linux/kernel/sys.c, branch v2.6.38

Fix prlimit64 for suid/sgid processes

2011-01-31T03:01:27Z

Since check_prlimit_permission always fails in the case of SUID/GUID processes, such processes are not able to read or set their own limits. This commit changes this by assuming that process can always read/change its own limits. Signed-off-by: Kacper Kornet Acked-by: Jiri Slaby Signed-off-by: Linus Torvalds

kmsg_dump: add kmsg_dump() calls to the reboot, halt, poweroff and emergency_restart paths

2011-01-13T16:03:07Z

We need to know the reason why system rebooted in support service. However, we can't inform our customers of the reason because final messages are lost on current Linux kernel. This patch improves the situation above because the final messages are saved by adding kmsg_dump() to reboot, halt, poweroff and emergency_restart path. Signed-off-by: Seiji Aguchi Cc: David Woodhouse Cc: Marco Stornelli Reviewed-by: Artem Bityutskiy Reviewed-by: KOSAKI Motohiro Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

sched: Add 'autogroup' scheduling feature: automated per session task groups

2010-11-30T15:03:35Z

A recurring complaint from CFS users is that parallel kbuild has a negative impact on desktop interactivity. This patch implements an idea from Linus, to automatically create task groups. Currently, only per session autogroups are implemented, but the patch leaves the way open for enhancement. Implementation: each task's signal struct contains an inherited pointer to a refcounted autogroup struct containing a task group pointer, the default for all tasks pointing to the init_task_group. When a task calls setsid(), a new task group is created, the process is moved into the new task group, and a reference to the preveious task group is dropped. Child processes inherit this task group thereafter, and increase it's refcount. When the last thread of a process exits, the process's reference is dropped, such that when the last process referencing an autogroup exits, the autogroup is destroyed. At runqueue selection time, IFF a task has no cgroup assignment, its current autogroup is used. Autogroup bandwidth is controllable via setting it's nice level through the proc filesystem: cat /proc//autogroup Displays the task's group and the group's nice level. echo > /proc//autogroup Sets the task group's shares to the weight of nice task. Setting nice level is rate limited for !admin users due to the abuse risk of task group locking. The feature is enabled from boot by default if CONFIG_SCHED_AUTOGROUP=y is selected, but can be disabled via the boot option noautogroup, and can also be turned on/off on the fly via: echo [01] > /proc/sys/kernel/sched_autogroup_enabled ... which will automatically move tasks to/from the root task group. Signed-off-by: Mike Galbraith Acked-by: Linus Torvalds Acked-by: Peter Zijlstra Cc: Markus Trippelsdorf Cc: Mathieu Desnoyers Cc: Paul Turner Cc: Oleg Nesterov [ Removed the task_group_path() debug code, and fixed !EVENTFD build failure. ] Signed-off-by: Ingo Molnar LKML-Reference: <1290281700.28711.9.camel@maggy.simson.net> Signed-off-by: Ingo Molnar

pid: make setpgid() system call use RCU read-side critical section

2010-09-01T00:00:18Z

[ 23.584719] [ 23.584720] =================================================== [ 23.585059] [ INFO: suspicious rcu_dereference_check() usage. ] [ 23.585176] --------------------------------------------------- [ 23.585176] kernel/pid.c:419 invoked rcu_dereference_check() without protection! [ 23.585176] [ 23.585176] other info that might help us debug this: [ 23.585176] [ 23.585176] [ 23.585176] rcu_scheduler_active = 1, debug_locks = 1 [ 23.585176] 1 lock held by rc.sysinit/728: [ 23.585176] #0: (tasklist_lock){.+.+..}, at: [] sys_setpgid+0x5f/0x193 [ 23.585176] [ 23.585176] stack backtrace: [ 23.585176] Pid: 728, comm: rc.sysinit Not tainted 2.6.36-rc2 #2 [ 23.585176] Call Trace: [ 23.585176] [] lockdep_rcu_dereference+0x99/0xa2 [ 23.585176] [] find_task_by_pid_ns+0x50/0x6a [ 23.585176] [] find_task_by_vpid+0x1d/0x1f [ 23.585176] [] sys_setpgid+0x67/0x193 [ 23.585176] [] system_call_fastpath+0x16/0x1b [ 24.959669] type=1400 audit(1282938522.956:4): avc: denied { module_request } for pid=766 comm="hwclock" kmod="char-major-10-135" scontext=system_u:system_r:hwclock_t:s0 tcontext=system_u:system_r:kernel_t:s0 tclas It turns out that the setpgid() system call fails to enter an RCU read-side critical section before doing a PID-to-task_struct translation. This commit therefore does rcu_read_lock() before the translation, and also does rcu_read_unlock() after the last use of the returned pointer. Reported-by: Andrew Morton Signed-off-by: Paul E. McKenney Acked-by: David Howells

rlimits: implement prlimit64 syscall

2010-07-16T07:48:48Z

This patch adds the code to support the sys_prlimit64 syscall which modifies-and-returns the rlim values of a selected process atomically. The first parameter, pid, being 0 means current process. Unlike the current implementation, it is a generic interface, architecture indepentent so that we needn't handle compat stuff anymore. In the future, after glibc start to use this we can deprecate sys_setrlimit and sys_getrlimit in favor to clean up the code finally. It also adds a possibility of changing limits of other processes. We check the user's permissions to do that and if it succeeds, the new limits are propagated online. This is good for large scale applications such as SAP or databases where administrators need to change limits time by time (e.g. on crashes increase core size). And it is unacceptable to restart the service. For safety, all rlim users now either use accessors or doesn't need them due to - locking - the fact a process was just forked and nobody else knows about it yet (and nobody can't thus read/write limits) hence it is safe to modify limits now. The limitation is that we currently stay at ulong internal representation. So the rlim64_is_infinity check is used where value is compared against ULONG_MAX on 32-bit which is the maximum value there. And since internally the limits are held in struct rlimit, converters which are used before and after do_prlimit call in sys_prlimit64 are introduced. Signed-off-by: Jiri Slaby

rlimits: switch more rlimit syscalls to do_prlimit

2010-07-16T07:48:48Z

After we added more generic do_prlimit, switch sys_getrlimit to that. Also switch compat handling, so we can get rid of ugly __user casts and avoid setting process' address limit to kernel data and back. Signed-off-by: Jiri Slaby

rlimits: redo do_setrlimit to more generic do_prlimit

2010-07-16T07:48:48Z

It now allows also reading of limits. I.e. all read and writes will later use this function. It takes two parameters, new and old limits which can be both NULL. If new is non-NULL, the value in it is set to rlimits. If old is non-NULL, current rlimits are stored there. If both are non-NULL, old are stored prior to setting the new ones, atomically. (Similar to sigaction.) Signed-off-by: Jiri Slaby

rlimits: do security check under task_lock

2010-07-16T07:48:47Z

Do security_task_setrlimit under task_lock. Other tasks may change limits under our hands while we are checking limits inside the function. From now on, they can't. Note that all the security work is done under a spinlock here now. Security hooks count with that, they are called from interrupt context (like security_task_kill) and with spinlocks already held (e.g. capable->security_capable). Signed-off-by: Jiri Slaby Acked-by: James Morris Cc: Heiko Carstens

rlimits: allow setrlimit to non-current tasks

2010-07-16T07:48:47Z

Add locking to allow setrlimit accept task parameter other than current. Namely, lock tasklist_lock for read and check whether the task structure has sighand non-null. Do all the signal processing under that lock still held. There are some points: 1) security_task_setrlimit is now called with that lock held. This is not new, many security_* functions are called with this lock held already so it doesn't harm (all this security_* stuff does almost the same). 2) task->sighand->siglock (in update_rlimit_cpu) is nested in tasklist_lock. This dependence is already existing. 3) tsk->alloc_lock is nested in tasklist_lock. This is OK too, already existing dependence. Signed-off-by: Jiri Slaby Cc: Oleg Nesterov

rlimits: split sys_setrlimit

2010-07-16T07:48:46Z

Create do_setrlimit from sys_setrlimit and declare do_setrlimit in the resource header. This is the first phase to have generic do_prlimit which allows to be called from read, write and compat rlimits code. The new do_setrlimit also accepts a task pointer to change the limits of. Currently, it cannot be other than current, but this will change with locking later. Also pass tsk->group_leader to security_task_setrlimit to check whether current is allowed to change rlimits of the process and not its arbitrary thread because it makes more sense given that rlimit are per process and not per-thread. Signed-off-by: Jiri Slaby