With the new ring buffer infrastructure in ftrace, I'm trying to make
ftrace a little more light weight.
This patch converts a lot of the local_irq_save/restore into
preempt_disable/enable. The original preempt count in a lot of cases
has to be sent in as a parameter so that it can be recorded correctly.
Some places were recording it incorrectly before anyway.
This is also laying the ground work to make ftrace a little bit
more reentrant, and remove all locking. The function tracers must
still protect from reentrancy.
Note: All the function tracers must be careful when using preempt_disable.
It must do the following:
resched = need_resched();
preempt_disable_notrace();
[...]
if (resched)
preempt_enable_no_resched_notrace();
else
preempt_enable_notrace();
The reason is that if this function traces schedule() itself, the
preempt_enable_notrace() will cause a schedule, which will lead
us into a recursive failure.
If we needed to reschedule before calling preempt_disable, we
should have already scheduled. Since we did not, this is most
likely that we should not and are probably inside a schedule
function.
If resched was not set, we still need to catch the need resched
flag being set when preemption was off and the if case at the
end will catch that for us.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The current method of overlaying the page frame as the buffer page pointer
can be very dangerous and limits our ability to do other things with
a page from the buffer, like send it off to disk.
This patch allocates the buffer_page instead of overlaying the page's
page frame. The use of the buffer_page has hardly changed due to this.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The mmiotrace map had a bug that would typecast the entry from
the trace to the wrong type. That is a known danger of C typecasts,
there's absolutely zero checking done on them.
Help that problem a bit by using a GCC extension to implement a
type filter that restricts the types that a trace record can be
cast into, and by adding a dynamic check (in debug mode) to verify
the type of the entry.
This patch adds a macro to assign all entries of ftrace using the type
of the variable and checking the entry id. The typecasts are now done
in the macro for only those types that it knows about, which should
be all the types that are allowed to be read from the tracer.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The old "lock always" scheme had issues with lockdep, and was not very
efficient anyways.
This patch does a new design to be partially lockless on writes.
Writes will add new entries to the per cpu pages by simply disabling
interrupts. When a write needs to go to another page than it will
grab the lock.
A new "read page" has been added so that the reader can pull out a page
from the ring buffer to read without worrying about the writer writing over
it. This allows us to not take the lock for all reads. The lock is
now only taken when a read needs to go to a new page.
This is far from lockless, and interrupts still need to be disabled,
but it is a step towards a more lockless solution, and it also
solves a lot of the issues that were noticed by the first conversion
of ftrace to the ring buffers.
Note: the ring_buffer_{un}lock API has been removed.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The raw_local_irq_save causes issues with lockdep. We don't need it
so replace them with local_irq_save.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adapts the boot tracer to the new type of the
print_line callback.
It still relays entries it doesn't support to default output
functions.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Pekka Paalanen <pq@iki.fi>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Adapt mmiotrace to the new print_line type.
By default, it ignores (and consumes) types it doesn't support.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Pekka Paalanen <pq@iki.fi>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch fixes a bug which break the pipe when the seq is empty.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
We need a kind of disambiguation when a print_line callback
returns 0.
_There is not enough space to print all the entry.
Please flush the seq and retry.
_I can't handle this type of entry
This patch changes the type of this callback for better information.
Also some changes have been made in this V2.
_ Only relay to default functions after the print_line callback fails.
_ This patch doesn't fix the issue with the broken pipe (see patch 2/4 for that)
Some things are still in discussion:
_ Find better names for the enum print_line_t values
_ Change the type of print_trace_line into boolean.
Patches to change that can be sent later.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Pekka Paalanen <pq@iki.fi>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Now that the underlining ring buffer for ftrace now hold variable length
entries, we can take advantage of this by only storing the size of the
actual event into the buffer. This happens to increase the number of
entries in the buffer dramatically.
We can also get rid of the "trace_cont" operation, but I'm keeping that
until we have no more users. Some of the ftrace tracers can now change
their code to adapt to this new feature.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Mathieu Desnoyers pointed out that the freeing of the page frame needs
to be reset otherwise we might trigger BUG_ON in the page free code.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If for some strange reason the buffer_page gets bigger, or the page struct
gets smaller, I want to know this ASAP. The best way is to not let the
kernel compile.
This patch adds code to test the size of the struct buffer_page against the
page struct and will cause compile issues if the buffer_page ever gets bigger
than the page struct.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is a unified tracing buffer that implements a ring buffer that
hopefully everyone will eventually be able to use.
The events recorded into the buffer have the following structure:
struct ring_buffer_event {
u32 type:2, len:3, time_delta:27;
u32 array[];
};
The minimum size of an event is 8 bytes. All events are 4 byte
aligned inside the buffer.
There are 4 types (all internal use for the ring buffer, only
the data type is exported to the interface users).
RINGBUF_TYPE_PADDING: this type is used to note extra space at the end
of a buffer page.
RINGBUF_TYPE_TIME_EXTENT: This type is used when the time between events
is greater than the 27 bit delta can hold. We add another
32 bits, and record that in its own event (8 byte size).
RINGBUF_TYPE_TIME_STAMP: (Not implemented yet). This will hold data to
help keep the buffer timestamps in sync.
RINGBUF_TYPE_DATA: The event actually holds user data.
The "len" field is only three bits. Since the data must be
4 byte aligned, this field is shifted left by 2, giving a
max length of 28 bytes. If the data load is greater than 28
bytes, the first array field holds the full length of the
data load and the len field is set to zero.
Example, data size of 7 bytes:
type = RINGBUF_TYPE_DATA
len = 2
time_delta: <time-stamp> - <prev_event-time-stamp>
array[0..1]: <7 bytes of data> <1 byte empty>
This event is saved in 12 bytes of the buffer.
An event with 82 bytes of data:
type = RINGBUF_TYPE_DATA
len = 0
time_delta: <time-stamp> - <prev_event-time-stamp>
array[0]: 84 (Note the alignment)
array[1..14]: <82 bytes of data> <2 bytes empty>
The above event is saved in 92 bytes (if my math is correct).
82 bytes of data, 2 bytes empty, 4 byte header, 4 byte length.
Do not reference the above event struct directly. Use the following
functions to gain access to the event table, since the
ring_buffer_event structure may change in the future.
ring_buffer_event_length(event): get the length of the event.
This is the size of the memory used to record this
event, and not the size of the data pay load.
ring_buffer_time_delta(event): get the time delta of the event
This returns the delta time stamp since the last event.
Note: Even though this is in the header, there should
be no reason to access this directly, accept
for debugging.
ring_buffer_event_data(event): get the data from the event
This is the function to use to get the actual data
from the event. Note, it is only a pointer to the
data inside the buffer. This data must be copied to
another location otherwise you risk it being written
over in the buffer.
ring_buffer_lock: A way to lock the entire buffer.
ring_buffer_unlock: unlock the buffer.
ring_buffer_alloc: create a new ring buffer. Can choose between
overwrite or consumer/producer mode. Overwrite will
overwrite old data, where as consumer producer will
throw away new data if the consumer catches up with the
producer. The consumer/producer is the default.
ring_buffer_free: free the ring buffer.
ring_buffer_resize: resize the buffer. Changes the size of each cpu
buffer. Note, it is up to the caller to provide that
the buffer is not being used while this is happening.
This requirement may go away but do not count on it.
ring_buffer_lock_reserve: locks the ring buffer and allocates an
entry on the buffer to write to.
ring_buffer_unlock_commit: unlocks the ring buffer and commits it to
the buffer.
ring_buffer_write: writes some data into the ring buffer.
ring_buffer_peek: Look at a next item in the cpu buffer.
ring_buffer_consume: get the next item in the cpu buffer and
consume it. That is, this function increments the head
pointer.
ring_buffer_read_start: Start an iterator of a cpu buffer.
For now, this disables the cpu buffer, until you issue
a finish. This is just because we do not want the iterator
to be overwritten. This restriction may change in the future.
But note, this is used for static reading of a buffer which
is usually done "after" a trace. Live readings would want
to use the ring_buffer_consume above, which will not
disable the ring buffer.
ring_buffer_read_finish: Finishes the read iterator and reenables
the ring buffer.
ring_buffer_iter_peek: Look at the next item in the cpu iterator.
ring_buffer_read: Read the iterator and increment it.
ring_buffer_iter_reset: Reset the iterator to point to the beginning
of the cpu buffer.
ring_buffer_iter_empty: Returns true if the iterator is at the end
of the cpu buffer.
ring_buffer_size: returns the size in bytes of each cpu buffer.
Note, the real size is this times the number of CPUs.
ring_buffer_reset_cpu: Sets the cpu buffer to empty
ring_buffer_reset: sets all cpu buffers to empty
ring_buffer_swap_cpu: swaps a cpu buffer from one buffer with a
cpu buffer of another buffer. This is handy when you
want to take a snap shot of a running trace on just one
cpu. Having a backup buffer, to swap with facilitates this.
Ftrace max latencies use this.
ring_buffer_empty: Returns true if the ring buffer is empty.
ring_buffer_empty_cpu: Returns true if the cpu buffer is empty.
ring_buffer_record_disable: disable all cpu buffers (read only)
ring_buffer_record_disable_cpu: disable a single cpu buffer (read only)
ring_buffer_record_enable: enable all cpu buffers.
ring_buffer_record_enabl_cpu: enable a single cpu buffer.
ring_buffer_entries: The number of entries in a ring buffer.
ring_buffer_overruns: The number of entries removed due to writing wrap.
ring_buffer_time_stamp: Get the time stamp used by the ring buffer
ring_buffer_normalize_time_stamp: normalize the ring buffer time stamp
into nanosecs.
I still need to implement the GTOD feature. But we need support from
the cpu frequency infrastructure. But this can be done at a later
time without affecting the ring buffer interface.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
It is possible that the testing thread in the ftrace wakeup test does not
run before we stop the trace. This will cause the trace to fail since nothing
will be in the buffers.
This patch adds a small wait in the wakeup test to allow for the woken task
to run and be traced.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When the boot tracer can't handle an entry output, it returns 1.
It should return 0 to relay on other output functions.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The tracing engine resets the ring buffer and the tracers touch it
too during self-tests. These self-tests happen during tracers registering
and work against boot tracing which is logging initcalls.
We have to disable tracing self-tests if the boot-tracer is selected.
Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Bring the entry to choose the boot tracer on the kernel config.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The tracing engine have now to be init in early_initcall to set the
boot tracer. Only the debugfs settings will be initialized at
fs_initcall time.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add the boot/initcall tracer.
It's primary purpose is to be able to trace the initcalls.
It is intended to be used with scripts/bootgraph.pl after some small
improvements.
Note that it is not active after its init. To avoid tracing (and so
crashing) before the whole tracing engine init, you have to explicitly
call start_boot_trace() after do_pre_smp_initcalls() to enable it.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
bit-field is not thread-safe nor smp-safe.
struct marker_entry.rcu_pending is not protected by any lock
in rcu-callback free_old_closure().
so we must turn it into a safe type.
detail:
I suppose rcu_pending and ptype are store in struct marker_entry.tmp1
free_old_closure() side: change ptype side:
| load struct marker_entry.tmp1
--------------------------------|--------------------------------
| change ptype bit in tmp1
load struct marker_entry.tmp1 |
change rcu_pending bit in tmp1 |
store tmp1 |
--------------------------------|--------------------------------
| store tmp1
now this result equals that free_old_closure() do not change rcu_pending
bit, bug! This bug will cause redundant rcu_barrier_sched() called.
not too harmful.
----- corresponding:
free_old_closure() side: change ptype side:
load struct marker_entry.tmp1 |
--------------------------------|--------------------------------
| load struct marker_entry.tmp1
change rcu_pending bit in tmp1 |
| change ptype bit in tmp1
| store tmp1
--------------------------------|--------------------------------
store tmp1 |
now this result equals that change ptype side do not change ptype
bit, bug! this bug cause marker_probe_cb() access to invalid memory.
oops!
see also: http://en.wikipedia.org/wiki/Bit_field
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
when the second, third... probe is registered, its format is
not checked, this patch fixes it.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Lai Jiangshan discovered a reentrancy issue with markers and fixed it by
adding synchronize_sched() calls at each registration/unregistraiton.
It works, but it removes the ability to do batch
registration/unregistration and can cause registration of ~100 markers
to take about 30 seconds on a loaded machine (synchronize_sched() is
much slower on such workloads).
This patch implements a version of the fix which won't slow down marker batch
registration/unregistration. It also go back to the original non-synchronized
reg/unreg.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Use the new rcu_read_lock_sched/unlock_sched() in marker code around the call
site instead of preempt_disable/enable(). It helps reviewing the code more
easily.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The tracepoints had the same problem markers did have wrt reentrancy. Apply a
similar fix using a rcu_barrier after each tracepoint mutex lock.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
unregister bug:
codes using makers are typically calling marker_probe_unregister()
and then destroying the data that marker_probe_func needs(or
unloading this module). This is bug when the corresponding
marker_probe_func is still running(on other cpus),
it is using the destroying/ed data.
we should call synchronize_sched() after marker_update_probes().
reenter bug:
marker_probe_register(), marker_probe_unregister() and
marker_probe_unregister_private_data() are not reentrant safe
functions. these 3 functions release markers_mutex and then
require it again and do "entry->oldptr = old; ...", but entry->oldptr
maybe is using now for these 3 functions may reenter when markers_mutex
is released.
we use synchronize_sched() instead of call_rcu_sched() to fix
this bug. actually we can do:
"
if (entry->rcu_pending)
rcu_barrier_sched();
"
after require markers_mutex again. but synchronize_sched()
is better and simpler. For these 3 functions are not critical path.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
With the recent updates to ftrace, there should not be any failures when
modifying the code. If there is, then we need to warn about it.
This patch has a cleaned up version of the code that I used to discover
that the weak symbols were causing failures.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Replace "none" tracer by the recently created "nop" tracer.
Both are pretty similar except that nop accepts TRACE_PRINT
or TRACE_SPECIAL entries.
And as a consequence, changing the size of the ring buffer now
requires that tracing has already been disabled.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Now that the nop tracer is used as the default tracer by
replacing the "none" tracer, tracing engine depends on it.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If nop tracer is selected, some old entries from the previous tracer
could still be enqueued. Tracing have to be reset.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The functions are already 'extern' anyway, so there's no problem
with linkage. Removing these ifdefs also helps find any potential
compiler errors.
Suggested by Andrew Morton.
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When CONFIG_DYNAMIC_FTRACE isn't used, neither is mcount_addr. This
patch eliminates that warning.
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
A no-op tracer which can serve two purposes:
1. A template for development of a new tracer.
2. A convenient way to see ftrace_printk() calls without
an irrelevant trace making the output messy.
[ mingo@elte.hu: resolved conflicts ]
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Allow a user to inject a marker (TRACE_PRINT entry) into the trace ring
buffer. The related file operations are derived from code by Frédéric
Weisbecker <fweisbec@gmail.com>.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Also make trace_seq_print_cont() non-static, and add a newline if the
seq buffer can't hold all data.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Offer mmiotrace users a function to inject markers from inside the kernel.
This depends on the trace_vprintk() patch.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
trace_vprintk() for easier implementation of tracer specific *_printk
functions. Add check check for no_tracer, and implement
__ftrace_printk() as a wrapper.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Moves the mmiotrace specific functions from trace.c to
trace_mmiotrace.c. Functions trace_wake_up(), tracing_get_trace_entry(),
and tracing_generic_entry_update() are therefore made available outside
trace.c.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This must be brown paper bag week for Steven Rostedt!
While working on ftrace for PPC, I discovered that the hash locking done
when CONFIG_FTRACE_MCOUNT_RECORD is not set, is totally incorrect.
With a cut and paste error, I had the hash lock macro to lock for both
hash_lock _and_ hash_unlock!
This bug did not affect x86 since this bug was introduced when
CONFIG_FTRACE_MCOUNT_RECORD was added to x86.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
make most of the tracers depend on DEBUG_KERNEL - that's their intended
purpose. (most distributions have DEBUG_KERNEL enabled anyway so this is
not a practical limitation - but it simplifies the tracing menu in the
normal case)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While profiling the smp behaviour of the scheduler it was needed to know to
which cpu a task got woken.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently ftrace_printk only works with the ftrace tracer, switch it to an
iter_ctrl setting so we can make us of them with other tracers too.
[rostedt@redhat.com: tweak to the disable condition]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
An item in the trace buffer that is bigger than one entry may be split
up using the TRACE_CONT entry. This makes it a virtual single entry.
The current code increments the iterator index even while traversing
TRACE_CONT entries, making it look like the iterator is further than
it actually is.
This patch adds code to not increment the iterator index while skipping
over TRACE_CONT entries.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Peter Zijlstra provided me with a nice brown paper bag while letting me know
that I was doing a logical AND and not a binary one, making a condition
true more often than it should be.
Luckily, a false true is handled by the calling function and no harm is
done. But this needs to be fixed regardless.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently some of the ftrace output goes skewiff if you have more
than 9 cpus, and some if you have more than 99.
Twiddle with the headers and format strings to make up to 999 cpus
display without causing spacing problems.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The mcount record method of ftrace scans objdump for references to mcount.
Using mcount as the reference to test if the calls to mcount being replaced
are indeed calls to mcount, this use of mcount was also caught as a
location to change. Using a variable that points to the mcount address
moves this reference into the data section that is not scanned, and
we do not use a false location to try and modify.
The warn on code was what was used to detect this bug.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is another tracer using the ftrace infrastructure, that examines
at each function call the size of the stack. If the stack use is greater
than the previous max it is recorded.
You can always see (and set) the max stack size seen. By setting it
to zero will start the recording again. The backtrace is also available.
For example:
# cat /debug/tracing/stack_max_size
1856
# cat /debug/tracing/stack_trace
[<c027764d>] stack_trace_call+0x8f/0x101
[<c021b966>] ftrace_call+0x5/0x8
[<c02553cc>] clocksource_get_next+0x12/0x48
[<c02542a5>] update_wall_time+0x538/0x6d1
[<c0245913>] do_timer+0x23/0xb0
[<c0257657>] tick_do_update_jiffies64+0xd9/0xf1
[<c02576b9>] tick_sched_timer+0x4a/0xad
[<c0250fe6>] __run_hrtimer+0x3e/0x75
[<c02518ed>] hrtimer_interrupt+0xf1/0x154
[<c022c870>] smp_apic_timer_interrupt+0x71/0x84
[<c021b7e9>] apic_timer_interrupt+0x2d/0x34
[<c0238597>] finish_task_switch+0x29/0xa0
[<c05abd13>] schedule+0x765/0x7be
[<c05abfca>] schedule_timeout+0x1b/0x90
[<c05ab4d4>] wait_for_common+0xab/0x101
[<c05ab5ac>] wait_for_completion+0x12/0x14
[<c033cfc3>] blk_execute_rq+0x84/0x99
[<c0402470>] scsi_execute+0xc2/0x105
[<c040250a>] scsi_execute_req+0x57/0x7f
[<c043afe0>] sr_test_unit_ready+0x3e/0x97
[<c043bbd6>] sr_media_change+0x43/0x205
[<c046b59f>] media_changed+0x48/0x77
[<c046b5ff>] cdrom_media_changed+0x31/0x37
[<c043b091>] sr_block_media_changed+0x16/0x18
[<c02b9e69>] check_disk_change+0x1b/0x63
[<c046f4c3>] cdrom_open+0x7a1/0x806
[<c043b148>] sr_block_open+0x78/0x8d
[<c02ba4c0>] do_open+0x90/0x257
[<c02ba869>] blkdev_open+0x2d/0x56
[<c0296a1f>] __dentry_open+0x14d/0x23c
[<c0296b32>] nameidata_to_filp+0x24/0x38
[<c02a1c68>] do_filp_open+0x347/0x626
[<c02967ef>] do_sys_open+0x47/0xbc
[<c02968b0>] sys_open+0x23/0x2b
[<c021aadd>] sysenter_do_call+0x12/0x26
I've tested this on both x86_64 and i386.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
After disabling FTRACE_MCOUNT_RECORD via a patch, a dormant build
failure surfaced:
kernel/trace/ftrace.c: In function 'ftrace_record_ip':
kernel/trace/ftrace.c:416: error: incompatible type for argument 1 of '_spin_lock_irqsave'
kernel/trace/ftrace.c:433: error: incompatible type for argument 1 of '_spin_lock_irqsave'
Introduced by commit 6dad8e07f4c10b17b038e84d29f3ca41c2e55cd0 ("ftrace:
add necessary locking for ftrace records").
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The new design of pre-recorded mcounts and updating the code outside of
kstop_machine has changed the way the records themselves are protected.
This patch uses the ftrace_lock to protect the records. Note, the lock
still does not need to be taken within calls that are only called via
kstop_machine, since the that code can not run while the spin lock is held.
Also removed the hash_lock needed for the daemon when MCOUNT_RECORD is
configured. Also did a slight cleanup of an unused variable.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If one of the self tests of ftrace has disabled the function tracer,
do not run the code to convert the mcount calls in modules.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch fixes some mistakes on the tracer in warning messages when
debugfs fails to create tracing files.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: srostedt@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
At OLS I had a lot of interest to be able to have the ftrace buffers
dumped on panic. Usually one would expect to uses kexec and examine
the buffers after a new kernel is loaded. But sometimes the resources
do not permit kdump and kexec, so having an option to still see the
sequence of events up to the crash is very advantageous.
This patch adds the option to have the ftrace buffers dumped to the
console in the latency_trace format on a panic. When the option is set,
the default entries per CPU buffer are lowered to 16384, since the writing
to the serial (if that is the console) may take an awful long time
otherwise.
[
Changes since -v1:
Got alpine to send correctly (as well as spell check working).
Removed config option.
Moved the static variables into ftrace_dump itself.
Gave printk a log level.
]
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs
with the ftrace_printk macro that should be used. Not with the
__ftrace_printk internal function.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds a feature that can help kernel developers debug their
code using ftrace.
int ftrace_printk(const char *fmt, ...);
This records into the ftrace buffer using printf formatting. The entry
size in the buffers are still a fixed length. A new type has been added
that allows for more entries to be used for a single recording.
The start of the print is still the same as the other entries.
It returns the number of characters written to the ftrace buffer.
For example:
Having a module with the following code:
static int __init ftrace_print_test(void)
{
ftrace_printk("jiffies are %ld\n", jiffies);
return 0;
}
Gives me:
insmod-5441 3...1 7569us : ftrace_print_test: jiffies are 4296626666
for the latency_trace file and:
insmod-5441 [03] 1959.370498: ftrace_print_test jiffies are 4296626666
for the trace file.
Note: Only the infrastructure should go into the kernel. It is to help
facilitate debugging for other kernel developers. Calls to ftrace_printk
is not intended to be left in the kernel, and should be frowned upon just
like scattering printks around in the code.
But having this easily at your fingertips helps the debugging go faster
and bugs be solved quicker.
Maybe later on, we can hook this with markers and have their printf format
be sucked into ftrace output.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Some tracers will need to work with more than one entry. In order to do this
the trace_entry structure was split into two fields. One for the start of
all entries, and one to continue an existing entry.
The trace_entry structure now has a "field" entry that consists of the previous
content of the trace_entry, and a "cont" entry that is just a string buffer
the size of the "field" entry.
Thanks to Andrew Morton for suggesting this idea.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When a mcount pointer is recorded into a table, it is used to add or
remove calls to mcount (replacing them with nops). If the code is removed
via removing a module, the pointers still exist. At modifying the code
a check is always made to make sure the code being replaced is the code
expected. In-other-words, the code being replaced is compared to what
it is expected to be before being replaced.
There is a very small chance that the code being replaced just happens
to look like code that calls mcount (very small since the call to mcount
is relative). To remove this chance, this patch adds ftrace_release to
allow module unloading to remove the pointers to mcount within the module.
Another change for init calls is made to not trace calls marked with
__init. The tracing can not be started until after init is done anyway.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Seems that freed records can appear in the available_filter_functions list.
This patch fixes that.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch enables the loading of the __mcount_section of modules and
changing all the callers of mcount into nops.
The modification is done before the init_module function is called, so
again, we do not need to use kstop_machine to make these changes.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is the infrastructure to the converting the mcount call sites
recorded by the __mcount_loc section into nops on boot. It also allows
for using these sites to enable tracing as normal. When the __mcount_loc
section is used, the "ftraced" kernel thread is disabled.
This uses the current infrastructure to record the mcount call sites
as well as convert them to nops. The mcount function is kept as a stub
on boot up and not converted to the ftrace_record_ip function. We use the
ftrace_record_ip to only record from the table.
This patch does not handle modules. That comes with a later patch.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch creates a section in the kernel called "__mcount_loc".
This will hold a list of pointers to the mcount relocation for
each call site of mcount.
For example:
objdump -dr init/main.o
[...]
Disassembly of section .text:
0000000000000000 <do_one_initcall>:
0: 55 push %rbp
[...]
000000000000017b <init_post>:
17b: 55 push %rbp
17c: 48 89 e5 mov %rsp,%rbp
17f: 53 push %rbx
180: 48 83 ec 08 sub $0x8,%rsp
184: e8 00 00 00 00 callq 189 <init_post+0xe>
185: R_X86_64_PC32 mcount+0xfffffffffffffffc
[...]
We will add a section to point to each function call.
.section __mcount_loc,"a",@progbits
[...]
.quad .text + 0x185
[...]
The offset to of the mcount call site in init_post is an offset from
the start of the section, and not the start of the function init_post.
The mcount relocation is at the call site 0x185 from the start of the
.text section.
.text + 0x185 == init_post + 0xa
We need a way to add this __mcount_loc section in a way that we do not
lose the relocations after final link. The .text section here will
be attached to all other .text sections after final link and the
offsets will be meaningless. We need to keep track of where these
.text sections are.
To do this, we use the start of the first function in the section.
do_one_initcall. We can make a tmp.s file with this function as a reference
to the start of the .text section.
.section __mcount_loc,"a",@progbits
[...]
.quad do_one_initcall + 0x185
[...]
Then we can compile the tmp.s into a tmp.o
gcc -c tmp.s -o tmp.o
And link it into back into main.o.
ld -r main.o tmp.o -o tmp_main.o
mv tmp_main.o main.o
But we have a problem. What happens if the first function in a section
is not exported, and is a static function. The linker will not let
the tmp.o use it. This case exists in main.o as well.
Disassembly of section .init.text:
0000000000000000 <set_reset_devices>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: e8 00 00 00 00 callq 9 <set_reset_devices+0x9>
5: R_X86_64_PC32 mcount+0xfffffffffffffffc
The first function in .init.text is a static function.
00000000000000a8 t __setup_set_reset_devices
000000000000105f t __setup_str_set_reset_devices
0000000000000000 t set_reset_devices
The lowercase 't' means that set_reset_devices is local and is not exported.
If we simply try to link the tmp.o with the set_reset_devices we end
up with two symbols: one local and one global.
.section __mcount_loc,"a",@progbits
.quad set_reset_devices + 0x10
00000000000000a8 t __setup_set_reset_devices
000000000000105f t __setup_str_set_reset_devices
0000000000000000 t set_reset_devices
U set_reset_devices
We still have an undefined reference to set_reset_devices, and if we try
to compile the kernel, we will end up with an undefined reference to
set_reset_devices, or even worst, it could be exported someplace else,
and then we will have a reference to the wrong location.
To handle this case, we make an intermediate step using objcopy.
We convert set_reset_devices into a global exported symbol before linking
it with tmp.o and set it back afterwards.
00000000000000a8 t __setup_set_reset_devices
000000000000105f t __setup_str_set_reset_devices
0000000000000000 T set_reset_devices
00000000000000a8 t __setup_set_reset_devices
000000000000105f t __setup_str_set_reset_devices
0000000000000000 T set_reset_devices
00000000000000a8 t __setup_set_reset_devices
000000000000105f t __setup_str_set_reset_devices
0000000000000000 t set_reset_devices
Now we have a section in main.o called __mcount_loc that we can place
somewhere in the kernel using vmlinux.ld.S and access it to convert
all these locations that call mcount into nops before starting SMP
and thus, eliminating the need to do this with kstop_machine.
Note, A well documented perl script (scripts/recordmcount.pl) is used
to do all this in one location.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kprobes already has an extensive list of annotations for functions
that should not be instrumented. Add notrace annotations to these
functions as well.
This is particularly useful for functions called by the NMI path.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
do not expose users to CONFIG_TRACEPOINTS - tracers can select it
just fine.
update ftrace to select CONFIG_TRACEPOINTS.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Instrument the scheduler activity (sched_switch, migration, wakeups,
wait for a task, signal delivery) and process/thread
creation/destruction (fork, exit, kthread stop). Actually, kthread
creation is not instrumented in this patch because it is architecture
dependent. It allows to connect tracers such as ftrace which detects
scheduling latencies, good/bad scheduler decisions. Tools like LTTng can
export this scheduler information along with instrumentation of the rest
of the kernel activity to perform post-mortem analysis on the scheduler
activity.
About the performance impact of tracepoints (which is comparable to
markers), even without immediate values optimizations, tests done by
Hideo Aoki on ia64 show no regression. His test case was using hackbench
on a kernel where scheduler instrumentation (about 5 events in code
scheduler code) was added. See the "Tracepoints" patch header for
performance result detail.
Changelog :
- Change instrumentation location and parameter to match ftrace
instrumentation, previously done with kernel markers.
[ mingo@elte.hu: conflict resolutions ]
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: 'Peter Zijlstra' <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Implementation of kernel tracepoints. Inspired from the Linux Kernel
Markers. Allows complete typing verification by declaring both tracing
statement inline functions and probe registration/unregistration static
inline functions within the same macro "DEFINE_TRACE". No format string
is required. See the tracepoint Documentation and Samples patches for
usage examples.
Taken from the documentation patch :
"A tracepoint placed in code provides a hook to call a function (probe)
that you can provide at runtime. A tracepoint can be "on" (a probe is
connected to it) or "off" (no probe is attached). When a tracepoint is
"off" it has no effect, except for adding a tiny time penalty (checking
a condition for a branch) and space penalty (adding a few bytes for the
function call at the end of the instrumented function and adds a data
structure in a separate section). When a tracepoint is "on", the
function you provide is called each time the tracepoint is executed, in
the execution context of the caller. When the function provided ends its
execution, it returns to the caller (continuing from the tracepoint
site).
You can put tracepoints at important locations in the code. They are
lightweight hooks that can pass an arbitrary number of parameters, which
prototypes are described in a tracepoint declaration placed in a header
file."
Addition and removal of tracepoints is synchronized by RCU using the
scheduler (and preempt_disable) as guarantees to find a quiescent state
(this is really RCU "classic"). The update side uses rcu_barrier_sched()
with call_rcu_sched() and the read/execute side uses
"preempt_disable()/preempt_enable()".
We make sure the previous array containing probes, which has been
scheduled for deletion by the rcu callback, is indeed freed before we
proceed to the next update. It therefore limits the rate of modification
of a single tracepoint to one update per RCU period. The objective here
is to permit fast batch add/removal of probes on _different_
tracepoints.
Changelog :
- Use #name ":" #proto as string to identify the tracepoint in the
tracepoint table. This will make sure not type mismatch happens due to
connexion of a probe with the wrong type to a tracepoint declared with
the same name in a different header.
- Add tracepoint_entry_free_old.
- Change __TO_TRACE to get rid of the 'i' iterator.
Masami Hiramatsu <mhiramat@redhat.com> :
Tested on x86-64.
Performance impact of a tracepoint : same as markers, except that it
adds about 70 bytes of instructions in an unlikely branch of each
instrumented function (the for loop, the stack setup and the function
call). It currently adds a memory read, a test and a conditional branch
at the instrumentation site (in the hot path). Immediate values will
eventually change this into a load immediate, test and branch, which
removes the memory read which will make the i-cache impact smaller
(changing the memory read for a load immediate removes 3-4 bytes per
site on x86_32 (depending on mov prefixes), or 7-8 bytes on x86_64, it
also saves the d-cache hit).
About the performance impact of tracepoints (which is comparable to
markers), even without immediate values optimizations, tests done by
Hideo Aoki on ia64 show no regression. His test case was using hackbench
on a kernel where scheduler instrumentation (about 5 events in code
scheduler code) was added.
Quoting Hideo Aoki about Markers :
I evaluated overhead of kernel marker using linux-2.6-sched-fixes git
tree, which includes several markers for LTTng, using an ia64 server.
While the immediate trace mark feature isn't implemented on ia64, there
is no major performance regression. So, I think that we don't have any
issues to propose merging marker point patches into Linus's tree from
the viewpoint of performance impact.
I prepared two kernels to evaluate. The first one was compiled without
CONFIG_MARKERS. The second one was enabled CONFIG_MARKERS.
I downloaded the original hackbench from the following URL:
http://devresources.linux-foundation.org/craiger/hackbench/src/hackbench.c
I ran hackbench 5 times in each condition and calculated the average and
difference between the kernels.
The parameter of hackbench: every 50 from 50 to 800
The number of CPUs of the server: 2, 4, and 8
Below is the results. As you can see, major performance regression
wasn't found in any case. Even if number of processes increases,
differences between marker-enabled kernel and marker- disabled kernel
doesn't increase. Moreover, if number of CPUs increases, the differences
doesn't increase either.
Curiously, marker-enabled kernel is better than marker-disabled kernel
in more than half cases, although I guess it comes from the difference
of memory access pattern.
* 2 CPUs
Number of | without | with | diff | diff |
processes | Marker [Sec] | Marker [Sec] | [Sec] | [%] |
--------------------------------------------------------------
50 | 4.811 | 4.872 | +0.061 | +1.27 |
100 | 9.854 | 10.309 | +0.454 | +4.61 |
150 | 15.602 | 15.040 | -0.562 | -3.6 |
200 | 20.489 | 20.380 | -0.109 | -0.53 |
250 | 25.798 | 25.652 | -0.146 | -0.56 |
300 | 31.260 | 30.797 | -0.463 | -1.48 |
350 | 36.121 | 35.770 | -0.351 | -0.97 |
400 | 42.288 | 42.102 | -0.186 | -0.44 |
450 | 47.778 | 47.253 | -0.526 | -1.1 |
500 | 51.953 | 52.278 | +0.325 | +0.63 |
550 | 58.401 | 57.700 | -0.701 | -1.2 |
600 | 63.334 | 63.222 | -0.112 | -0.18 |
650 | 68.816 | 68.511 | -0.306 | -0.44 |
700 | 74.667 | 74.088 | -0.579 | -0.78 |
750 | 78.612 | 79.582 | +0.970 | +1.23 |
800 | 85.431 | 85.263 | -0.168 | -0.2 |
--------------------------------------------------------------
* 4 CPUs
Number of | without | with | diff | diff |
processes | Marker [Sec] | Marker [Sec] | [Sec] | [%] |
--------------------------------------------------------------
50 | 2.586 | 2.584 | -0.003 | -0.1 |
100 | 5.254 | 5.283 | +0.030 | +0.56 |
150 | 8.012 | 8.074 | +0.061 | +0.76 |
200 | 11.172 | 11.000 | -0.172 | -1.54 |
250 | 13.917 | 14.036 | +0.119 | +0.86 |
300 | 16.905 | 16.543 | -0.362 | -2.14 |
350 | 19.901 | 20.036 | +0.135 | +0.68 |
400 | 22.908 | 23.094 | +0.186 | +0.81 |
450 | 26.273 | 26.101 | -0.172 | -0.66 |
500 | 29.554 | 29.092 | -0.461 | -1.56 |
550 | 32.377 | 32.274 | -0.103 | -0.32 |
600 | 35.855 | 35.322 | -0.533 | -1.49 |
650 | 39.192 | 38.388 | -0.804 | -2.05 |
700 | 41.744 | 41.719 | -0.025 | -0.06 |
750 | 45.016 | 44.496 | -0.520 | -1.16 |
800 | 48.212 | 47.603 | -0.609 | -1.26 |
--------------------------------------------------------------
* 8 CPUs
Number of | without | with | diff | diff |
processes | Marker [Sec] | Marker [Sec] | [Sec] | [%] |
--------------------------------------------------------------
50 | 2.094 | 2.072 | -0.022 | -1.07 |
100 | 4.162 | 4.273 | +0.111 | +2.66 |
150 | 6.485 | 6.540 | +0.055 | +0.84 |
200 | 8.556 | 8.478 | -0.078 | -0.91 |
250 | 10.458 | 10.258 | -0.200 | -1.91 |
300 | 12.425 | 12.750 | +0.325 | +2.62 |
350 | 14.807 | 14.839 | +0.032 | +0.22 |
400 | 16.801 | 16.959 | +0.158 | +0.94 |
450 | 19.478 | 19.009 | -0.470 | -2.41 |
500 | 21.296 | 21.504 | +0.208 | +0.98 |
550 | 23.842 | 23.979 | +0.137 | +0.57 |
600 | 26.309 | 26.111 | -0.198 | -0.75 |
650 | 28.705 | 28.446 | -0.259 | -0.9 |
700 | 31.233 | 31.394 | +0.161 | +0.52 |
750 | 34.064 | 33.720 | -0.344 | -1.01 |
800 | 36.320 | 36.114 | -0.206 | -0.57 |
--------------------------------------------------------------
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: 'Peter Zijlstra' <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'proc' of git://git.kernel.org/pub/scm/linux/kernel/git/adobriyan/proc:
proc: remove kernel.maps_protect
proc: remove now unneeded ADDBUF macro
[PATCH] proc: show personality via /proc/pid/personality
[PATCH] signal, procfs: some lock_task_sighand() users do not need rcu_read_lock()
proc: move PROC_PAGE_MONITOR to fs/proc/Kconfig
proc: make grab_header() static
proc: remove unused get_dma_list()
proc: remove dummy vmcore_open()
proc: proc_sys_root tweak
proc: fix return value of proc_reg_open() in "too late" case
Fixed up trivial conflict in removed file arch/sparc/include/asm/dma_32.h
Various people outside the tty layer still stick their noses in behind the
scenes. We need to make sure they also obey the locking and referencing rules.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is pure tty code so put it in the tty layer where it can be with the
locking relevant material it uses
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Introduce a kref to the tty structure and use it to protect the tty->signal
tty references. For now we don't introduce it for anything else.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There's a small race/chance that, while hrtimers are enabled globally,
they're later not enabled when we're calling the hrtimer_interrupt() function,
which then BUG_ON()'s for that. This patch closes that race/gap.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
* 'x86-v28-for-linus-phase4-D' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (186 commits)
x86, debug: print more information about unknown CPUs
x86 setup: handle more than 8 CPU flag words
x86: cpuid, fix typo
x86: move transmeta cap read to early_init_transmeta()
x86: identify_cpu_without_cpuid v2
x86: extended "flags" to show virtualization HW feature in /proc/cpuinfo
x86: move VMX MSRs to msr-index.h
x86: centaur_64.c remove duplicated setting of CONSTANT_TSC
x86: intel.c put workaround for old cpus together
x86: let intel 64-bit use intel.c
x86: make intel_64.c the same as intel.c
x86: make intel.c have 64-bit support code
x86: little clean up of intel.c/intel_64.c
x86: make 64 bit to use amd.c
x86: make amd_64 have 32 bit code
x86: make amd.c have 64bit support code
x86: merge header in amd_64.c
x86: add srat_detect_node for amd64
x86: remove duplicated force_mwait
x86: cpu make amd.c more like amd_64.c v2
...
* 'x86-v28-for-linus-phase3-B' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (74 commits)
AMD IOMMU: use iommu_device_max_index, fix
AMD IOMMU: use iommu_device_max_index
x86: add PCI IDs for AMD Barcelona PCI devices
x86/iommu: use __GFP_ZERO instead of memset for GART
x86/iommu: convert GART need_flush to bool
x86/iommu: make GART driver checkpatch clean
x86 gart: remove unnecessary initialization
x86: restore old GART alloc_coherent behavior
revert "x86: make GART to respect device's dma_mask about virtual mappings"
x86: export pci-nommu's alloc_coherent
iommu: remove fullflush and nofullflush in IOMMU generic option
x86: remove set_bit_string()
iommu: export iommu_area_reserve helper function
AMD IOMMU: use coherent_dma_mask in alloc_coherent
add AMD IOMMU tree to MAINTAINERS file
AMD IOMMU: use cmd_buf_size when freeing the command buffer
AMD IOMMU: calculate IVHD size with a function
AMD IOMMU: remove unnecessary cast to u64 in the init code
AMD IOMMU: free domain bitmap with its allocation order
AMD IOMMU: simplify dma_mask_to_pages
...
* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
[CPUFREQ] Fix BUG: using smp_processor_id() in preemptible code
[CPUFREQ] Don't export governors for default governor
[CPUFREQ][6/6] cpufreq: Add idle microaccounting in ondemand governor
[CPUFREQ][5/6] cpufreq: Changes to get_cpu_idle_time_us(), used by ondemand governor
[CPUFREQ][4/6] cpufreq_ondemand: Parameterize down differential
[CPUFREQ][3/6] cpufreq: get_cpu_idle_time() changes in ondemand for idle-microaccounting
[CPUFREQ][2/6] cpufreq: Change load calculation in ondemand for software coordination
[CPUFREQ][1/6] cpufreq: Add cpu number parameter to __cpufreq_driver_getavg()
[CPUFREQ] use deferrable delayed work init in conservative governor
[CPUFREQ] drivers/cpufreq/cpufreq.c: Adjust error handling code involving cpufreq_cpu_put
[CPUFREQ] add error handling for cpufreq_register_governor() error
[CPUFREQ] acpi-cpufreq: add error handling for cpufreq_register_driver() error
[CPUFREQ] Coding style fixes to arch/x86/kernel/cpu/cpufreq/powernow-k6.c
[CPUFREQ] Coding style fixes to arch/x86/kernel/cpu/cpufreq/elanfreq.c
We need to add a flag for all code that is in the drivers/staging/
directory to prevent all other kernel developers from worrying about
issues here, and to notify users that the drivers might not be as good
as they are normally used to.
Based on code from Andreas Gruenbacher and Jeff Mahoney to provide a
TAINT flag for the support level of a kernel module in the Novell
enterprise kernel release.
This is the kernel portion of this feature, the ability for the flag to
be set needs to be done in the build process and will happen in a
follow-up patch.
Cc: Andreas Gruenbacher <agruen@suse.de>
Cc: Jeff Mahoney <jeffm@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* 'rcu-v28-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (21 commits)
rcu: RCU-based detection of stalled CPUs for Classic RCU, fix
rcu: RCU-based detection of stalled CPUs for Classic RCU
rcu: add rcu_read_lock_sched() / rcu_read_unlock_sched()
rcu: fix sparse shadowed variable warning
doc/RCU: fix pseudocode in rcuref.txt
rcuclassic: fix compiler warning
rcu: use irq-safe locks
rcuclassic: fix compilation NG
rcu: fix locking cleanup fallout
rcu: remove redundant ACCESS_ONCE definition from rcupreempt.c
rcu: fix classic RCU locking cleanup lockdep problem
rcu: trace fix possible mem-leak
rcu: just rename call_rcu_bh instead of making it a macro
rcu: remove list_for_each_rcu()
rcu: fixes to include/linux/rcupreempt.h
rcu: classic RCU locking and memory-barrier cleanups
rcu: prevent console flood when one CPU sees another AWOL via RCU
rcu, debug: detect stalled grace periods, cleanups
rcu, debug: detect stalled grace periods
rcu classic: new algorithm for callbacks-processing(v2)
...
When sched_clock_cpu() couples the clocks between two cpus, it may
increment scd->clock beyond the GTOD tick window that __update_sched_clock()
uses to clamp the clock. A later call to __update_sched_clock() may move
the clock back to scd->tick_gtod + TICK_NSEC, violating the clock's
monotonic property.
This patch ensures that scd->clock will not be set backward.
Signed-off-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
After commit 831830b5a2 aka
"restrict reading from /proc/<pid>/maps to those who share ->mm or can ptrace"
sysctl stopped being relevant because commit moved security checks from ->show
time to ->start time (mm_for_maps()).
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Kees Cook <kees.cook@canonical.com>
lock_task_sighand() make sure task->sighand is being protected,
so we do not need rcu_read_lock().
[ exec() will get task->sighand->siglock before change task->sighand! ]
But code using rcu_read_lock() _just_ to protect lock_task_sighand()
only appear in procfs. (and some code in procfs use lock_task_sighand()
without such redundant protection.)
Other subsystem may put lock_task_sighand() into rcu_read_lock()
critical region, but these rcu_read_lock() are used for protecting
"for_each_process()", "find_task_by_vpid()" etc. , not for protecting
lock_task_sighand().
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
[ok from Oleg]
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
export get_cpu_idle_time_us() for it to be used in ondemand governor.
Last update time can be current time when the CPU is currently non-idle,
accounting for the busy time since last idle.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dave Jones <davej@redhat.com>
add /proc/sys/kernel/sched_domain/cpu0/domain0/name, to make
it easier to see which specific scheduler domain remained at
that entry.
Since we process the scheduler domain tree and
simplify it, it's not always immediately clear during debugging
which domain came from where.
depends on CONFIG_SCHED_DEBUG=y.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While looking at the code I wondered why we always do:
sync && avg_overlap < migration_cost
Which is a bit odd, since the overlap test was meant to detect sync wakeups
so using it to specialize sync wakeups doesn't make much sense.
Hence change the code to do:
sync || avg_overlap < migration_cost
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The softlockup watchdog needs to be touched when resuming the from the
kgdb stopped state to avoid the printk that a CPU is stuck if the
debugger was active for longer than the softlockup threshold.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
css will be initialized by cgroup core.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While working on the new version of the code for SCHED_SPORADIC I
noticed something strange in the present throttling mechanism. More
specifically in the throttling timer handler in sched_rt.c
(do_sched_rt_period_timer()) and in rt_rq_enqueue().
The problem is that, when unthrottling a runqueue, rt_rq_enqueue() only
asks for rescheduling if the runqueue has a sched_entity associated to
it (i.e., rt_rq->rt_se != NULL).
Now, if the runqueue is the root rq (which has a rt_se = NULL)
rescheduling does not take place, and it is delayed to some undefined
instant in the future.
This imply some random bandwidth usage by the RT tasks under throttling.
For instance, setting rt_runtime_us/rt_period_us = 950ms/1000ms an RT
task will get less than 95%. In our tests we got something varying
between 70% to 95%.
Using smaller time values, e.g., 95ms/100ms, things are even worse, and
I can see values also going down to 20-25%!!
The tests we performed are simply running 'yes' as a SCHED_FIFO task,
and checking the CPU usage with top, but we can investigate thoroughly
if you think it is needed.
Things go much better, for us, with the attached patch... Don't know if
it is the best approach, but it solved the issue for us.
Signed-off-by: Dario Faggioli <raistlin@linux.it>
Signed-off-by: Michael Trimarchi <trimarchimichael@yahoo.it>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: jiffies increment too fast.
Hugh Dickins noted that with NOHZ=n and HIGHRES=n jiffies get
incremented too fast. The reason is a wrong check in the broadcast
enter/exit code, which keeps the local apic timer in periodic mode
when the switch happens.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
This fixes a warning on latest -tip:
kernel/cpuset.c: Dans la fonction «scan_for_empty_cpusets» :
kernel/cpuset.c:1932: attention : passing argument 1 of «list_add_tail» discards qualifiers from pointer target type
Actually the struct cpuset *root passed in parameter to scan_for_empty_cpusets
is not supposed to be const since an entry is added on the tail of its list.
Just correct the qualifier.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Last -tip gives this warning:
kernel/softirq.c: Dans la fonction «__do_softirq» :
kernel/softirq.c:216: attention : format «%ld» expects type «long int», but argument 2 has type «int»
This patch corrects the format type, and a small mistake in the "softirq" word.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
fix the !CONFIG_RCU_CPU_STALL_DETECTOR path:
kernel/rcuclassic.c: In function '__rcu_pending':
kernel/rcuclassic.c:609: error: too few arguments to function 'check_cpu_stall'
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds stalled-CPU detection to Classic RCU. This capability
is enabled by a new config variable CONFIG_RCU_CPU_STALL_DETECTOR, which
defaults disabled.
This is a debugging feature to detect infinite loops in kernel code, not
something that non-kernel-hackers would be expected to care about.
This feature can detect looping CPUs in !PREEMPT builds and looping CPUs
with preemption disabled in PREEMPT builds. This is essentially a port of
this functionality from the treercu patch, replacing the stall debug patch
that is already in tip/core/rcu (commit 67182ae1c4).
The changes from the patch in tip/core/rcu include making the config
variable name match that in treercu, changing from seconds to jiffies to
avoid spurious warnings, and printing a boot message when this feature
is enabled.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Found by static checker (http://repo.or.cz/w/smatch.git).
Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
if a preempt count leaks out of a softirq handler it can be very hard
to figure it out. Add a debug check for this.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Genirq hasn't previously recorded the trigger type used by any given IRQ,
although some irq_chip support has done so. That data can be useful when
troubleshooting. This patch records it in the relevant irq_desc.status
bits, and improves consistency between the two driver-visible calls
affected:
- Make set_irq_type() usage match request_irq() usage:
* IRQ_TYPE_NONE should be a NOP; succeed, so irq_chip methods
won't have to handle that case any more (many do it wrong).
* IRQ_TYPE_PROBE is ignored; any buggy out-of-tree callers
might need to switch over to the real IRQ probing code.
* emit the same diagnostics (from shared utility code)
- Their kerneldoc now reflects usage:
* request_irq() flags include IRQF_TRIGGER_* to specify
active edge(s)/level ... docs previously omitted that
* set_irq_type() is declared in <linux/irq.h> so callers
should use the (bit-equivalent) IRQ_TYPE_* symbols there
Also: adds a warning about shared IRQs that don't end up using the
requested trigger mode; and fix an unrelated "sparse" warning.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
hrtimer: prevent migration of per CPU hrtimers
hrtimer: mark migration state
hrtimer: fix migration of CB_IRQSAFE_NO_SOFTIRQ hrtimers
hrtimer: migrate pending list on cpu offline
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This patch does following:
o Removes unused variable and argument "rq".
o Optimizes one of the "if" conditions in wake_affine() - i.e. if
"balanced" is true, we need not do rest of the calculations in the
condition.
o If this cpu is same as the previous cpu (on which woken up task
was running when it went to sleep), no need to call wake_affine at all.
Signed-off-by: Amit K Arora <aarora@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds the CONFIG_FILE_LOCKING option which allows to remove
support for advisory locks. With this patch enabled, the flock()
system call, the F_GETLK, F_SETLK and F_SETLKW operations of fcntl()
and NFS support are disabled. These features are not necessarly needed
on embedded systems. It allows to save ~11 Kb of kernel code and data:
text data bss dec hex filename
1125436 118764 212992 1457192 163c28 vmlinux.old
1114299 118564 212992 1445855 160fdf vmlinux
-11137 -200 0 -11337 -2C49 +/-
This patch has originally been written by Matt Mackall
<mpm@selenic.com>, and is part of the Linux Tiny project.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Matt Mackall <mpm@selenic.com>
Cc: matthew@wil.cx
Cc: linux-fsdevel@vger.kernel.org
Cc: mpm@selenic.com
Cc: akpm@linux-foundation.org
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
There's a race between mm->owner assignment and swapoff, more easily
seen when task slab poisoning is turned on. The condition occurs when
try_to_unuse() runs in parallel with an exiting task. A similar race
can occur with callers of get_task_mm(), such as /proc/<pid>/<mmstats>
or ptrace or page migration.
CPU0 CPU1
try_to_unuse
looks at mm = task0->mm
increments mm->mm_users
task 0 exits
mm->owner needs to be updated, but no
new owner is found (mm_users > 1, but
no other task has task->mm = task0->mm)
mm_update_next_owner() leaves
mmput(mm) decrements mm->mm_users
task0 freed
dereferencing mm->owner fails
The fix is to notify the subsystem via mm_owner_changed callback(),
if no new owner is found, by specifying the new task as NULL.
Jiri Slaby:
mm->owner was set to NULL prior to calling cgroup_mm_owner_callbacks(), but
must be set after that, so as not to pass NULL as old owner causing oops.
Daisuke Nishimura:
mm_update_next_owner() may set mm->owner to NULL, but mem_cgroup_from_task()
and its callers need to take account of this situation to avoid oops.
Hugh Dickins:
Lockdep warning and hang below exec_mmap() when testing these patches.
exit_mm() up_reads mmap_sem before calling mm_update_next_owner(),
so exec_mmap() now needs to do the same. And with that repositioning,
there's now no point in mm_need_new_owner() allowing for NULL mm.
Reported-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Impact: per CPU hrtimers can be migrated from a dead CPU
The hrtimer code has no knowledge about per CPU timers, but we need to
prevent the migration of such timers and warn when such a timer is
active at migration time.
Explicitely mark the timers as per CPU and use a more understandable
mode descriptor for the interrupts safe unlocked callback mode, which
is used by hrtimer_sleeper and the scheduler code.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: during migration active hrtimers can be seen as inactive
The migration code removes the hrtimers from the queues of the dead
CPU and sets the state temporary to INACTIVE. The enqueue code sets it
to ACTIVE/PENDING again.
Prevent that the wrong state can be seen by using a separate migration
state bit.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: Stale timers after a CPU went offline.
commit 37bb6cb409
hrtimer: unlock hrtimer_wakeup
changed the hrtimer sleeper callback mode to CB_IRQSAFE_NO_SOFTIRQ due
to locking problems. A result of this change is that when enqueue is
called for an already expired hrtimer the callback function is not
longer called directly from the enqueue code. The normal callers have
been fixed in the code, but the migration code which moves hrtimers
from a dead CPU to a live CPU was not made aware of this.
This can be fixed by checking the timer state after the call to
enqueue in the migration code.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: hrtimers which are on the pending list are not migrated at cpu
offline and can be stale forever
Add the pending list migration when CONFIG_HIGH_RES_TIMERS is enabled
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
- fix UP lockup
- another set of UP/SMP cleanups and simplifications
Signed-off-by: Frank Mayhar <fmayhar@google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
On the x86 arch, user space single step exceptions should be ignored
if they occur in the kernel space, such as ptrace stepping through a
system call.
First check if it is kgdb that is executing a single step, then ensure
it is not an accidental traversal into the user space, while in kgdb,
any other time the TIF_SINGLESTEP is set, kgdb should ignore the
exception.
On x86, arm, mips and powerpc, the kgdb_contthread usage was
inconsistent with the way single stepping is implemented in the kgdb
core. The arch specific stub should always set the
kgdb_cpu_doing_single_step correctly if it is single stepping. This
allows kgdb to correctly process an instruction steps if ptrace
happens to be requesting an instruction step over a system call.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
On the ARM architecture, kgdb will crash the kernel if the last byte
of valid memory is written due to a flush_icache_range flushing
beyond the memory boundary.
Signed-off-by: Atsuo Igarashi <atsuo_igarashi@tripeaks.co.jp>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
fix this build error:
kernel/resource.c: In function 'iomem_map_sanity_check':
kernel/resource.c:842: error: implicit declaration of function 'r_next'
kernel/resource.c:842: warning: assignment makes pointer from integer without a cast
r_next() was only available if CONFIG_PROCFS was enabled.
and fix this build warning:
kernel/resource.c:855: warning: format '%llx' expects type 'long long unsigned int', but argument 2 has type 'resource_size_t'
kernel/resource.c:855: warning: format '%llx' expects type 'long long unsigned int', but argument 3 has type 'long unsigned int'
kernel/resource.c:855: warning: format '%llx' expects type 'long long unsigned int', but argument 4 has type 'resource_size_t'
kernel/resource.c:855: warning: format '%llx' expects type 'long long unsigned int', but argument 5 has type 'resource_size_t'
resource_t can be 32 bits.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Go through the iomem resource tree to check if any of the ioremap()
requests span more than any slot in the iomem resource tree and do
a WARN_ON() if we hit this check.
This will raise a red-flag, if some driver is mapping more than what
is needed. And hopefully identify possible corruptions much earlier.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
cfs_rq->tasks list is used by the load balancer to iterate
over all the tasks. Currently it holds all the entities
(both task and group entities) because of which there is
a need to check for group entities explicitly during load
balancing. This patch changes the cfs_rq->tasks list to
hold only task entities.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Change PPM_SCALE_INV_SHIFT so that it doesn't throw away any input bits
(19 is the amount of the factor 2 in PPM_SCALE), the output frequency
can then be calculated back to its input value, as the inverse divide
produce a slightly larger value, which is then correctly rounded by the
final shift.
Reported-by: Martin Ziegler <ziegler@uni-freiburg.de>
Signed-off-by: Roman Zippel <zippel@linux-m68k.org>
Cc: John Stultz <johnstul@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Due to a rounding problem during a clock update it's possible for readers
to observe the clock jumping back by 1nsec. The following simplified
example demonstrates the problem:
cycle xtime
0 0
1000 999999.6
2000 1999999.2
3000 2999998.8
...
1500 = 1499999.4
= 0.0 + 1499999.4
= 999999.6 + 499999.8
When reading the clock only the full nanosecond part is used, while
timekeeping internally keeps nanosecond fractions. If the clock is now
updated at cycle 1500 here, a nanosecond is missing due to the truncation.
The simple fix is to round up the xtime value during the update, this also
changes the distance to the reference time, but the adjustment will
automatically take care that it stays under control.
Signed-off-by: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
This is a change that makes the 11-minute RTC update be run in the process
context. This is so that update_persistent_clock() can sleep, which may
be required for certain types of RTC hardware -- most notably I2C devices.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: David Brownell <david-b@pacbell.net>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cleanup. Imho makes the code much more understandable. At least this
patch lessens both the source and compiled code.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
lock_timer() checks that the timer found by idr_find(timer_id) has ->it_id
== timer_id. This buys nothing. This check can fail only if
sys_timer_create() unlocked idr_lock after idr_get_new(), but didn't set
->it_id = new_timer_id yet. But in that case ->it_process == NULL so
lock_timer() can't succeed anyway.
Also remove a couple of unneeded typecasts.
Note that with or without this patch we have a small problem.
sys_timer_create() doesn't ensure that the result of setting (say)
->it_sigev_notify must be visible if lock_timer() succeeds.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
With the recent changes ->it_sigev_signo and ->it_sigev_value are only
used in sys_timer_create(), kill them.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cleanup.
- sys_timer_create() is big and complicated. The code above the "out:"
label relies on the fact that "error" must be == 0. This is not very
robust, make the code more explicit. Remove the unneeded initialization
of error.
- If idr_get_new() succeeds (as it normally should), we check the returned
value twice. Move the "-EAGAIN" check under "if (error)".
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
posix_timer_event() always populates timer->sigq with the same numbers,
move this code into sys_timer_create().
Note that with this patch we can kill it_sigev_signo and it_sigev_value.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
- Change the code to do rcu_read_lock() instead of taking tasklist_lock,
it is safe to get_task_struct(p) if p was found under RCU.
However, now we must not use process's sighand/signal, they may be NULL.
We can use current->sighand/signal instead, this "process" must belong
to the current's thread-group.
- Factor out the common code for 2 "if (timer_event_spec)" branches, the
!timer_event_spec case can use current too.
- use spin_lock_irq() instead of _irqsave(), kill "flags".
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
sys_timer_create() return -EINVAL if the target thread has PF_EXITING.
This doesn't really make sense, the sub-thread can die right after unlock.
And in fact, this is just wrong. Without SIGEV_THREAD_ID good_sigevent()
returns ->group_leader, and it is very possible that the leader is already
dead. This is OK, we shouldn't return the error in this case.
Remove this check and the comment. Note that the "process" was found
under tasklist_lock, it must have ->sighand != NULL.
Also, remove a couple of unneeded initializations.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Change the code to get/put timer->it_process regardless of
SIGEV_THREAD_ID. This streamlines the create/destroy paths and allows us
to simplify the usage of exit_itimers() in de_thread().
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
posix_timer_event() drops SIGEV_THREAD_ID and switches to ->group_leader
if send_sigqueue() fails.
This is not very useful and doesn't work reliably. send_sigqueue() can
only fail if ->it_process is dead. But it can die before it dequeues the
SI_TIMER signal, in that case the timer stops anyway.
Remove this code. I guess it was needed a long ago to ensure that the
timer is not destroyed when when its creator thread dies.
Q: perhaps it makes sense to change sys_timer_settime() to return an error
if ->it_process is dead?
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: mingo@elte.hu
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
timers: fix build error in !oneshot case
x86: c1e_idle: don't mark TSC unstable if CPU has invariant TSC
x86: prevent C-states hang on AMD C1E enabled machines
clockevents: prevent mode mismatch on cpu online
clockevents: check broadcast device not tick device
clockevents: prevent stale tick_next_period for onlining CPUs
x86: prevent stale state of c1e_mask across CPU offline/online
clockevents: prevent cpu online to interfere with nohz
A segmentation fault can occur in kimage_add_entry in kexec.c when loading
a kernel image into memory. The fault occurs because a page is requested
by calling kimage_alloc_page with gfp_mask GFP_KERNEL and the function may
actually return a page with gfp_mask GFP_HIGHUSER. The high mem page is
returned because it was swapped with the kernel page due to the kernel
page being a page that will shortly be copied to.
This patch ensures that kimage_alloc_page returns a page that was created
with the correct gfp flags.
I have verified the change and fixed the whitespace damage of the original
patch. Jonathan did a great job of tracking this down after he hit the
problem. -- Eric
Signed-off-by: Jonathan Steel <jon.steel@esentire.com>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We should set the buddy even though we might already have the
TIF_RESCHED flag set.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While playing around with it, I noticed we missed some sanity checks.
Also add some comments while we're there.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
We should not only correct the increment for the initial group, but should
be consistent and do so for all the groups we encounter.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Rework the wakeup preemption to work on real runtime instead of
the virtual runtime. This greatly simplifies the code.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is the second resubmission of the posix timer rework patch, posted
a few days ago.
This includes the changes from the previous resubmittion, which addressed
Oleg Nesterov's comments, removing the RCU stuff from the patch and
un-inlining the thread_group_cputime() function for SMP.
In addition, per Ingo Molnar it simplifies the UP code, consolidating much
of it with the SMP version and depending on lower-level SMP/UP handling to
take care of the differences.
It also cleans up some UP compile errors, moves the scheduler stats-related
macros into kernel/sched_stats.h, cleans up a merge error in
kernel/fork.c and has a few other minor fixes and cleanups as suggested
by Oleg and Ingo. Thanks for the review, guys.
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>
kernel/time/tick-common.c: In function ‘tick_setup_periodic’:
kernel/time/tick-common.c:113: error: implicit declaration of function ‘tick_broadcast_oneshot_active’
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: timer hang on CPU online observed on AMD C1E systems
When a CPU is brought online then the broadcast machinery can
be in the one shot state already. Check this and setup the timer
device of the new CPU in one shot mode so the broadcast code
can pick up the next_event value correctly.
Another AMD C1E oddity, as we switch to broadcast immediately and
not after the full bring up via the ACPI cpu idle code.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: Possible hang on CPU online observed on AMD C1E machines.
The broadcast setup code looks at the mode of the tick device to
determine whether it needs to be shut down or setup. This is wrong
when the broadcast mode is set to one shot already. This can happen
when a CPU is brought online as it goes through the periodic setup
first.
The problem went unnoticed as sane systems do not call into that code
before the switch to one shot for the clock event device happens.
The AMD C1E idle routine switches over immediately and thereby shuts
down the just setup device before the first interrupt happens.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: possible hang on CPU onlining in timer one shot mode.
The tick_next_period variable is only used during boot on nohz/highres
enabled systems, but for CPU onlining it needs to be maintained when
the per cpu clock events device operates in one shot mode.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Impact: rare hang which can be triggered on CPU online.
tick_do_timer_cpu keeps track of the CPU which updates jiffies
via do_timer. The value -1 is used to signal, that currently no
CPU is doing this. There are two cases, where the variable can
have this state:
boot:
necessary for systems where the boot cpu id can be != 0
nohz long idle sleep:
When the CPU which did the jiffies update last goes into
a long idle sleep it drops the update jiffies duty so
another CPU which is not idle can pick it up and keep
jiffies going.
Using the same value for both situations is wrong, as the CPU online
code can see the -1 state when the timer of the newly onlined CPU is
setup. The setup for a newly onlined CPU goes through periodic mode
and can pick up the do_timer duty without being aware of the nohz /
highres mode of the already running system.
Use two separate states and make them constants to avoid magic
numbers confusion.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
kernel/rcuclassic.c:564:18: warning: symbol 'flags' shadows an earlier one
kernel/rcuclassic.c:527:16: originally declared here
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
- Add some comments to try to make the ifdef puzzle a bit clearer
- Explicitly inline one of the three init_hrtick() implementations.
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
LD kernel/built-in.o
WARNING: kernel/built-in.o(.text+0x326): Section mismatch in reference
from the function init_hrtick() to the variable
.cpuinit.data:hotplug_hrtick_nb.8
The function init_hrtick() references
the variable __cpuinitdata hotplug_hrtick_nb.8.
This is often because init_hrtick lacks a __cpuinitdata
annotation or the annotation of hotplug_hrtick_nb.8 is wrong.
Signed-off-by: Md.Rakib H. Mullick <rakib.mullick@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
load_balance_fair() calls rcu_read_lock() but then traverses the list
using the regular list traversal routine. This patch converts the
list traversal to use the _rcu version.
Signed-off-by: Chris Friesen <cfriesen@nortel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Lin Ming reported a 10% OLTP regression against 2.6.27-rc4.
The difference seems to come from different preemption agressiveness,
which affects the cache footprint of the workload and its effective
cache trashing.
Aggresively preempt a task if its avg overlap is very small, this should
avoid the task going to sleep and find it still running when we schedule
back to it - saving a wakeup.
Reported-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Peter Zijlstra noticed this 8 months ago and I just noticed
it again.
hrtimer_clock_base::get_softirq_time() is currently unused
in the entire tree. In fact, looking at the logs, it appears
as if it was never used. Remove it.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
sched: fix deadlock in setting scheduler parameter to zero
sched: fix 2.6.27-rc5 couldn't boot on tulsa machine randomly
* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
clockevents: make device shutdown robust
clocksource, acpi_pm.c: fix check for monotonicity
clockevents: remove WARN_ON which was used to gather information
The device shut down does not cleanup the next_event variable of the
clock event device. So when the device is reactivated the possible
stale next_event value can prevent the device to be reprogrammed as it
claims to wait on a event already.
This is the root cause of the resurfacing suspend/resume problem,
where systems need key press to come back to life.
Fix this by setting next_event to KTIME_MAX when the device is shut
down. Use a separate function for shutdown which takes care of that
and only keep the direct set mode call in the broadcast code, where we
can not touch the next_event value.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
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>
After the patch:
commit 0b2f630a28
Author: Miao Xie <miaox@cn.fujitsu.com>
Date: Fri Jul 25 01:47:21 2008 -0700
cpusets: restructure the function update_cpumask() and update_nodemask()
It might happen that 'echo 0 > /cpuset/sub/cpus' returned failure but 'cpus'
has been changed, because cpus was changed before calling heap_init() which
may return -ENOMEM.
This patch restores the orginal behavior.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Paul Menage <menage@google.com>
Cc: Paul Jackson <pj@sgi.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
As part of going idle, we already look at the time of the next timer event to determine
which C-state to select etc.
This patch adds functionality that causes the timers that are past their
soft expire time, to fire at this time, before we calculate the next wakeup
time. This functionality will thus avoid wakeups by running timers before
going idle rather than specially waking up for it.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
This patch makes the futex() system call use the per process
slack value; with this users are able to externally control existing
applications to reduce the wakeup rate.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
This patch makes the nanosleep() system call use the per process
slack value; with this users are able to externally control existing
applications to reduce the wakeup rate.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
On my tulsa x86-64 machine, kernel 2.6.25-rc5 couldn't boot randomly.
Basically, function __enable_runtime forgets to reset rt_rq->rt_throttled
to 0. When every cpu is up, per-cpu migration_thread is created and it runs
very fast, sometimes to mark the corresponding rt_rq->rt_throttled to 1 very
quickly. After all cpus are up, with below calling chain:
sched_init_smp => arch_init_sched_domains => build_sched_domains => ...
=> cpu_attach_domain => rq_attach_root => set_rq_online => ...
=> _enable_runtime
_enable_runtime is called against every rt_rq again, so rt_rq->rt_time is
reset to 0, but rt_rq->rt_throttled might be still 1. Later on function
do_sched_rt_period_timer couldn't reset it, and all RT tasks couldn't be
scheduled to run on that cpu. here is RT task migration_thread which is
woken up when a task is migrated to another cpu.
Below patch fixes it against 2.6.27-rc5.
Signed-off-by: Zhang Yanmin <yanmin_zhang@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The issue of the endless reprogramming loop due to a too small
min_delta_ns was fixed with the previous updates of the clock events
code, but we had no information about the spread of this problem. I
added a WARN_ON to get automated information via kerneloops.org and to
get some direct reports, which allowed me to analyse the affected
machines.
The WARN_ON has served its purpose and would be annoying for a release
kernel. Remove it and just keep the information about the increase of
the min_delta_ns value.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
sched: arch_reinit_sched_domains() must destroy domains to force rebuild
sched, cpuset: rework sched domains and CPU hotplug handling (v4)
Right now, there is no notifier that is called on a new cpu, before the new
cpu begins processing interrupts/softirqs.
Various kernel function would need that notification, e.g. kvm works around
by calling smp_call_function_single(), rcu polls cpu_online_map.
The patch adds a CPU_STARTING notification. It also adds a helper function
that sends the message to all cpu_chain handlers.
Tested on x86-64.
All other archs are untested. Especially on sparc, I'm not sure if I got
it right.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
to help debugging and visibility of timer ranges, show them
in the existing timer list in /proc/timer_list
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
this patch adds a _range version of hrtimer_start() so that range timers
can be created; the hrtimer_start() function is just a wrapper around this.
In addition, hrtimer_start_expires() will now preserve existing ranges.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
This patch clarifies usage of irq_chip->startup() callback:
1. The "if (startup) startup(); else enabled();" code in setup_irq()
is unnecessary, as startup() falls back to enabled() via
default callbacks, set by irq_chip_set_defaults().
2. When using set_irq_chained_handler() the startup() was never called,
which is not good at all... Fixed. And again - when startup() is not
defined the call will fall back to enable() than to unmask() via
default callbacks.
Signed-off-by: Pawel Moll <pawel.moll@st.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
We don't need whole 32 of them, only NR_SOFTIRQS.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
What I realized recently is that calling rebuild_sched_domains() in
arch_reinit_sched_domains() by itself is not enough when cpusets are enabled.
partition_sched_domains() code is trying to avoid unnecessary domain rebuilds
and will not actually rebuild anything if new domain masks match the old ones.
What this means is that doing
echo 1 > /sys/devices/system/cpu/sched_mc_power_savings
on a system with cpusets enabled will not take affect untill something changes
in the cpuset setup (ie new sets created or deleted).
This patch fixes restore correct behaviour where domains must be rebuilt in
order to enable MC powersaving flags.
Test on quad-core Core2 box with both CONFIG_CPUSETS and !CONFIG_CPUSETS.
Also tested on dual-core Core2 laptop. Lockdep is happy and things are working
as expected.
Signed-off-by: Max Krasnyansky <maxk@qualcomm.com>
Tested-by: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When a cpu is taken offline, the CPU_DYING notifiers are called on the
dying cpu. According to <linux/notifiers.h>, the cpu should be "not
running any task, not handling interrupts, soon dead".
For the current implementation, this is not true:
- __cpu_disable can fail. If it fails, then the cpu will remain alive
and happy.
- At least on x86, __cpu_disable() briefly enables the local interrupts
to handle any outstanding interrupts.
What about moving CPU_DYING down a few lines, behind the __cpu_disable()
line?
There are only two CPU_DYING handlers in the kernel right now: one in
kvm, one in the scheduler. Both should work with the patch applied
[and: I'm not sure if either one handles a failing __cpu_disable()]
The patch survives simple offlining a cpu. kvm untested due to lack
of a test setup.
Signed-off-By: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The __load_balance_iterator() returns a NULL when there's only one
sched_entity which is a task. It is caused by the following code-path.
/* Skip over entities that are not tasks */
do {
se = list_entry(next, struct sched_entity, group_node);
next = next->next;
} while (next != &cfs_rq->tasks && !entity_is_task(se));
if (next == &cfs_rq->tasks)
return NULL;
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This will return NULL even when se is a task.
As a side-effect, there was a regression in sched_mc behavior since 2.6.25,
since iter_move_one_task() when it calls load_balance_start_fair(),
would not get any tasks to move!
Fix this by checking if the last entity was a task or not.
Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
We have a bug in the calculation of the next jiffie to trigger the RTC
synchronisation. The aim here is to run sync_cmos_clock() as close as
possible to the middle of a second. Which means we want this function to
be called less than or equal to half a jiffie away from when now.tv_nsec
equals 5e8 (500000000).
If this is not the case for a given call to the function, for this purpose
instead of updating the RTC we calculate the offset in nanoseconds to the
next point in time where now.tv_nsec will be equal 5e8. The calculated
offset is then converted to jiffies as these are the unit used by the
timer.
Hovewer timespec_to_jiffies() used here uses a ceil()-type rounding mode,
where the resulting value is rounded up. As a result the range of
now.tv_nsec when the timer will trigger is from 5e8 to 5e8 + TICK_NSEC
rather than the desired 5e8 - TICK_NSEC / 2 to 5e8 + TICK_NSEC / 2.
As a result if for example sync_cmos_clock() happens to be called at the
time when now.tv_nsec is between 5e8 + TICK_NSEC / 2 and 5e8 to 5e8 +
TICK_NSEC, it will simply be rescheduled HZ jiffies later, falling in the
same range of now.tv_nsec again. Similarly for cases offsetted by an
integer multiple of TICK_NSEC.
This change addresses the problem by subtracting TICK_NSEC / 2 from the
nanosecond offset to the next point in time where now.tv_nsec will be
equal 5e8, effectively shifting the following rounding in
timespec_to_jiffies() so that it produces a rounded-to-nearest result.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
I found that 2.6.27-rc5-mm1 does not compile with gcc 3.4.6.
The error is:
CC kernel/sched.o
kernel/sched.c: In function `start_rt_bandwidth':
kernel/sched.c:208: sorry, unimplemented: inlining failed in call to 'rt_bandwidth_enabled': function body not available
kernel/sched.c:214: sorry, unimplemented: called from here
make[1]: *** [kernel/sched.o] Error 1
make: *** [kernel] Error 2
It seems that the gcc 3.4.6 requires full inline definition before first usage.
The patch below fixes the compilation problem.
Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl> (if needed>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Until the C1E patches arrived there where no users of periodic broadcast
before switching to oneshot mode. Now we need to trigger a possible
waiter for a periodic broadcast when switching to oneshot mode.
Otherwise we can starve them for ever.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
We want to be able to control the default "rounding" that is used by
select() and poll() and friends. This is a per process property
(so that we can have a "nice" like program to start certain programs with
a looser or stricter rounding) that can be set/get via a prctl().
For this purpose, a field called "timer_slack_ns" is added to the task
struct. In addition, a field called "default_timer_slack"ns" is added
so that tasks easily can temporarily to a more/less accurate slack and then
back to the default.
The default value of the slack is set to 50 usec; this is significantly less
than 2.6.27's average select() and poll() timing error but still allows
the kernel to group timers somewhat to preserve power behavior. Applications
and admins can override this via the prctl()
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
this patch turns hrtimers into range timers; they have 2 expire points
1) the soft expire point
2) the hard expire point
the kernel will do it's regular best effort attempt to get the timer run
at the hard expire point. However, if some other time fires after the soft
expire point, the kernel now has the freedom to fire this timer at this point,
and thus grouping the events and preventing a power-expensive wakeup in the
future.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
In order to be able to do range hrtimers we need to use accessor functions
to the "expire" member of the hrtimer struct.
This patch converts kernel/* to these accessors.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
For the select() rework, it's important to be able to add timespec
structures in an overflow-safe manner.
This patch adds a timespec_add_safe() function for this which is similar in
operation to ktime_add_safe(), but works on a struct timespec.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Steven Rostedt