct (used for earliest expiration
of thread timers). The "thread_group_cputime" structure contains process-wide
timers that are updated via account_user_time() and friends. In the non-SMP
case the structure is a simple aggregator; unfortunately in the SMP case that
simplicity was not achievable due to cache-line contention between CPUs (in
one measured case performance was actually _worse_ on a 16-cpu system than
the same test on a 4-cpu system, due to this contention). For SMP, the
thread_group_cputime counters are maintained as a per-cpu structure allocated
using alloc_percpu(). The timer functions update only the timer field in
the structure corresponding to the running CPU, obtained using per_cpu_ptr().
We define a set of inline functions in sched.h that we use to maintain the
thread_group_cputime structure and hide the differences between UP and SMP
implementations from the rest of the kernel. The thread_group_cputime_init()
function initializes the thread_group_cputime structure for the given task.
The thread_group_cputime_alloc() is a no-op for UP; for SMP it calls the
out-of-line function thread_group_cputime_alloc_smp() to allocate and fill
in the per-cpu structures and fields. The thread_group_cputime_free()
function, also a no-op for UP, in SMP frees the per-cpu structures. The
thread_group_cputime_clone_thread() function (also a UP no-op) for SMP calls
thread_group_cputime_alloc() if the per-cpu structures haven't yet been
allocated. The thread_group_cputime() function fills the task_cputime
structure it is passed with the contents of the thread_group_cputime fields;
in UP it's that simple but in SMP it must also safely check that tsk->signal
is non-NULL (if it is it just uses the appropriate fields of task_struct) and,
if so, sums the per-cpu values for each online CPU. Finally, the three
functions account_group_user_time(), account_group_system_time() and
account_group_exec_runtime() are used by timer functions to update the
respective fields of the thread_group_cputime structure.
Non-SMP operation is trivial and will not be mentioned further.
The per-cpu structure is always allocated when a task creates its first new
thread, via a call to thread_group_cputime_clone_thread() from copy_signal().
It is freed at process exit via a call to thread_group_cputime_free() from
cleanup_signal().
All functions that formerly summed utime/stime/sum_sched_runtime values from
from all threads in the thread group now use thread_group_cputime() to
snapshot the values in the thread_group_cputime structure or the values in
the task structure itself if the per-cpu structure hasn't been allocated.
Finally, the code in kernel/posix-cpu-timers.c has changed quite a bit.
The run_posix_cpu_timers() function has been split into a fast path and a
slow path; the former safely checks whether there are any expired thread
timers and, if not, just returns, while the slow path does the heavy lifting.
With the dedicated thread group fields, timers are no longer "rebalanced" and
the process_timer_rebalance() function and related code has gone away. All
summing loops are gone and all code that used them now uses the
thread_group_cputime() inline. When process-wide timers are set, the new
task_cputime structure in signal_struct is used to cache the earliest
expiration; this is checked in the fast path.
Performance
The fix appears not to add significant overhead to existing operations. It
generally performs the same as the current code except in two cases, one in
which it performs slightly worse (Case 5 below) and one in which it performs
very significantly better (Case 2 below). Overall it's a wash except in those
two cases.
I've since done somewhat more involved testing on a dual-core Opteron system.
Case 1: With no itimer running, for a test with 100,000 threads, the fixed
kernel took 1428.5 seconds, 513 seconds more than the unfixed system,
all of which was spent in the system. There were twice as many
voluntary context switches with the fix as without it.
Case 2: With an itimer running at .01 second ticks and 4000 threads (the most
an unmodified kernel can handle), the fixed kernel ran the test in
eight percent of the time (5.8 seconds as opposed to 70 seconds) and
had better tick accuracy (.012 seconds per tick as opposed to .023
seconds per tick).
Case 3: A 4000-thread test with an initial timer tick of .01 second and an
interval of 10,000 seconds (i.e. a timer that ticks only once) had
very nearly the same performance in both cases: 6.3 seconds elapsed
for the fixed kernel versus 5.5 seconds for the unfixed kernel.
With fewer threads (eight in these tests), the Case 1 test ran in essentially
the same time on both the modified and unmodified kernels (5.2 seconds versus
5.8 seconds). The Case 2 test ran in about the same time as well, 5.9 seconds
versus 5.4 seconds but again with much better tick accuracy, .013 seconds per
tick versus .025 seconds per tick for the unmodified kernel.
Since the fix affected the rlimit code, I also tested soft and hard CPU limits.
Case 4: With a hard CPU limit of 20 seconds and eight threads (and an itimer
running), the modified kernel was very slightly favored in that while
it killed the process in 19.997 seconds of CPU time (5.002 seconds of
wall time), only .003 seconds of that was system time, the rest was
user time. The unmodified kernel killed the process in 20.001 seconds
of CPU (5.014 seconds of wall time) of which .016 seconds was system
time. Really, though, the results were too close to call. The results
were essentially the same with no itimer running.
Case 5: With a soft limit of 20 seconds and a hard limit of 2000 seconds
(where the hard limit would never be reached) and an itimer running,
the modified kernel exhibited worse tick accuracy than the unmodified
kernel: .050 seconds/tick versus .028 seconds/tick. Otherwise,
performance was almost indistinguishable. With no itimer running this
test exhibited virtually identical behavior and times in both cases.
In times past I did some limited performance testing. those results are below.
On a four-cpu Opteron system without this fix, a sixteen-thread test executed
in 3569.991 seconds, of which user was 3568.435s and system was 1.556s. On
the same system with the fix, user and elapsed time were about the same, but
system time dropped to 0.007 seconds. Performance with eight, four and one
thread were comparable. Interestingly, the timer ticks with the fix seemed
more accurate: The sixteen-thread test with the fix received 149543 ticks
for 0.024 seconds per tick, while the same test without the fix received 58720
for 0.061 seconds per tick. Both cases were configured for an interval of
0.01 seconds. Again, the other tests were comparable. Each thread in this
test computed the primes up to 25,000,000.
I also did a test with a large number of threads, 100,000 threads, which is
impossible without the fix. In this case each thread computed the primes only
up to 10,000 (to make the runtime manageable). System time dominated, at
1546.968 seconds out of a total 2176.906 seconds (giving a user time of
629.938s). It received 147651 ticks for 0.015 seconds per tick, still quite
accurate. There is obviously no comparable test without the fix.
Signed-off-by: Frank Mayhar <fmayhar@google.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add support for mv643xx_eth versions that have no transmit bandwidth
control registers at all, such as the ethernet block found in the
Marvell 88F6183 ARM SoC.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Currently, the receive processing reads ->byte_cnt twice (once to
update interface statistics and once to properly size the data area
of the received skb), but since receive descriptors live in uncached
memory, caching this value in a local variable saves one uncached
access, and increases routing performance a tiny little bit more.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
introduce do_rt_sigreturn(), to collect common part of sys_rt_sigreturn().
No change in functionality intended.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When setup frame fails, force_sigsegv is called and returns -EFAULT.
There is similar code in ia32_setup_frame(), ia32_setup_rt_frame(),
__setup_frame() and __setup_rt_frame().
Make them identical.
No change in functionality intended.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Since the size of the receive queue is directly related to the data
cache footprint of the driver (between refilling a receive ring entry
with a fresh skb and receiving a packet in that entry, queue_size - 1
other skbs will have been touched), shrink the default receive queue
size to a saner number of entries, as 400 is definite overkill for
almost all workloads.
While we are at it, trim the default transmit queue size a bit as well.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Krzysztof found some old cyrix cpu where an mtrr-alike cpu feature was
not detected properly.
this one is based on Krzysztof' patch, and we call ->c_identify() in
early_identify_cpu.
need to call c_identify() for cpus without cpuid even earlier ...
v2: Krzysztof point out need to give cyrix another chance about cpuid
checking again, after ->c_identify() enables cpuid for it
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Get rid of the skb pointer array that we currently use for transmit
reclaim, and replace it with an skb queue, to which skbuffs are appended
when they are passed to the xmit function, and removed from the front
and freed when we do transmit queue reclaim and hit a descriptor with
the 'owned by device' bit clear and 'last descriptor' bit set.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
By moving DMA unmapping during transmit reclaim back under the netif
tx lock, we avoid the situation where we read the DMA address and buffer
length from the descriptor under the lock and then not do anything with
that data after dropping the lock on platforms where the DMA unmapping
routines are all NOPs (which is the case on all ARM platforms that
mv643xx_eth is used on at least).
This saves two uncached reads, which makes a small but measurable
performance difference in routing benchmarks.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Since our ->hard_start_xmit() method is already called under spinlock
protection (the netif tx queue lock), we can simply make that lock
cover the private transmit state (descriptor ring indexes et al.) as
well, which avoids having to use a private lock to protect that state.
Since this was the last user of the driver-private spinlock, it can
be killed off.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Move link status handling, transmit reclaim and TX_END handling from
the interrupt handler to the napi poll handler. This allows switching
->lock over to a non-IRQ-safe lock and removes all explicit interrupt
disabling from the driver.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
The break after the return serves no purpose.
Signed-off-by: Julia Lawall <julia@diku.dk>
Reviewed-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
* Rename 'next' to 'dep'
* Move the channel switch check inside the loop to simplify
termination
Acked-by: Ilya Yanok <yanok@emcraft.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
There are a few commits that misencoded my name (or used "oe" instead of
"ö"). So add a correct version to .mailmap.
Signed-off-by: Uwe Kleine-König <ukleinek@informatik.uni-freiburg.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* master.kernel.org:/home/rmk/linux-2.6-arm:
[ARM] Fix PCI_DMA_BUS_IS_PHYS for ARM
[ARM] 5247/1: tosa: SW_EAR_IN support
[ARM] 5246/1: tosa: add proper clock alias for tc6393xb clock
[ARM] 5245/1: Fix warning about unused return value in drivers/pcmcia
[ARM] OMAP: Fix MMC device data
imx serial: fix rts handling for non imx1 based hardware
imx serial: set RXD mux bit on i.MX27 and i.MX31
i.MX serial: fix init failure
pcm037: add rts/cts support for serial port
