snowball/igloo-kernel.git/fs/select.c, branch u0

lttng-instrumentation/lttng-instrumentation-fs

2011-03-16T23:04:34+00:00

LTTng instrumentation - fs Core filesystem tracepoints. Tracepoints added : fs_buffer_wait_end fs_buffer_wait_start fs_close fs_exec fs_ioctl fs_llseek fs_lseek fs_open fs_poll fs_pread64 fs_pwrite64 fs_read fs_readv fs_select fs_write fs_writev Instrument the core FS events, extracting the information when it is available. e.g. the instrumentation of "reads" is inserted _after_ the read, when the information is ready. This would not be possible if the instrumentation would be placed elsewhere (at the beginning of the system call for instance). Those tracepoints are used by LTTng. 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. Signed-off-by: Mathieu Desnoyers CC: Alexander Viro CC: Masami Hiramatsu CC: 'Peter Zijlstra' CC: "Frank Ch. Eigler" CC: 'Ingo Molnar' CC: 'Hideo AOKI' CC: Takashi Nishiie CC: 'Steven Rostedt' CC: Eduard - Gabriel Munteanu

poll-wait-exclusive

2011-03-16T23:04:23+00:00

Poll : add poll_wait_set_exclusive Problem description : In LTTng, all lttd readers are polling all the available debugfs files for data. This is principally because the number of reader threads is user-defined and there are typical workloads where a single CPU is producing most of the tracing data and all other CPUs are idle, available to consume data. It therefore makes sense not to tie those threads to specific buffers. However, when the number of threads grows, we face a "thundering herd" problem where many threads can be woken up and put back to sleep, leaving only a single thread doing useful work. Solution : I just created a patch which adds a poll_wait_set_exclusive() primitive to poll(), so the code which implements the pollfd operation can specify that only a single waiter must be woken up. poll_wait_set_exclusive : set poll wait queue to exclusive Sets up a poll wait queue to use exclusive wakeups. This is useful to wake up only one waiter at each wakeup. Used to work-around "thundering herd" problem. Signed-off-by: Mathieu Desnoyers CC: William Lee Irwin III CC: Ingo Molnar

fs/select.c: fix information leak to userspace

2011-01-13T16:03:12+00:00

On some architectures __kernel_suseconds_t is int. On these archs struct timeval has padding bytes at the end. This struct is copied to userspace with these padding bytes uninitialized. This leads to leaking of contents of kernel stack memory. This bug was added with v2.6.27-rc5-286-gb773ad4. [akpm@linux-foundation.org: avoid the memset on architectures which don't need it] Signed-off-by: Vasiliy Kulikov Cc: Al Viro Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

epoll: make epoll_wait() use the hrtimer range feature

2010-10-28T01:03:18+00:00

This make epoll use hrtimers for the timeout value which prevents epoll_wait() from timing out up to a millisecond early. This mirrors the behavior of select() and poll(). Signed-off-by: Shawn Bohrer Cc: Al Viro Acked-by: Davide Libenzi Cc: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

select: rename estimate_accuracy() to select_estimate_accuracy()

2010-10-28T01:03:18+00:00

Make it a subsystem-specific identifier because we wish to amke it non-static in the next patch ("epoll: make epoll_wait() use the hrtimer range feature"). Cc: Shawn Bohrer Cc: Al Viro Cc: Davide Libenzi Cc: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Add generic sys_old_select()

2010-03-12T23:52:32+00:00

Add a generic implementation of the old select() syscall, which expects its argument in a memory block and switch all architectures over to use it. Signed-off-by: Christoph Hellwig Cc: Ralf Baechle Cc: Benjamin Herrenschmidt Cc: Paul Mundt Cc: Jeff Dike Cc: Hirokazu Takata Cc: Thomas Gleixner Cc: Ingo Molnar Reviewed-by: H. Peter Anvin Cc: Al Viro Cc: Arnd Bergmann Cc: Heiko Carstens Cc: Martin Schwidefsky Cc: "Luck, Tony" Cc: James Morris Acked-by: Andreas Schwab Acked-by: Russell King Acked-by: Greg Ungerer Acked-by: David Howells Cc: Andreas Schwab Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

fs: use rlimit helpers

2010-03-06T19:26:29+00:00

Make sure compiler won't do weird things with limits. E.g. fetching them twice may return 2 different values after writable limits are implemented. I.e. either use rlimit helpers added in commit 3e10e716abf3 ("resource: add helpers for fetching rlimits") or ACCESS_ONCE if not applicable. Signed-off-by: Jiri Slaby Cc: Alexander Viro Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

headers: remove sched.h from poll.h

2009-10-04T22:05:10+00:00

Signed-off-by: Alexey Dobriyan Signed-off-by: Linus Torvalds

poll/select: avoid arithmetic overflow in __estimate_accuracy()

2009-09-23T14:39:27+00:00

__estimate_accuracy() was prone to integer overflow, for example if *tv == {2147, 483648000} on a 32 bit computer (or even for delays as small as {429, 500000000} if the task is niced). Because the result was already forced between 0 and 100ms, the effect of the overflow was not too problematic, but the use of the hrtimer range feature was not optimal in overflow cases. This patch ensures that there can not be an integer overflow in this function. Signed-off-by: Guillaume Knispel Cc: Alexander Viro Cc: Arjan van de Ven Cc: Thomas Gleixner Cc: Heiko Carstens Cc: Tejun Heo Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

poll/select: initialize triggered field of struct poll_wqueues

2009-08-16T01:40:11+00:00

The triggered field of struct poll_wqueues introduced in commit 5f820f648c92a5ecc771a96b3c29aa6e90013bba ("poll: allow f_op->poll to sleep"). It was first set to 1 in pollwake() (now __pollwake() ), tested and later set to 0 in poll_schedule_timeout(), but not initialized before. As a result when the process needs to sleep, triggered was likely to be non-zero even if pollwake() is not called before the first poll_schedule_timeout(), meaning schedule_hrtimeout_range() would not be called and an extra loop calling all ->poll() would be done. This patch initialize triggered to 0 in poll_initwait() so the ->poll() are not called twice before the process goes to sleep when it needs to. Signed-off-by: Guillaume Knispel Acked-by: Thomas Gleixner Acked-by: Tejun Heo Cc: stable@kernel.org Signed-off-by: Linus Torvalds