Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2017-04-22 Thread zhong jiang
Hi, Dashi The same issue I had occured every other week. Do you have solve it . I want to know how it is fixed. The patch exist in the mainline. Thanks zhongjiang On 2016/12/23 10:38, Dashi DS1 Cao wrote: > I'd expected that one or more tasks doing the free were the current task of > other

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2017-04-22 Thread zhong jiang
Hi, Dashi The same issue I had occured every other week. Do you have solve it . I want to know how it is fixed. The patch exist in the mainline. Thanks zhongjiang On 2016/12/23 10:38, Dashi DS1 Cao wrote: > I'd expected that one or more tasks doing the free were the current task of > other

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-24 Thread Dashi DS1 Cao
It's a CentOS 7.2, so there is point in asking RHT. I'll try to persuade the customer to have a try with kernel version 4.9, if only I can get it work with CentOS 7.2. Dashi Cao -Original Message- From: Peter Zijlstra [mailto:pet...@infradead.org] Sent: Friday, December 23, 2016 10:20

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-24 Thread Dashi DS1 Cao
It's a CentOS 7.2, so there is point in asking RHT. I'll try to persuade the customer to have a try with kernel version 4.9, if only I can get it work with CentOS 7.2. Dashi Cao -Original Message- From: Peter Zijlstra [mailto:pet...@infradead.org] Sent: Friday, December 23, 2016 10:20

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-23 Thread Peter Zijlstra
On Fri, Dec 23, 2016 at 02:02:14AM +, Dashi DS1 Cao wrote: > The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the > latest kernel release of CentOS 7.2 at that time, or maybe still now. This would be the point where we ask you to run a recent upstream kernel and try and

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-23 Thread Peter Zijlstra
On Fri, Dec 23, 2016 at 02:02:14AM +, Dashi DS1 Cao wrote: > The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the > latest kernel release of CentOS 7.2 at that time, or maybe still now. This would be the point where we ask you to run a recent upstream kernel and try and

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Hugh Dickins
On Fri, 23 Dec 2016, Dashi DS1 Cao wrote: > The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the > latest kernel release of CentOS 7.2 at that time, or maybe still now. Okay, thanks: so, basically a v3.10 kernel, with lots of added patches, but also lacking many more recent

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Hugh Dickins
On Fri, 23 Dec 2016, Dashi DS1 Cao wrote: > The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the > latest kernel release of CentOS 7.2 at that time, or maybe still now. Okay, thanks: so, basically a v3.10 kernel, with lots of added patches, but also lacking many more recent

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
I'd expected that one or more tasks doing the free were the current task of other cpu cores, but only one of the four dumps has two swapd task that are concurrently at execution on different cpu. This is the task leading to the crash: PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND:

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
I'd expected that one or more tasks doing the free were the current task of other cpu cores, but only one of the four dumps has two swapd task that are concurrently at execution on different cpu. This is the task leading to the crash: PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND:

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the latest kernel release of CentOS 7.2 at that time, or maybe still now. I've tried to print the value of anon_vma from other three dumps, but the content is not available in the dump. "gdb: page excluded: kernel virtual

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
The kernel version is "RELEASE: 3.10.0-327.36.3.el7.x86_64". It was the latest kernel release of CentOS 7.2 at that time, or maybe still now. I've tried to print the value of anon_vma from other three dumps, but the content is not available in the dump. "gdb: page excluded: kernel virtual

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Hugh Dickins
On Thu, 22 Dec 2016, Peter Zijlstra wrote: > On Wed, Dec 21, 2016 at 03:43:43PM +0100, Michal Hocko wrote: > > anon_vma locking is clever^Wsubtle as hell. CC Peter... > > > > On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > > > I've collected four crash dumps with similar backtrace. > > > > > >

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Hugh Dickins
On Thu, 22 Dec 2016, Peter Zijlstra wrote: > On Wed, Dec 21, 2016 at 03:43:43PM +0100, Michal Hocko wrote: > > anon_vma locking is clever^Wsubtle as hell. CC Peter... > > > > On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > > > I've collected four crash dumps with similar backtrace. > > > > > >

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Peter Zijlstra
On Wed, Dec 21, 2016 at 03:43:43PM +0100, Michal Hocko wrote: > anon_vma locking is clever^Wsubtle as hell. CC Peter... > > On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > > I've collected four crash dumps with similar backtrace. > > > > PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND:

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Peter Zijlstra
On Wed, Dec 21, 2016 at 03:43:43PM +0100, Michal Hocko wrote: > anon_vma locking is clever^Wsubtle as hell. CC Peter... > > On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > > I've collected four crash dumps with similar backtrace. > > > > PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND:

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
Value of anon_vma: print *((struct anon_vma *)0x8820833ed940) $2 = { root = 0x0, rwsem = { count = 0, wait_lock = { raw_lock = { { head_tail = 0, tickets = { head = 0, tail = 0 } } } },

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
Value of anon_vma: print *((struct anon_vma *)0x8820833ed940) $2 = { root = 0x0, rwsem = { count = 0, wait_lock = { raw_lock = { { head_tail = 0, tickets = { head = 0, tail = 0 } } } },

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Michal Hocko
On Thu 22-12-16 11:53:26, Dashi DS1 Cao wrote: > I've used another dump with similar backtrace. Please also dump the anon_vma of the page as well. -- Michal Hocko SUSE Labs

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Michal Hocko
On Thu 22-12-16 11:53:26, Dashi DS1 Cao wrote: > I've used another dump with similar backtrace. Please also dump the anon_vma of the page as well. -- Michal Hocko SUSE Labs

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
I've used another dump with similar backtrace. PID: 246TASK: 881fd27df300 CPU: 0 COMMAND: "kswapd0" #0 [881fcfb23748] machine_kexec at 81051e9b #1 [881fcfb237a8] crash_kexec at 810f27e2 #2 [881fcfb23878] oops_end at 8163f448 #3 [881fcfb238a0]

RE: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-22 Thread Dashi DS1 Cao
I've used another dump with similar backtrace. PID: 246TASK: 881fd27df300 CPU: 0 COMMAND: "kswapd0" #0 [881fcfb23748] machine_kexec at 81051e9b #1 [881fcfb237a8] crash_kexec at 810f27e2 #2 [881fcfb23878] oops_end at 8163f448 #3 [881fcfb238a0]

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-21 Thread Michal Hocko
anon_vma locking is clever^Wsubtle as hell. CC Peter... On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > I've collected four crash dumps with similar backtrace. > > PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND: "kswapd1" > #0 [881fcfad7978] machine_kexec at 81051e9b > #1

Re: A small window for a race condition in mm/rmap.c:page_lock_anon_vma_read

2016-12-21 Thread Michal Hocko
anon_vma locking is clever^Wsubtle as hell. CC Peter... On Tue 20-12-16 09:32:27, Dashi DS1 Cao wrote: > I've collected four crash dumps with similar backtrace. > > PID: 247TASK: 881fcfad8000 CPU: 14 COMMAND: "kswapd1" > #0 [881fcfad7978] machine_kexec at 81051e9b > #1