From 64ca5ab913f1594ef316556e65f5eae63ff50cee Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 4 Mar 2009 12:11:56 -0800 Subject: rcu: increment quiescent state counter in ksoftirqd() If a machine is flooded by network frames, a cpu can loop 100% of its time inside ksoftirqd() without calling schedule(). This can delay RCU grace period to insane values. Adding rcu_qsctr_inc() call in ksoftirqd() solves this problem. Paul: "This regression was a result of the recent change from "schedule()" to "cond_resched()", which got rid of that quiescent state in the common case where a reschedule is not needed". Signed-off-by: Eric Dumazet Reviewed-by: Paul E. McKenney Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/softirq.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/softirq.c b/kernel/softirq.c index bdbe9de9cd8d..9041ea7948fe 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -626,6 +626,7 @@ static int ksoftirqd(void * __bind_cpu) preempt_enable_no_resched(); cond_resched(); preempt_disable(); + rcu_qsctr_inc((long)__bind_cpu); } preempt_enable(); set_current_state(TASK_INTERRUPTIBLE); -- cgit v1.2.3 From 6d5b5acca9e566515ef3f1ed617e7295c4f94345 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Mon, 9 Mar 2009 13:31:59 +0100 Subject: Fix fixpoint divide exception in acct_update_integrals Frans Pop reported the crash below when running an s390 kernel under Hercules: Kernel BUG at 000738b4 verbose debug info unavailable! fixpoint divide exception: 0009 #1! SMP Modules linked in: nfs lockd nfs_acl sunrpc ctcm fsm tape_34xx cu3088 tape ccwgroup tape_class ext3 jbd mbcache dm_mirror dm_log dm_snapshot dm_mod dasd_eckd_mod dasd_mod CPU: 0 Not tainted 2.6.27.19 #13 Process awk (pid: 2069, task: 0f9ed9b8, ksp: 0f4f7d18) Krnl PSW : 070c1000 800738b4 (acct_update_integrals+0x4c/0x118) R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:1 PM:0 Krnl GPRS: 00000000 000007d0 7fffffff fffff830 00000000 ffffffff 00000002 0f9ed9b8 00000000 00008ca0 00000000 0f9ed9b8 0f9edda4 8007386e 0f4f7ec8 0f4f7e98 Krnl Code: 800738aa: a71807d0 lhi %r1,2000 800738ae: 8c200001 srdl %r2,1 800738b2: 1d21 dr %r2,%r1 >800738b4: 5810d10e l %r1,270(%r13) 800738b8: 1823 lr %r2,%r3 800738ba: 4130f060 la %r3,96(%r15) 800738be: 0de1 basr %r14,%r1 800738c0: 5800f060 l %r0,96(%r15) Call Trace: ( <000000000004fdea>! blocking_notifier_call_chain+0x1e/0x2c) <0000000000038502>! do_exit+0x106/0x7c0 <0000000000038c36>! do_group_exit+0x7a/0xb4 <0000000000038c8e>! SyS_exit_group+0x1e/0x30 <0000000000021c28>! sysc_do_restart+0x12/0x16 <0000000077e7e924>! 0x77e7e924 Reason for this is that cpu time accounting usually only happens from interrupt context, but acct_update_integrals gets also called from process context with interrupts enabled. So in acct_update_integrals we may end up with the following scenario: Between reading tsk->stime/tsk->utime and tsk->acct_timexpd an interrupt happens which updates accouting values. This causes acct_timexpd to be greater than the former stime + utime. The subsequent calculation of dtime = cputime_sub(time, tsk->acct_timexpd); will be negative and the division performed by cputime_to_jiffies(dtime) will generate an exception since the result won't fit into a 32 bit register. In order to fix this just always disable interrupts while accessing any of the accounting values. Reported by: Frans Pop Tested by: Frans Pop Cc: stable@kernel.org Cc: Martin Schwidefsky Signed-off-by: Heiko Carstens Signed-off-by: Linus Torvalds --- kernel/tsacct.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/tsacct.c b/kernel/tsacct.c index 43f891b05a4b..00d59d048edf 100644 --- a/kernel/tsacct.c +++ b/kernel/tsacct.c @@ -122,8 +122,10 @@ void acct_update_integrals(struct task_struct *tsk) if (likely(tsk->mm)) { cputime_t time, dtime; struct timeval value; + unsigned long flags; u64 delta; + local_irq_save(flags); time = tsk->stime + tsk->utime; dtime = cputime_sub(time, tsk->acct_timexpd); jiffies_to_timeval(cputime_to_jiffies(dtime), &value); @@ -131,10 +133,12 @@ void acct_update_integrals(struct task_struct *tsk) delta = delta * USEC_PER_SEC + value.tv_usec; if (delta == 0) - return; + goto out; tsk->acct_timexpd = time; tsk->acct_rss_mem1 += delta * get_mm_rss(tsk->mm); tsk->acct_vm_mem1 += delta * tsk->mm->total_vm; + out: + local_irq_restore(flags); } } -- cgit v1.2.3 From 2d5516cbb9daf7d0e342a2e3b0fc6f8c39a81205 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Mon, 2 Mar 2009 22:58:45 +0100 Subject: copy_process: fix CLONE_PARENT && parent_exec_id interaction CLONE_PARENT can fool the ->self_exec_id/parent_exec_id logic. If we re-use the old parent, we must also re-use ->parent_exec_id to make sure exit_notify() sees the right ->xxx_exec_id's when the CLONE_PARENT'ed task exits. Also, move down the "p->parent_exec_id = p->self_exec_id" thing, to place two different cases together. Signed-off-by: Oleg Nesterov Cc: Roland McGrath Cc: Andrew Morton Cc: David Howells Cc: Serge E. Hallyn Signed-off-by: Linus Torvalds --- kernel/fork.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index a66fbde20715..4854c2c4a82e 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1179,10 +1179,6 @@ static struct task_struct *copy_process(unsigned long clone_flags, #endif clear_all_latency_tracing(p); - /* Our parent execution domain becomes current domain - These must match for thread signalling to apply */ - p->parent_exec_id = p->self_exec_id; - /* ok, now we should be set up.. */ p->exit_signal = (clone_flags & CLONE_THREAD) ? -1 : (clone_flags & CSIGNAL); p->pdeath_signal = 0; @@ -1220,10 +1216,13 @@ static struct task_struct *copy_process(unsigned long clone_flags, set_task_cpu(p, smp_processor_id()); /* CLONE_PARENT re-uses the old parent */ - if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) + if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) { p->real_parent = current->real_parent; - else + p->parent_exec_id = current->parent_exec_id; + } else { p->real_parent = current; + p->parent_exec_id = current->self_exec_id; + } spin_lock(¤t->sighand->siglock); -- cgit v1.2.3 From be50b8342dead8cacf57d4839240106b225d31f5 Mon Sep 17 00:00:00 2001 From: Dhaval Giani Date: Tue, 10 Mar 2009 12:55:56 -0700 Subject: kernel/user.c: fix a memory leak when freeing up non-init usernamespaces users We were returning early in the sysfs directory cleanup function if the user belonged to a non init usernamespace. Due to this a lot of the cleanup was not done and we were left with a leak. Fix the leak. Reported-by: Serge Hallyn Signed-off-by: Dhaval Giani Acked-by: Serge Hallyn Tested-by: Serge Hallyn Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/user.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/user.c b/kernel/user.c index 6a9b696128c8..fbb300e6191f 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -286,14 +286,12 @@ int __init uids_sysfs_init(void) /* work function to remove sysfs directory for a user and free up * corresponding structures. */ -static void remove_user_sysfs_dir(struct work_struct *w) +static void cleanup_user_struct(struct work_struct *w) { struct user_struct *up = container_of(w, struct user_struct, work); unsigned long flags; int remove_user = 0; - if (up->user_ns != &init_user_ns) - return; /* Make uid_hash_remove() + sysfs_remove_file() + kobject_del() * atomic. */ @@ -312,9 +310,11 @@ static void remove_user_sysfs_dir(struct work_struct *w) if (!remove_user) goto done; - kobject_uevent(&up->kobj, KOBJ_REMOVE); - kobject_del(&up->kobj); - kobject_put(&up->kobj); + if (up->user_ns == &init_user_ns) { + kobject_uevent(&up->kobj, KOBJ_REMOVE); + kobject_del(&up->kobj); + kobject_put(&up->kobj); + } sched_destroy_user(up); key_put(up->uid_keyring); @@ -335,7 +335,7 @@ static void free_user(struct user_struct *up, unsigned long flags) atomic_inc(&up->__count); spin_unlock_irqrestore(&uidhash_lock, flags); - INIT_WORK(&up->work, remove_user_sysfs_dir); + INIT_WORK(&up->work, cleanup_user_struct); schedule_work(&up->work); } -- cgit v1.2.3