When print-fatal-signals is enabled it's possible to dump any memory
reachable by the kernel to the log by simply jumping to that address from
user space.
Or crash the system if there's some hardware with read side effects.
The fatal signals handler will dump 16 bytes at the execution address,
which is fully controlled by ring 3.
In addition when something jumps to a unmapped address there will be up to
16 additional useless page faults, which might be potentially slow (and at
least is not very efficient)
Fortunately this option is off by default and only there on i386.
But fix it by checking for kernel addresses and also stopping when there's
a page fault.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The LTP cgroup test suite generates a "kernel BUG at kernel/cgroup.c:790!"
here in cgroup_diput():
/*
* if we're getting rid of the cgroup, refcount should ensure
* that there are no pidlists left.
*/
BUG_ON(!list_empty(&cgrp->pidlists));
The cgroup pidlist rework in 2.6.32 generates the BUG_ON, which is caused
when pidlist_array_load() calls cgroup_pidlist_find():
(1) if a matching cgroup_pidlist is found, it down_write's the mutex of the
pre-existing cgroup_pidlist, and increments its use_count.
(2) if no matching cgroup_pidlist is found, then a new one is allocated, it
down_write's its mutex, and the use_count is set to 0.
(3) the matching, or new, cgroup_pidlist gets returned back to pidlist_array_load(),
which increments its use_count -- regardless whether new or pre-existing --
and up_write's the mutex.
So if a matching list is ever encountered by cgroup_pidlist_find() during
the life of a cgroup directory, it results in an inflated use_count value,
preventing it from ever getting released by cgroup_release_pid_array().
Then if the directory is subsequently removed, cgroup_diput() hits the
BUG_ON() when it finds that the directory's cgroup is still populated with
a pidlist.
The patch simply removes the use_count increment when a matching pidlist
is found by cgroup_pidlist_find(), because it gets bumped by the calling
pidlist_array_load() function while still protected by the list's mutex.
Signed-off-by: Dave Anderson <anderson@redhat.com>
Reviewed-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Ben Blum <bblum@andrew.cmu.edu>
Cc: Paul Menage <menage@google.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix resource (write-pipe file) leak in call_usermodehelper_pipe().
When call_usermodehelper_exec() fails, write-pipe file is opened and
call_usermodehelper_pipe() just returns an error. Since it is hard for
caller to determine whether the error occured when opening the pipe or
executing the helper, the caller cannot close the pipe by themselves.
I've found this resoruce leak when testing coredump. You can check how
the resource leaks as below;
$ echo "|nocommand" > /proc/sys/kernel/core_pattern
$ ulimit -c unlimited
$ while [ 1 ]; do ./segv; done &> /dev/null &
$ cat /proc/meminfo (<- repeat it)
where segv.c is;
//-----
int main () {
char *p = 0;
*p = 1;
}
//-----
This patch closes write-pipe file if call_usermodehelper_exec() failed.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If the very unlikely case happens where the writer moves the head by one
between where the head page is read and where the new reader page
is assigned _and_ the writer then writes and wraps the entire ring buffer
so that the head page is back to what was originally read as the head page,
the page to be swapped will have a corrupted next pointer.
Simple solution is to wrap the assignment of the next pointer with a
rb_list_head().
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This reference at the end of rb_get_reader_page() was causing off-by-one
writes to the prev pointer of the page after the reader page when that
page is the head page, and therefore the reader page has the RB_PAGE_HEAD
flag in its list.next pointer. This eventually results in a GPF in a
subsequent call to rb_set_head_page() (usually from rb_get_reader_page())
when that prev pointer is dereferenced. The dereferenced register would
characteristically have an address that appears shifted left by one byte
(eg, ffxxxxxxxxxxxxyy instead of ffffxxxxxxxxxxxx) due to being written at
an address one byte too high.
Signed-off-by: David Sharp <dhsharp@google.com>
LKML-Reference: <1262826727-9090-1-git-send-email-dhsharp@google.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
If the ftrace stacktrace option is set, then add the stack dumps to
trace_printk.
Requested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
At the beginning, access to the ring buffer was fully serialized
by trace_types_lock. Patch d7350c3f45 gives more freedom to readers,
and patch b04cc6b1f6 adds code to protect trace_pipe and cpu#/trace_pipe.
But actually it is not enough, ring buffer readers are not always
read-only, they may consume data.
This patch makes accesses to trace, trace_pipe, trace_pipe_raw
cpu#/trace, cpu#/trace_pipe and cpu#/trace_pipe_raw serialized.
And removes tracing_reader_cpumask which is used to protect trace_pipe.
Details:
Ring buffer serializes readers, but it is low level protection.
The validity of the events (which returns by ring_buffer_peek() ..etc)
are not protected by ring buffer.
The content of events may become garbage if we allow another process to consume
these events concurrently:
A) the page of the consumed events may become a normal page
(not reader page) in ring buffer, and this page will be rewritten
by the events producer.
B) The page of the consumed events may become a page for splice_read,
and this page will be returned to system.
This patch adds trace_access_lock() and trace_access_unlock() primitives.
These primitives allow multi process access to different cpu ring buffers
concurrently.
These primitives don't distinguish read-only and read-consume access.
Multi read-only access is also serialized.
And we don't use these primitives when we open files,
we only use them when we read files.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B447D52.1050602@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The previous patches added the use of print_fmt string and changes
the trace_define_field() function to also create the fields and
format output for the event format files.
text data bss dec hex filename
5857201 1355780 9336808 16549789 fc879d vmlinux
5884589 1351684 9337896 16574169 fce6d9 vmlinux-orig
The above shows the size of the vmlinux after this patch set
compared to the vmlinux-orig which is before the patch set.
This saves us 27k on text, 1k on bss and adds just 4k of data.
The total savings of 24k in size.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D4D.40604@cn.fujitsu.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The calls ftrace_format_##call() and ftrace_define_fields_##call()
are almost duplicate in functionality. With the addition of the
print_fmt in previous patches, these two functions can be merged
into one.
The trace_define_field() defines the fields and links them into
the struct ftrace_event_call. The previous patches introduced
the print_fmt field and this can now be used with the trace_define_field()
to create the event format file fields and print_fmt field.
The struct ftrace_event_call->fields are used to print the fields
The struct ftrace_event_call->print_fmt is used to print
the "print fmt: XXXXXXXXXXX" line.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D49.5000006@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
In the clean up of having all events call one specific function,
the syscall event init was changed to call this helper function.
With the new print_fmt updates, the syscalls need to do special
initializations. This patch converts the syscall events to call
its own init function again.
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This is part of a patch set that removes the show_format method
in the ftrace event macros.
Add the print_fmt initialization to the kprobe events.
The print_fmt is still not used, but will be in the follow up
patches.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D45.3080100@cn.fujitsu.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This is part of a patch set that removes the show_format method
in the ftrace event macros.
Add the print_fmt initialization to the syscall events.
The print_fmt is still not used, but will be in the follow up
patches.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D41.609@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This is part of a patch set that removes the show_format method
in the ftrace event macros.
The print_fmt field is added to hold the string that shows
the print_fmt in the event format files. This patch only adds
the field but it is currently not used. Later patches will use
this field to enable us to remove the show_format field
and function.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D3E.2000704@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This is part of a patch set that removes the show_format method
in the ftrace event macros.
This patch set requires that all fields are added to the
ftrace_event_call->fields. This patch changes __dynamic_array()
to call trace_define_field() to include fields that use __dynamic_array().
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D36.8090100@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Commit 35dead4 "modules: don't export section names of empty sections
via sysfs" changed the set of sections that have attributes, but did
not change the iteration over these attributes in add_notes_attrs().
This can lead to add_notes_attrs() creating attributes with the wrong
names or with null name pointers.
Introduce a sect_empty() function and use it in both add_sect_attrs()
and add_notes_attrs().
Reported-by: Martin Michlmayr <tbm@cyrius.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Tested-by: Martin Michlmayr <tbm@cyrius.com>
Cc: stable@kernel.org
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch introduces an interface to process data objects
in parallel. The parallelized objects return after serialization
in the same order as they were before the parallelization.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
ringbuffer*.c are the last users of local.h.
Remove the include from modules.h and add it to ringbuffer files.
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Use cpu ops to deal with the per cpu data instead of a local_t. Reduces memory
requirements, cache footprint and decreases cycle counts.
The this_cpu_xx operations are also used for !SMP mode. Otherwise we could
not drop the use of __module_ref_addr() which would make per cpu data handling
complicated. this_cpu_xx operations have their own fallback for !SMP.
V8-V9:
- Leave include asm/module.h since ringbuffer.c depends on it. Nothing else
does though. Another patch will deal with that.
- Remove spurious free.
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Tejun Heo <tj@kernel.org>
* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
perf kmem: Fix statistics typo
kprobes: Fix distinct type warning
perf: Rename perf_event_hw_event in design document
perf tools: Add missing header files to LIB_H Makefile variable
perf record: We should fork only if a program was specified to run
perf diff: Fix usage array, it must end with a NULL entry
* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
tracing: Fix sign fields in ftrace_define_fields_##call()
tracing/syscalls: Fix typo in SYSCALL_DEFINE0
tracing/kprobe: Show sign of fields in trace_kprobe format files
ksym_tracer: Remove trace_stat
ksym_tracer: Fix race when incrementing count
ksym_tracer: Fix to allow writing newline to ksym_trace_filter
ksym_tracer: Fix to make the tracer work
tracing: Kconfig spelling fixes and cleanups
tracing: Fix setting tracer specific options
Documentation: Update ftrace-design.txt
Documentation: Update tracepoint-analysis.txt
Documentation: Update mmiotrace.txt
crash_kexec gets called before kmsg_dump(KMSG_DUMP_OOPS) if
panic_on_oops is set, so the kernel log buffer is not stored
for this case.
This patch adds a KMSG_DUMP_KEXEC dump type which gets called
when crash_kexec() is invoked. To avoid getting double dumps,
the old KMSG_DUMP_PANIC is moved below crash_kexec(). The
mtdoops driver is modified to handle KMSG_DUMP_KEXEC in the
same way as a panic.
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Simon Kagstrom <simon.kagstrom@netinsight.net>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Liming found a NULL deref when a task has a perf context but no
counters when it forks.
This can occur in two cases, a race during construction where
the fork hits after installing the context but before the first
counter gets inserted, or more reproducably, a fork after the
last counter is closed (which leaves the context around).
Reported-by: Wang Liming <liming.wang@windriver.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
CC: <stable@kernel.org>
LKML-Reference: <1262185684.7135.222.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add is_signed_type() call to trace_define_field() in ftrace macros.
The code previously just passed in 0 (false), disregarding whether
or not the field was actually a signed type.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D3A.6020007@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The format files of trace_kprobe do not show the sign of the fields.
The other format files show the field signed type of the fields and
this patch makes the trace_kprobe formats consistent with the others.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
LKML-Reference: <4B273D27.5040009@cn.fujitsu.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Quoted from Ingo:
| This reminds me - i think we should eliminate CONFIG_EVENT_PROFILE -
| it's an unnecessary Kconfig complication. If both PERF_EVENTS and
| EVENT_TRACING is enabled we should expose generic tracepoints.
|
| Nor is it limited to event 'profiling', so it has become a misnomer as
| well.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <4B2F1557.2050705@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Every time I see this:
kernel/kprobes.c: In function 'register_kretprobe':
kernel/kprobes.c:1038: warning: comparison of distinct pointer types lacks a cast
I'm wondering if something changed in common code and we need to
do something for s390. Apparently that's not the case.
Let's get rid of this annoying warning.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Masami Hiramatsu <mhiramat@redhat.com>
LKML-Reference: <20091221120224.GA4471@osiris.boeblingen.de.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Since we only ever schedule the local cpu, there is no need to pass the
cpu number to the perf sched hooks.
This micro-optimizes things a bit.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'sysctl' of git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc-2.6:
SYSCTL: Add a mutex to the page_alloc zone order sysctl
SYSCTL: Print binary sysctl warnings (nearly) only once
When printing legacy sysctls print the warning message
for each of them only once. This way there is a guarantee
the syslog won't be flooded for any sane program.
The original attempt at this made the tables non const and stored
the flag inline.
Linus suggested using a separate hash table for this, this is based on a
code snippet from him.
The hash implies this is not exact and can sometimes not print a
new sysctl due to a hash collision, but in practice this should not
be a problem
I used a FNV32 hash over the binary string with a 32byte bitmap. This
gives relatively little collisions when all the predefined binary sysctls
are hashed:
size 256
bucket
length number
0: [25]
1: [67]
2: [88]
3: [47]
4: [22]
5: [6]
6: [1]
The worst case is a single collision of 6 hash values.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Effectively reverts 738d2be430.
As demonstrated by Eric, we really need to call __set_task_cpu()
early in the fork() path to properly initialize the various task
state -- specifically the cgroup state through set_task_rq().
[ we could probably fix this by explicitly calling
__set_task_cpu() from sched_fork(), but lets try that for the
next cycle and simply revert to the old behaviour for now. ]
Reported-by: Eric Paris <eparis@redhat.com>
Tested-by: Eric Paris <eparis@redhat.com>,
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: efault@gmx.de
LKML-Reference: <1261492999.4937.36.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add kfifo_in_rec() - puts some record data into the FIFO
Add kfifo_out_rec() - gets some record data from the FIFO
Add kfifo_from_user_rec() - puts some data from user space into the FIFO
Add kfifo_to_user_rec() - gets data from the FIFO and write it to user space
Add kfifo_peek_rec() - gets the size of the next FIFO record field
Add kfifo_skip_rec() - skip the next fifo out record
Add kfifo_avail_rec() - determinate the number of bytes available in a record FIFO
Signed-off-by: Stefani Seibold <stefani@seibold.net>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add kfifo_reset_out() for save lockless discard the fifo output
Add kfifo_skip() to skip a number of output bytes
Add kfifo_from_user() to copy user space data into the fifo
Add kfifo_to_user() to copy fifo data to user space
Signed-off-by: Stefani Seibold <stefani@seibold.net>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
rename kfifo_put... into kfifo_in... to prevent miss use of old non in
kernel-tree drivers
ditto for kfifo_get... -> kfifo_out...
Improve the prototypes of kfifo_in and kfifo_out to make the kerneldoc
annotations more readable.
Add mini "howto porting to the new API" in kfifo.h
Signed-off-by: Stefani Seibold <stefani@seibold.net>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Move the pointer to the spinlock out of struct kfifo. Most users in
tree do not actually use a spinlock, so the few exceptions now have to
call kfifo_{get,put}_locked, which takes an extra argument to a
spinlock.
Signed-off-by: Stefani Seibold <stefani@seibold.net>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is a new generic kernel FIFO implementation.
The current kernel fifo API is not very widely used, because it has to
many constrains. Only 17 files in the current 2.6.31-rc5 used it.
FIFO's are like list's a very basic thing and a kfifo API which handles
the most use case would save a lot of development time and memory
resources.
I think this are the reasons why kfifo is not in use:
- The API is to simple, important functions are missing
- A fifo can be only allocated dynamically
- There is a requirement of a spinlock whether you need it or not
- There is no support for data records inside a fifo
So I decided to extend the kfifo in a more generic way without blowing up
the API to much. The new API has the following benefits:
- Generic usage: For kernel internal use and/or device driver.
- Provide an API for the most use case.
- Slim API: The whole API provides 25 functions.
- Linux style habit.
- DECLARE_KFIFO, DEFINE_KFIFO and INIT_KFIFO Macros
- Direct copy_to_user from the fifo and copy_from_user into the fifo.
- The kfifo itself is an in place member of the using data structure, this save an
indirection access and does not waste the kernel allocator.
- Lockless access: if only one reader and one writer is active on the fifo,
which is the common use case, no additional locking is necessary.
- Remove spinlock - give the user the freedom of choice what kind of locking to use if
one is required.
- Ability to handle records. Three type of records are supported:
- Variable length records between 0-255 bytes, with a record size
field of 1 bytes.
- Variable length records between 0-65535 bytes, with a record size
field of 2 bytes.
- Fixed size records, which no record size field.
- Preserve memory resource.
- Performance!
- Easy to use!
This patch:
Since most users want to have the kfifo as part of another object,
reorganize the code to allow including struct kfifo in another data
structure. This requires changing the kfifo_alloc and kfifo_init
prototypes so that we pass an existing kfifo pointer into them. This
patch changes the implementation and all existing users.
[akpm@linux-foundation.org: fix warning]
Signed-off-by: Stefani Seibold <stefani@seibold.net>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit 7bc7d63745, as
requested by John Stultz. Quoting John:
"Petr Titěra reported an issue where he saw odd atime regressions with
2.6.33 where there were a full second worth of nanoseconds in the
nanoseconds field.
He also reviewed the time code and narrowed down the problem: unhandled
overflow of the nanosecond field caused by rounding up the
sub-nanosecond accumulated time.
Details:
* At the end of update_wall_time(), we currently round up the
sub-nanosecond portion of accumulated time when storing it into xtime.
This was added to avoid time inconsistencies caused when the
sub-nanosecond portion was truncated when storing into xtime.
Unfortunately we don't handle the possible second overflow caused by
that rounding.
* Previously the xtime_cache code hid this overflow by normalizing the
xtime value when storing into the xtime_cache.
* We could try to handle the second overflow after the rounding up, but
since this affects the timekeeping's internal state, this would further
complicate the next accumulation cycle, causing small errors in ntp
steering. As much as I'd like to get rid of it, the xtime_cache code is
known to work.
* The correct fix is really to include the sub-nanosecond portion in the
timekeeping accessor function, so we don't need to round up at during
accumulation. This would greatly simplify the accumulation code.
Unfortunately, we can't do this safely until the last three
non-GENERIC_TIME arches (sparc32, arm, cris) are converted (those
patches are in -mm) and we kill off the spots where arches set xtime
directly. This is all 2.6.34 material, so I think reverting the
xtime_cache change is the best approach for now.
Many thanks to Petr for both reporting and finding the issue!"
Reported-by: Petr Titěra <P.Titera@century.cz>
Requested-by: john stultz <johnstul@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* pull ACC_MODE to fs.h; we have several copies all over the place
* nightmarish expression calculating f_mode by f_flags deserves a helper
too (OPEN_FMODE(flags))
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
It seems a couple places such as arch/ia64/kernel/perfmon.c and
drivers/infiniband/core/uverbs_main.c could use anon_inode_getfile()
instead of a private pseudo-fs + alloc_file(), if only there were a way
to get a read-only file. So provide this by having anon_inode_getfile()
create a read-only file if we pass O_RDONLY in flags.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>