Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-26 Thread Mike Galbraith
On Sat, 2012-10-20 at 08:38 -0400, Mike Galbraith wrote: > So what I would do is either let the user decide once at boot, in which > case if off, creating groups would be stupid), or, just rip autogroup > completely out, since systemd is taking over the known universe anyway. I'm traveling, but

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-26 Thread Mike Galbraith
On Sat, 2012-10-20 at 08:38 -0400, Mike Galbraith wrote: So what I would do is either let the user decide once at boot, in which case if off, creating groups would be stupid), or, just rip autogroup completely out, since systemd is taking over the known universe anyway. I'm traveling, but

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-20 Thread Mike Galbraith
On Sat, 2012-10-20 at 14:42 +0800, Xiaotian Feng wrote: > On Fri, Oct 19, 2012 at 9:42 PM, Peter Zijlstra wrote: > > Always try and CC people who wrote the code.. > > > > On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: > >> There's a regression from commit 800d4d30, in

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-20 Thread Xiaotian Feng
On Fri, Oct 19, 2012 at 9:42 PM, Peter Zijlstra wrote: > Always try and CC people who wrote the code.. > > On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: >> There's a regression from commit 800d4d30, in autogroup_move_group() >> >> p->signal->autogroup = autogroup_kref_get(ag); >>

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-20 Thread Xiaotian Feng
On Fri, Oct 19, 2012 at 9:42 PM, Peter Zijlstra pet...@infradead.org wrote: Always try and CC people who wrote the code.. On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: There's a regression from commit 800d4d30, in autogroup_move_group() p-signal-autogroup =

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-20 Thread Mike Galbraith
On Sat, 2012-10-20 at 14:42 +0800, Xiaotian Feng wrote: On Fri, Oct 19, 2012 at 9:42 PM, Peter Zijlstra pet...@infradead.org wrote: Always try and CC people who wrote the code.. On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: There's a regression from commit 800d4d30, in

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-19 Thread Peter Zijlstra
Always try and CC people who wrote the code.. On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: > There's a regression from commit 800d4d30, in autogroup_move_group() > > p->signal->autogroup = autogroup_kref_get(ag); > > if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled)) >

[PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-19 Thread Xiaotian Feng
There's a regression from commit 800d4d30, in autogroup_move_group() p->signal->autogroup = autogroup_kref_get(ag); if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled)) goto out; ... out: autogroup_kref_put(prev); So kernel changed p's autogroup

[PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-19 Thread Xiaotian Feng
There's a regression from commit 800d4d30, in autogroup_move_group() p-signal-autogroup = autogroup_kref_get(ag); if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled)) goto out; ... out: autogroup_kref_put(prev); So kernel changed p's autogroup to

Re: [PATCH] sched, autogroup: fix kernel crashes caused by runtime disable autogroup

2012-10-19 Thread Peter Zijlstra
Always try and CC people who wrote the code.. On Fri, 2012-10-19 at 16:36 +0800, Xiaotian Feng wrote: There's a regression from commit 800d4d30, in autogroup_move_group() p-signal-autogroup = autogroup_kref_get(ag); if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled))