linux-sg2042/kernel/compat.c

1131 lines
28 KiB
C
Raw Normal View History

/*
* linux/kernel/compat.c
*
* Kernel compatibililty routines for e.g. 32 bit syscall support
* on 64 bit kernels.
*
* Copyright (C) 2002-2003 Stephen Rothwell, IBM Corporation
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
#include <linux/linkage.h>
#include <linux/compat.h>
#include <linux/errno.h>
#include <linux/time.h>
#include <linux/signal.h>
#include <linux/sched.h> /* for MAX_SCHEDULE_TIMEOUT */
#include <linux/syscalls.h>
#include <linux/unistd.h>
#include <linux/security.h>
#include <linux/timex.h>
#include <linux/migrate.h>
#include <linux/posix-timers.h>
timers: fix itimer/many thread hang 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>
2008-09-13 00:54:39 +08:00
#include <linux/times.h>
Allow times and time system calls to return small negative values At the moment, the times() system call will appear to fail for a period shortly after boot, while the value it want to return is between -4095 and -1. The same thing will also happen for the time() system call on 32-bit platforms some time in 2106 or so. On some platforms, such as x86, this is unavoidable because of the system call ABI, but other platforms such as powerpc have a separate error indication from the return value, so system calls can in fact return small negative values without indicating an error. On those platforms, force_successful_syscall_return() provides a way to indicate that the system call return value should not be treated as an error even if it is in the range which would normally be taken as a negative error number. This adds a force_successful_syscall_return() call to the time() and times() system calls plus their 32-bit compat versions, so that they don't erroneously indicate an error on those platforms whose system call ABI has a separate error indication. This will not affect anything on other platforms. Joakim Tjernlund added the fix for time() and the compat versions of time() and times(), after I did the fix for times(). Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se> Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: David S. Miller <davem@davemloft.net> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-07 06:41:02 +08:00
#include <linux/ptrace.h>
#include <asm/uaccess.h>
/*
* Note that the native side is already converted to a timespec, because
* that's what we want anyway.
*/
static int compat_get_timeval(struct timespec *o,
struct compat_timeval __user *i)
{
long usec;
if (get_user(o->tv_sec, &i->tv_sec) ||
get_user(usec, &i->tv_usec))
return -EFAULT;
o->tv_nsec = usec * 1000;
return 0;
}
static int compat_put_timeval(struct compat_timeval __user *o,
struct timeval *i)
{
return (put_user(i->tv_sec, &o->tv_sec) ||
put_user(i->tv_usec, &o->tv_usec)) ? -EFAULT : 0;
}
asmlinkage long compat_sys_gettimeofday(struct compat_timeval __user *tv,
struct timezone __user *tz)
{
if (tv) {
struct timeval ktv;
do_gettimeofday(&ktv);
if (compat_put_timeval(tv, &ktv))
return -EFAULT;
}
if (tz) {
if (copy_to_user(tz, &sys_tz, sizeof(sys_tz)))
return -EFAULT;
}
return 0;
}
asmlinkage long compat_sys_settimeofday(struct compat_timeval __user *tv,
struct timezone __user *tz)
{
struct timespec kts;
struct timezone ktz;
if (tv) {
if (compat_get_timeval(&kts, tv))
return -EFAULT;
}
if (tz) {
if (copy_from_user(&ktz, tz, sizeof(ktz)))
return -EFAULT;
}
return do_sys_settimeofday(tv ? &kts : NULL, tz ? &ktz : NULL);
}
int get_compat_timespec(struct timespec *ts, const struct compat_timespec __user *cts)
{
return (!access_ok(VERIFY_READ, cts, sizeof(*cts)) ||
__get_user(ts->tv_sec, &cts->tv_sec) ||
__get_user(ts->tv_nsec, &cts->tv_nsec)) ? -EFAULT : 0;
}
int put_compat_timespec(const struct timespec *ts, struct compat_timespec __user *cts)
{
return (!access_ok(VERIFY_WRITE, cts, sizeof(*cts)) ||
__put_user(ts->tv_sec, &cts->tv_sec) ||
__put_user(ts->tv_nsec, &cts->tv_nsec)) ? -EFAULT : 0;
}
static long compat_nanosleep_restart(struct restart_block *restart)
{
struct compat_timespec __user *rmtp;
struct timespec rmt;
mm_segment_t oldfs;
long ret;
restart->nanosleep.rmtp = (struct timespec __user *) &rmt;
oldfs = get_fs();
set_fs(KERNEL_DS);
ret = hrtimer_nanosleep_restart(restart);
set_fs(oldfs);
if (ret) {
rmtp = restart->nanosleep.compat_rmtp;
if (rmtp && put_compat_timespec(&rmt, rmtp))
return -EFAULT;
}
return ret;
}
asmlinkage long compat_sys_nanosleep(struct compat_timespec __user *rqtp,
struct compat_timespec __user *rmtp)
{
struct timespec tu, rmt;
mm_segment_t oldfs;
long ret;
if (get_compat_timespec(&tu, rqtp))
return -EFAULT;
if (!timespec_valid(&tu))
return -EINVAL;
oldfs = get_fs();
set_fs(KERNEL_DS);
ret = hrtimer_nanosleep(&tu,
rmtp ? (struct timespec __user *)&rmt : NULL,
HRTIMER_MODE_REL, CLOCK_MONOTONIC);
set_fs(oldfs);
if (ret) {
struct restart_block *restart
= &current_thread_info()->restart_block;
restart->fn = compat_nanosleep_restart;
restart->nanosleep.compat_rmtp = rmtp;
if (rmtp && put_compat_timespec(&rmt, rmtp))
return -EFAULT;
}
return ret;
}
static inline long get_compat_itimerval(struct itimerval *o,
struct compat_itimerval __user *i)
{
return (!access_ok(VERIFY_READ, i, sizeof(*i)) ||
(__get_user(o->it_interval.tv_sec, &i->it_interval.tv_sec) |
__get_user(o->it_interval.tv_usec, &i->it_interval.tv_usec) |
__get_user(o->it_value.tv_sec, &i->it_value.tv_sec) |
__get_user(o->it_value.tv_usec, &i->it_value.tv_usec)));
}
static inline long put_compat_itimerval(struct compat_itimerval __user *o,
struct itimerval *i)
{
return (!access_ok(VERIFY_WRITE, o, sizeof(*o)) ||
(__put_user(i->it_interval.tv_sec, &o->it_interval.tv_sec) |
__put_user(i->it_interval.tv_usec, &o->it_interval.tv_usec) |
__put_user(i->it_value.tv_sec, &o->it_value.tv_sec) |
__put_user(i->it_value.tv_usec, &o->it_value.tv_usec)));
}
asmlinkage long compat_sys_getitimer(int which,
struct compat_itimerval __user *it)
{
struct itimerval kit;
int error;
error = do_getitimer(which, &kit);
if (!error && put_compat_itimerval(it, &kit))
error = -EFAULT;
return error;
}
asmlinkage long compat_sys_setitimer(int which,
struct compat_itimerval __user *in,
struct compat_itimerval __user *out)
{
struct itimerval kin, kout;
int error;
if (in) {
if (get_compat_itimerval(&kin, in))
return -EFAULT;
} else
memset(&kin, 0, sizeof(kin));
error = do_setitimer(which, &kin, out ? &kout : NULL);
if (error || !out)
return error;
if (put_compat_itimerval(out, &kout))
return -EFAULT;
return 0;
}
timers: fix itimer/many thread hang 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>
2008-09-13 00:54:39 +08:00
static compat_clock_t clock_t_to_compat_clock_t(clock_t x)
{
return compat_jiffies_to_clock_t(clock_t_to_jiffies(x));
}
asmlinkage long compat_sys_times(struct compat_tms __user *tbuf)
{
if (tbuf) {
timers: fix itimer/many thread hang 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>
2008-09-13 00:54:39 +08:00
struct tms tms;
struct compat_tms tmp;
timers: fix itimer/many thread hang 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>
2008-09-13 00:54:39 +08:00
do_sys_times(&tms);
/* Convert our struct tms to the compat version. */
tmp.tms_utime = clock_t_to_compat_clock_t(tms.tms_utime);
tmp.tms_stime = clock_t_to_compat_clock_t(tms.tms_stime);
tmp.tms_cutime = clock_t_to_compat_clock_t(tms.tms_cutime);
tmp.tms_cstime = clock_t_to_compat_clock_t(tms.tms_cstime);
if (copy_to_user(tbuf, &tmp, sizeof(tmp)))
return -EFAULT;
}
Allow times and time system calls to return small negative values At the moment, the times() system call will appear to fail for a period shortly after boot, while the value it want to return is between -4095 and -1. The same thing will also happen for the time() system call on 32-bit platforms some time in 2106 or so. On some platforms, such as x86, this is unavoidable because of the system call ABI, but other platforms such as powerpc have a separate error indication from the return value, so system calls can in fact return small negative values without indicating an error. On those platforms, force_successful_syscall_return() provides a way to indicate that the system call return value should not be treated as an error even if it is in the range which would normally be taken as a negative error number. This adds a force_successful_syscall_return() call to the time() and times() system calls plus their 32-bit compat versions, so that they don't erroneously indicate an error on those platforms whose system call ABI has a separate error indication. This will not affect anything on other platforms. Joakim Tjernlund added the fix for time() and the compat versions of time() and times(), after I did the fix for times(). Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se> Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: David S. Miller <davem@davemloft.net> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-07 06:41:02 +08:00
force_successful_syscall_return();
return compat_jiffies_to_clock_t(jiffies);
}
/*
* Assumption: old_sigset_t and compat_old_sigset_t are both
* types that can be passed to put_user()/get_user().
*/
asmlinkage long compat_sys_sigpending(compat_old_sigset_t __user *set)
{
old_sigset_t s;
long ret;
mm_segment_t old_fs = get_fs();
set_fs(KERNEL_DS);
ret = sys_sigpending((old_sigset_t __user *) &s);
set_fs(old_fs);
if (ret == 0)
ret = put_user(s, set);
return ret;
}
asmlinkage long compat_sys_sigprocmask(int how, compat_old_sigset_t __user *set,
compat_old_sigset_t __user *oset)
{
old_sigset_t s;
long ret;
mm_segment_t old_fs;
if (set && get_user(s, set))
return -EFAULT;
old_fs = get_fs();
set_fs(KERNEL_DS);
ret = sys_sigprocmask(how,
set ? (old_sigset_t __user *) &s : NULL,
oset ? (old_sigset_t __user *) &s : NULL);
set_fs(old_fs);
if (ret == 0)
if (oset)
ret = put_user(s, oset);
return ret;
}
asmlinkage long compat_sys_setrlimit(unsigned int resource,
struct compat_rlimit __user *rlim)
{
struct rlimit r;
int ret;
mm_segment_t old_fs = get_fs ();
if (resource >= RLIM_NLIMITS)
return -EINVAL;
if (!access_ok(VERIFY_READ, rlim, sizeof(*rlim)) ||
__get_user(r.rlim_cur, &rlim->rlim_cur) ||
__get_user(r.rlim_max, &rlim->rlim_max))
return -EFAULT;
if (r.rlim_cur == COMPAT_RLIM_INFINITY)
r.rlim_cur = RLIM_INFINITY;
if (r.rlim_max == COMPAT_RLIM_INFINITY)
r.rlim_max = RLIM_INFINITY;
set_fs(KERNEL_DS);
ret = sys_setrlimit(resource, (struct rlimit __user *) &r);
set_fs(old_fs);
return ret;
}
#ifdef COMPAT_RLIM_OLD_INFINITY
asmlinkage long compat_sys_old_getrlimit(unsigned int resource,
struct compat_rlimit __user *rlim)
{
struct rlimit r;
int ret;
mm_segment_t old_fs = get_fs();
set_fs(KERNEL_DS);
ret = sys_old_getrlimit(resource, &r);
set_fs(old_fs);
if (!ret) {
if (r.rlim_cur > COMPAT_RLIM_OLD_INFINITY)
r.rlim_cur = COMPAT_RLIM_INFINITY;
if (r.rlim_max > COMPAT_RLIM_OLD_INFINITY)
r.rlim_max = COMPAT_RLIM_INFINITY;
if (!access_ok(VERIFY_WRITE, rlim, sizeof(*rlim)) ||
__put_user(r.rlim_cur, &rlim->rlim_cur) ||
__put_user(r.rlim_max, &rlim->rlim_max))
return -EFAULT;
}
return ret;
}
#endif
asmlinkage long compat_sys_getrlimit (unsigned int resource,
struct compat_rlimit __user *rlim)
{
struct rlimit r;
int ret;
mm_segment_t old_fs = get_fs();
set_fs(KERNEL_DS);
ret = sys_getrlimit(resource, (struct rlimit __user *) &r);
set_fs(old_fs);
if (!ret) {
if (r.rlim_cur > COMPAT_RLIM_INFINITY)
r.rlim_cur = COMPAT_RLIM_INFINITY;
if (r.rlim_max > COMPAT_RLIM_INFINITY)
r.rlim_max = COMPAT_RLIM_INFINITY;
if (!access_ok(VERIFY_WRITE, rlim, sizeof(*rlim)) ||
__put_user(r.rlim_cur, &rlim->rlim_cur) ||
__put_user(r.rlim_max, &rlim->rlim_max))
return -EFAULT;
}
return ret;
}
int put_compat_rusage(const struct rusage *r, struct compat_rusage __user *ru)
{
if (!access_ok(VERIFY_WRITE, ru, sizeof(*ru)) ||
__put_user(r->ru_utime.tv_sec, &ru->ru_utime.tv_sec) ||
__put_user(r->ru_utime.tv_usec, &ru->ru_utime.tv_usec) ||
__put_user(r->ru_stime.tv_sec, &ru->ru_stime.tv_sec) ||
__put_user(r->ru_stime.tv_usec, &ru->ru_stime.tv_usec) ||
__put_user(r->ru_maxrss, &ru->ru_maxrss) ||
__put_user(r->ru_ixrss, &ru->ru_ixrss) ||
__put_user(r->ru_idrss, &ru->ru_idrss) ||
__put_user(r->ru_isrss, &ru->ru_isrss) ||
__put_user(r->ru_minflt, &ru->ru_minflt) ||
__put_user(r->ru_majflt, &ru->ru_majflt) ||
__put_user(r->ru_nswap, &ru->ru_nswap) ||
__put_user(r->ru_inblock, &ru->ru_inblock) ||
__put_user(r->ru_oublock, &ru->ru_oublock) ||
__put_user(r->ru_msgsnd, &ru->ru_msgsnd) ||
__put_user(r->ru_msgrcv, &ru->ru_msgrcv) ||
__put_user(r->ru_nsignals, &ru->ru_nsignals) ||
__put_user(r->ru_nvcsw, &ru->ru_nvcsw) ||
__put_user(r->ru_nivcsw, &ru->ru_nivcsw))
return -EFAULT;
return 0;
}
asmlinkage long compat_sys_getrusage(int who, struct compat_rusage __user *ru)
{
struct rusage r;
int ret;
mm_segment_t old_fs = get_fs();
set_fs(KERNEL_DS);
ret = sys_getrusage(who, (struct rusage __user *) &r);
set_fs(old_fs);
if (ret)
return ret;
if (put_compat_rusage(&r, ru))
return -EFAULT;
return 0;
}
asmlinkage long
compat_sys_wait4(compat_pid_t pid, compat_uint_t __user *stat_addr, int options,
struct compat_rusage __user *ru)
{
if (!ru) {
return sys_wait4(pid, stat_addr, options, NULL);
} else {
struct rusage r;
int ret;
unsigned int status;
mm_segment_t old_fs = get_fs();
set_fs (KERNEL_DS);
ret = sys_wait4(pid,
(stat_addr ?
(unsigned int __user *) &status : NULL),
options, (struct rusage __user *) &r);
set_fs (old_fs);
if (ret > 0) {
if (put_compat_rusage(&r, ru))
return -EFAULT;
if (stat_addr && put_user(status, stat_addr))
return -EFAULT;
}
return ret;
}
}
asmlinkage long compat_sys_waitid(int which, compat_pid_t pid,
struct compat_siginfo __user *uinfo, int options,
struct compat_rusage __user *uru)
{
siginfo_t info;
struct rusage ru;
long ret;
mm_segment_t old_fs = get_fs();
memset(&info, 0, sizeof(info));
set_fs(KERNEL_DS);
ret = sys_waitid(which, pid, (siginfo_t __user *)&info, options,
uru ? (struct rusage __user *)&ru : NULL);
set_fs(old_fs);
if ((ret < 0) || (info.si_signo == 0))
return ret;
if (uru) {
ret = put_compat_rusage(&ru, uru);
if (ret)
return ret;
}
BUG_ON(info.si_code & __SI_MASK);
info.si_code |= __SI_CHLD;
return copy_siginfo_to_user32(uinfo, &info);
}
static int compat_get_user_cpu_mask(compat_ulong_t __user *user_mask_ptr,
unsigned len, struct cpumask *new_mask)
{
unsigned long *k;
if (len < cpumask_size())
memset(new_mask, 0, cpumask_size());
else if (len > cpumask_size())
len = cpumask_size();
k = cpumask_bits(new_mask);
return compat_get_bitmap(k, user_mask_ptr, len * 8);
}
asmlinkage long compat_sys_sched_setaffinity(compat_pid_t pid,
unsigned int len,
compat_ulong_t __user *user_mask_ptr)
{
cpumask_var_t new_mask;
int retval;
if (!alloc_cpumask_var(&new_mask, GFP_KERNEL))
return -ENOMEM;
retval = compat_get_user_cpu_mask(user_mask_ptr, len, new_mask);
if (retval)
goto out;
retval = sched_setaffinity(pid, new_mask);
out:
free_cpumask_var(new_mask);
return retval;
}
asmlinkage long compat_sys_sched_getaffinity(compat_pid_t pid, unsigned int len,
compat_ulong_t __user *user_mask_ptr)
{
int ret;
cpumask_var_t mask;
unsigned long *k;
unsigned int min_length = cpumask_size();
if (nr_cpu_ids <= BITS_PER_COMPAT_LONG)
min_length = sizeof(compat_ulong_t);
if (len < min_length)
return -EINVAL;
if (!alloc_cpumask_var(&mask, GFP_KERNEL))
return -ENOMEM;
ret = sched_getaffinity(pid, mask);
if (ret < 0)
goto out;
k = cpumask_bits(mask);
ret = compat_put_bitmap(user_mask_ptr, k, min_length * 8);
if (ret == 0)
ret = min_length;
out:
free_cpumask_var(mask);
return ret;
}
int get_compat_itimerspec(struct itimerspec *dst,
const struct compat_itimerspec __user *src)
{
if (get_compat_timespec(&dst->it_interval, &src->it_interval) ||
get_compat_timespec(&dst->it_value, &src->it_value))
return -EFAULT;
return 0;
}
int put_compat_itimerspec(struct compat_itimerspec __user *dst,
const struct itimerspec *src)
{
if (put_compat_timespec(&src->it_interval, &dst->it_interval) ||
put_compat_timespec(&src->it_value, &dst->it_value))
return -EFAULT;
return 0;
}
long compat_sys_timer_create(clockid_t which_clock,
struct compat_sigevent __user *timer_event_spec,
timer_t __user *created_timer_id)
{
struct sigevent __user *event = NULL;
if (timer_event_spec) {
struct sigevent kevent;
event = compat_alloc_user_space(sizeof(*event));
if (get_compat_sigevent(&kevent, timer_event_spec) ||
copy_to_user(event, &kevent, sizeof(*event)))
return -EFAULT;
}
return sys_timer_create(which_clock, event, created_timer_id);
}
long compat_sys_timer_settime(timer_t timer_id, int flags,
struct compat_itimerspec __user *new,
struct compat_itimerspec __user *old)
{
long err;
mm_segment_t oldfs;
struct itimerspec newts, oldts;
if (!new)
return -EINVAL;
if (get_compat_itimerspec(&newts, new))
return -EFAULT;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_timer_settime(timer_id, flags,
(struct itimerspec __user *) &newts,
(struct itimerspec __user *) &oldts);
set_fs(oldfs);
if (!err && old && put_compat_itimerspec(old, &oldts))
return -EFAULT;
return err;
}
long compat_sys_timer_gettime(timer_t timer_id,
struct compat_itimerspec __user *setting)
{
long err;
mm_segment_t oldfs;
struct itimerspec ts;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_timer_gettime(timer_id,
(struct itimerspec __user *) &ts);
set_fs(oldfs);
if (!err && put_compat_itimerspec(setting, &ts))
return -EFAULT;
return err;
}
long compat_sys_clock_settime(clockid_t which_clock,
struct compat_timespec __user *tp)
{
long err;
mm_segment_t oldfs;
struct timespec ts;
if (get_compat_timespec(&ts, tp))
return -EFAULT;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_clock_settime(which_clock,
(struct timespec __user *) &ts);
set_fs(oldfs);
return err;
}
long compat_sys_clock_gettime(clockid_t which_clock,
struct compat_timespec __user *tp)
{
long err;
mm_segment_t oldfs;
struct timespec ts;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_clock_gettime(which_clock,
(struct timespec __user *) &ts);
set_fs(oldfs);
if (!err && put_compat_timespec(&ts, tp))
return -EFAULT;
return err;
}
long compat_sys_clock_getres(clockid_t which_clock,
struct compat_timespec __user *tp)
{
long err;
mm_segment_t oldfs;
struct timespec ts;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_clock_getres(which_clock,
(struct timespec __user *) &ts);
set_fs(oldfs);
if (!err && tp && put_compat_timespec(&ts, tp))
return -EFAULT;
return err;
}
static long compat_clock_nanosleep_restart(struct restart_block *restart)
{
long err;
mm_segment_t oldfs;
struct timespec tu;
struct compat_timespec *rmtp = restart->nanosleep.compat_rmtp;
restart->nanosleep.rmtp = (struct timespec __user *) &tu;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = clock_nanosleep_restart(restart);
set_fs(oldfs);
if ((err == -ERESTART_RESTARTBLOCK) && rmtp &&
put_compat_timespec(&tu, rmtp))
return -EFAULT;
if (err == -ERESTART_RESTARTBLOCK) {
restart->fn = compat_clock_nanosleep_restart;
restart->nanosleep.compat_rmtp = rmtp;
}
return err;
}
long compat_sys_clock_nanosleep(clockid_t which_clock, int flags,
struct compat_timespec __user *rqtp,
struct compat_timespec __user *rmtp)
{
long err;
mm_segment_t oldfs;
struct timespec in, out;
struct restart_block *restart;
if (get_compat_timespec(&in, rqtp))
return -EFAULT;
oldfs = get_fs();
set_fs(KERNEL_DS);
err = sys_clock_nanosleep(which_clock, flags,
(struct timespec __user *) &in,
(struct timespec __user *) &out);
set_fs(oldfs);
if ((err == -ERESTART_RESTARTBLOCK) && rmtp &&
put_compat_timespec(&out, rmtp))
return -EFAULT;
if (err == -ERESTART_RESTARTBLOCK) {
restart = &current_thread_info()->restart_block;
restart->fn = compat_clock_nanosleep_restart;
restart->nanosleep.compat_rmtp = rmtp;
}
return err;
}
/*
* We currently only need the following fields from the sigevent
* structure: sigev_value, sigev_signo, sig_notify and (sometimes
* sigev_notify_thread_id). The others are handled in user mode.
* We also assume that copying sigev_value.sival_int is sufficient
* to keep all the bits of sigev_value.sival_ptr intact.
*/
int get_compat_sigevent(struct sigevent *event,
const struct compat_sigevent __user *u_event)
{
memset(event, 0, sizeof(*event));
return (!access_ok(VERIFY_READ, u_event, sizeof(*u_event)) ||
__get_user(event->sigev_value.sival_int,
&u_event->sigev_value.sival_int) ||
__get_user(event->sigev_signo, &u_event->sigev_signo) ||
__get_user(event->sigev_notify, &u_event->sigev_notify) ||
__get_user(event->sigev_notify_thread_id,
&u_event->sigev_notify_thread_id))
? -EFAULT : 0;
}
long compat_get_bitmap(unsigned long *mask, const compat_ulong_t __user *umask,
unsigned long bitmap_size)
{
int i, j;
unsigned long m;
compat_ulong_t um;
unsigned long nr_compat_longs;
/* align bitmap up to nearest compat_long_t boundary */
bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
if (!access_ok(VERIFY_READ, umask, bitmap_size / 8))
return -EFAULT;
nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
for (i = 0; i < BITS_TO_LONGS(bitmap_size); i++) {
m = 0;
for (j = 0; j < sizeof(m)/sizeof(um); j++) {
/*
* We dont want to read past the end of the userspace
* bitmap. We must however ensure the end of the
* kernel bitmap is zeroed.
*/
if (nr_compat_longs-- > 0) {
if (__get_user(um, umask))
return -EFAULT;
} else {
um = 0;
}
umask++;
m |= (long)um << (j * BITS_PER_COMPAT_LONG);
}
*mask++ = m;
}
return 0;
}
long compat_put_bitmap(compat_ulong_t __user *umask, unsigned long *mask,
unsigned long bitmap_size)
{
int i, j;
unsigned long m;
compat_ulong_t um;
unsigned long nr_compat_longs;
/* align bitmap up to nearest compat_long_t boundary */
bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
if (!access_ok(VERIFY_WRITE, umask, bitmap_size / 8))
return -EFAULT;
nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
for (i = 0; i < BITS_TO_LONGS(bitmap_size); i++) {
m = *mask++;
for (j = 0; j < sizeof(m)/sizeof(um); j++) {
um = m;
/*
* We dont want to write past the end of the userspace
* bitmap.
*/
if (nr_compat_longs-- > 0) {
if (__put_user(um, umask))
return -EFAULT;
}
umask++;
m >>= 4*sizeof(um);
m >>= 4*sizeof(um);
}
}
return 0;
}
void
sigset_from_compat (sigset_t *set, compat_sigset_t *compat)
{
switch (_NSIG_WORDS) {
case 4: set->sig[3] = compat->sig[6] | (((long)compat->sig[7]) << 32 );
case 3: set->sig[2] = compat->sig[4] | (((long)compat->sig[5]) << 32 );
case 2: set->sig[1] = compat->sig[2] | (((long)compat->sig[3]) << 32 );
case 1: set->sig[0] = compat->sig[0] | (((long)compat->sig[1]) << 32 );
}
}
asmlinkage long
compat_sys_rt_sigtimedwait (compat_sigset_t __user *uthese,
struct compat_siginfo __user *uinfo,
struct compat_timespec __user *uts, compat_size_t sigsetsize)
{
compat_sigset_t s32;
sigset_t s;
int sig;
struct timespec t;
siginfo_t info;
long ret, timeout = 0;
if (sigsetsize != sizeof(sigset_t))
return -EINVAL;
if (copy_from_user(&s32, uthese, sizeof(compat_sigset_t)))
return -EFAULT;
sigset_from_compat(&s, &s32);
sigdelsetmask(&s,sigmask(SIGKILL)|sigmask(SIGSTOP));
signotset(&s);
if (uts) {
if (get_compat_timespec (&t, uts))
return -EFAULT;
if (t.tv_nsec >= 1000000000L || t.tv_nsec < 0
|| t.tv_sec < 0)
return -EINVAL;
}
spin_lock_irq(&current->sighand->siglock);
sig = dequeue_signal(current, &s, &info);
if (!sig) {
timeout = MAX_SCHEDULE_TIMEOUT;
if (uts)
timeout = timespec_to_jiffies(&t)
+(t.tv_sec || t.tv_nsec);
if (timeout) {
current->real_blocked = current->blocked;
sigandsets(&current->blocked, &current->blocked, &s);
recalc_sigpending();
spin_unlock_irq(&current->sighand->siglock);
timeout = schedule_timeout_interruptible(timeout);
spin_lock_irq(&current->sighand->siglock);
sig = dequeue_signal(current, &s, &info);
current->blocked = current->real_blocked;
siginitset(&current->real_blocked, 0);
recalc_sigpending();
}
}
spin_unlock_irq(&current->sighand->siglock);
if (sig) {
ret = sig;
if (uinfo) {
if (copy_siginfo_to_user32(uinfo, &info))
ret = -EFAULT;
}
}else {
ret = timeout?-EINTR:-EAGAIN;
}
return ret;
}
#ifdef __ARCH_WANT_COMPAT_SYS_TIME
/* compat_time_t is a 32 bit "long" and needs to get converted. */
asmlinkage long compat_sys_time(compat_time_t __user * tloc)
{
compat_time_t i;
struct timeval tv;
do_gettimeofday(&tv);
i = tv.tv_sec;
if (tloc) {
if (put_user(i,tloc))
Allow times and time system calls to return small negative values At the moment, the times() system call will appear to fail for a period shortly after boot, while the value it want to return is between -4095 and -1. The same thing will also happen for the time() system call on 32-bit platforms some time in 2106 or so. On some platforms, such as x86, this is unavoidable because of the system call ABI, but other platforms such as powerpc have a separate error indication from the return value, so system calls can in fact return small negative values without indicating an error. On those platforms, force_successful_syscall_return() provides a way to indicate that the system call return value should not be treated as an error even if it is in the range which would normally be taken as a negative error number. This adds a force_successful_syscall_return() call to the time() and times() system calls plus their 32-bit compat versions, so that they don't erroneously indicate an error on those platforms whose system call ABI has a separate error indication. This will not affect anything on other platforms. Joakim Tjernlund added the fix for time() and the compat versions of time() and times(), after I did the fix for times(). Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se> Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: David S. Miller <davem@davemloft.net> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-07 06:41:02 +08:00
return -EFAULT;
}
Allow times and time system calls to return small negative values At the moment, the times() system call will appear to fail for a period shortly after boot, while the value it want to return is between -4095 and -1. The same thing will also happen for the time() system call on 32-bit platforms some time in 2106 or so. On some platforms, such as x86, this is unavoidable because of the system call ABI, but other platforms such as powerpc have a separate error indication from the return value, so system calls can in fact return small negative values without indicating an error. On those platforms, force_successful_syscall_return() provides a way to indicate that the system call return value should not be treated as an error even if it is in the range which would normally be taken as a negative error number. This adds a force_successful_syscall_return() call to the time() and times() system calls plus their 32-bit compat versions, so that they don't erroneously indicate an error on those platforms whose system call ABI has a separate error indication. This will not affect anything on other platforms. Joakim Tjernlund added the fix for time() and the compat versions of time() and times(), after I did the fix for times(). Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se> Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: David S. Miller <davem@davemloft.net> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-07 06:41:02 +08:00
force_successful_syscall_return();
return i;
}
asmlinkage long compat_sys_stime(compat_time_t __user *tptr)
{
struct timespec tv;
int err;
if (get_user(tv.tv_sec, tptr))
return -EFAULT;
tv.tv_nsec = 0;
err = security_settime(&tv, NULL);
if (err)
return err;
do_settimeofday(&tv);
return 0;
}
#endif /* __ARCH_WANT_COMPAT_SYS_TIME */
#ifdef __ARCH_WANT_COMPAT_SYS_RT_SIGSUSPEND
asmlinkage long compat_sys_rt_sigsuspend(compat_sigset_t __user *unewset, compat_size_t sigsetsize)
{
sigset_t newset;
compat_sigset_t newset32;
/* XXX: Don't preclude handling different sized sigset_t's. */
if (sigsetsize != sizeof(sigset_t))
return -EINVAL;
if (copy_from_user(&newset32, unewset, sizeof(compat_sigset_t)))
return -EFAULT;
sigset_from_compat(&newset, &newset32);
sigdelsetmask(&newset, sigmask(SIGKILL)|sigmask(SIGSTOP));
spin_lock_irq(&current->sighand->siglock);
current->saved_sigmask = current->blocked;
current->blocked = newset;
recalc_sigpending();
spin_unlock_irq(&current->sighand->siglock);
current->state = TASK_INTERRUPTIBLE;
schedule();
set_restore_sigmask();
return -ERESTARTNOHAND;
}
#endif /* __ARCH_WANT_COMPAT_SYS_RT_SIGSUSPEND */
asmlinkage long compat_sys_adjtimex(struct compat_timex __user *utp)
{
struct timex txc;
int ret;
memset(&txc, 0, sizeof(struct timex));
if (!access_ok(VERIFY_READ, utp, sizeof(struct compat_timex)) ||
__get_user(txc.modes, &utp->modes) ||
__get_user(txc.offset, &utp->offset) ||
__get_user(txc.freq, &utp->freq) ||
__get_user(txc.maxerror, &utp->maxerror) ||
__get_user(txc.esterror, &utp->esterror) ||
__get_user(txc.status, &utp->status) ||
__get_user(txc.constant, &utp->constant) ||
__get_user(txc.precision, &utp->precision) ||
__get_user(txc.tolerance, &utp->tolerance) ||
__get_user(txc.time.tv_sec, &utp->time.tv_sec) ||
__get_user(txc.time.tv_usec, &utp->time.tv_usec) ||
__get_user(txc.tick, &utp->tick) ||
__get_user(txc.ppsfreq, &utp->ppsfreq) ||
__get_user(txc.jitter, &utp->jitter) ||
__get_user(txc.shift, &utp->shift) ||
__get_user(txc.stabil, &utp->stabil) ||
__get_user(txc.jitcnt, &utp->jitcnt) ||
__get_user(txc.calcnt, &utp->calcnt) ||
__get_user(txc.errcnt, &utp->errcnt) ||
__get_user(txc.stbcnt, &utp->stbcnt))
return -EFAULT;
ret = do_adjtimex(&txc);
if (!access_ok(VERIFY_WRITE, utp, sizeof(struct compat_timex)) ||
__put_user(txc.modes, &utp->modes) ||
__put_user(txc.offset, &utp->offset) ||
__put_user(txc.freq, &utp->freq) ||
__put_user(txc.maxerror, &utp->maxerror) ||
__put_user(txc.esterror, &utp->esterror) ||
__put_user(txc.status, &utp->status) ||
__put_user(txc.constant, &utp->constant) ||
__put_user(txc.precision, &utp->precision) ||
__put_user(txc.tolerance, &utp->tolerance) ||
__put_user(txc.time.tv_sec, &utp->time.tv_sec) ||
__put_user(txc.time.tv_usec, &utp->time.tv_usec) ||
__put_user(txc.tick, &utp->tick) ||
__put_user(txc.ppsfreq, &utp->ppsfreq) ||
__put_user(txc.jitter, &utp->jitter) ||
__put_user(txc.shift, &utp->shift) ||
__put_user(txc.stabil, &utp->stabil) ||
__put_user(txc.jitcnt, &utp->jitcnt) ||
__put_user(txc.calcnt, &utp->calcnt) ||
__put_user(txc.errcnt, &utp->errcnt) ||
__put_user(txc.stbcnt, &utp->stbcnt) ||
__put_user(txc.tai, &utp->tai))
ret = -EFAULT;
return ret;
}
#ifdef CONFIG_NUMA
asmlinkage long compat_sys_move_pages(pid_t pid, unsigned long nr_pages,
compat_uptr_t __user *pages32,
const int __user *nodes,
int __user *status,
int flags)
{
const void __user * __user *pages;
int i;
pages = compat_alloc_user_space(nr_pages * sizeof(void *));
for (i = 0; i < nr_pages; i++) {
compat_uptr_t p;
if (get_user(p, pages32 + i) ||
put_user(compat_ptr(p), pages + i))
return -EFAULT;
}
return sys_move_pages(pid, nr_pages, pages, nodes, status, flags);
}
asmlinkage long compat_sys_migrate_pages(compat_pid_t pid,
compat_ulong_t maxnode,
const compat_ulong_t __user *old_nodes,
const compat_ulong_t __user *new_nodes)
{
unsigned long __user *old = NULL;
unsigned long __user *new = NULL;
nodemask_t tmp_mask;
unsigned long nr_bits;
unsigned long size;
nr_bits = min_t(unsigned long, maxnode - 1, MAX_NUMNODES);
size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
if (old_nodes) {
if (compat_get_bitmap(nodes_addr(tmp_mask), old_nodes, nr_bits))
return -EFAULT;
old = compat_alloc_user_space(new_nodes ? size * 2 : size);
if (new_nodes)
new = old + size / sizeof(unsigned long);
if (copy_to_user(old, nodes_addr(tmp_mask), size))
return -EFAULT;
}
if (new_nodes) {
if (compat_get_bitmap(nodes_addr(tmp_mask), new_nodes, nr_bits))
return -EFAULT;
if (new == NULL)
new = compat_alloc_user_space(size);
if (copy_to_user(new, nodes_addr(tmp_mask), size))
return -EFAULT;
}
return sys_migrate_pages(pid, nr_bits + 1, old, new);
}
#endif
struct compat_sysinfo {
s32 uptime;
u32 loads[3];
u32 totalram;
u32 freeram;
u32 sharedram;
u32 bufferram;
u32 totalswap;
u32 freeswap;
u16 procs;
u16 pad;
u32 totalhigh;
u32 freehigh;
u32 mem_unit;
char _f[20-2*sizeof(u32)-sizeof(int)];
};
asmlinkage long
compat_sys_sysinfo(struct compat_sysinfo __user *info)
{
struct sysinfo s;
do_sysinfo(&s);
/* Check to see if any memory value is too large for 32-bit and scale
* down if needed
*/
if ((s.totalram >> 32) || (s.totalswap >> 32)) {
int bitcount = 0;
while (s.mem_unit < PAGE_SIZE) {
s.mem_unit <<= 1;
bitcount++;
}
s.totalram >>= bitcount;
s.freeram >>= bitcount;
s.sharedram >>= bitcount;
s.bufferram >>= bitcount;
s.totalswap >>= bitcount;
s.freeswap >>= bitcount;
s.totalhigh >>= bitcount;
s.freehigh >>= bitcount;
}
if (!access_ok(VERIFY_WRITE, info, sizeof(struct compat_sysinfo)) ||
__put_user (s.uptime, &info->uptime) ||
__put_user (s.loads[0], &info->loads[0]) ||
__put_user (s.loads[1], &info->loads[1]) ||
__put_user (s.loads[2], &info->loads[2]) ||
__put_user (s.totalram, &info->totalram) ||
__put_user (s.freeram, &info->freeram) ||
__put_user (s.sharedram, &info->sharedram) ||
__put_user (s.bufferram, &info->bufferram) ||
__put_user (s.totalswap, &info->totalswap) ||
__put_user (s.freeswap, &info->freeswap) ||
__put_user (s.procs, &info->procs) ||
__put_user (s.totalhigh, &info->totalhigh) ||
__put_user (s.freehigh, &info->freehigh) ||
__put_user (s.mem_unit, &info->mem_unit))
return -EFAULT;
return 0;
}