(cc'ing scheduler folks)

On Sun, Sep 20, 2015 at 10:45:25AM -0400, Sasha Levin wrote:

On 09/13/2015 02:59 PM, Tejun Heo wrote:

quoted

task_struct->memcg_oom is a sub-struct containing fields which are
used for async memcg oom handling.  Most task_struct fields aren't
packaged this way and it can lead to unnecessary alignment paddings.
This patch flattens it.

* task.memcg_oom.memcg          -> task.memcg_in_oom
* task.memcg_oom.gfp_mask	-> task.memcg_oom_gfp_mask
* task.memcg_oom.order          -> task.memcg_oom_order
* task.memcg_oom.may_oom        -> task.memcg_may_oom

...

I've started seeing these warnings:

[1598889.250160] WARNING: CPU: 3 PID: 11648 at include/linux/memcontrol.h:414 handle_mm_fault+0x1020/0x3fa0()

...

[1598892.247256] dump_stack (lib/dump_stack.c:52)
[1598892.249105] warn_slowpath_common (kernel/panic.c:448)
[1598892.253202] warn_slowpath_null (kernel/panic.c:482)
[1598892.255148] handle_mm_fault (include/linux/memcontrol.h:414 mm/memory.c:3430)
[1598892.268151] __do_page_fault (arch/x86/mm/fault.c:1239)
[1598892.269022] trace_do_page_fault (arch/x86/mm/fault.c:1331 include/linux/jump_label.h:133 include/linux/context_tracking_state.h:30 include/linux/context_tracking.h:46 arch/x86/mm/fault.c:1332)
[1598892.269894] do_async_page_fault (arch/x86/kernel/kvm.c:280)
[1598892.270792] async_page_fault (arch/x86/entry/entry_64.S:989)

Not sure if it's because of this patch or not, but I haven't seen them before.

So, the only way the patch could have caused the above is if someone
who isn't the task itself is writing to the bitfields while the task
is running.  Looking through the fields, ->sched_reset_on_fork seems a
bit suspicious.  __sched_setscheduler() looks like it can modify the
bit while the target task is running.  Peter, am I misreading the
code?

Thanks.

-- 
tejun