Re: [PATCH mmotm] memcg: avoid using buggy kmap at swap_cgroup

["KAMEZAWA Hiroyuki" <kamezawa.hiroyu@jp.fujitsu.com>]

Hugh Dickins said:
> On Tue, 18 Nov 2008, Hugh Dickins wrote:
>> On Wed, 19 Nov 2008, KAMEZAWA Hiroyuki wrote:
>>
>> >  2. later, add kmap_atomic + HighMem buffer support in explicit style.
>> >     maybe KM_BOUNCE_READ...can be used.....
>>
>> It's hardly appropriate (there's no bouncing here), and you could only
>> use it if you disable interrupts.  Oh, you do disable interrupts:
>> why's that?
>
> In fact, why do you even need the spinlock?  I can see that you would
> need it if in future you reduce the size of the elements of the array
> from pointers; but at present, aren't you already in trouble if there's
> a race on the pointer?
>
Hmm, I originally added it just for doing exchange-entry.
Now, lookup and exchange operation is implemented for swap_cgroup.

This field is touched when
  - try to map swap cache (try_charge_swapin) -> lookup
  - after swapcache is mapped -> exchange
  - swap cache is read by shmem -> lookup/exchange
  - swap cache is dropped -> exchange
  - swap entry is freed. -> exchange
  ...
Hmm.....
  When accessed via SwapCache -> SwapCache is locked -> no race..
  When accessed vid swap_free -> no user of swap -> no race....

Then, maybe lock is not needed...I'll review and prepare rework patch
for patch-in-mmotn+fix1234.

Thank you for pointing out.
-Kame





> Hugh
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org.  For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
>


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>