2011/10/2 Samuel Thibault <samuel.thiba...@gnu.org>:
> Sergio López, le Thu 29 Sep 2011 11:44:40 +0200, a écrit :
>> Sync'ing the pager before clearing that flag ensures that there aren't
>> dirty pages in the object before its termination.
>
> Do you mean that the writes will be completely avoided for objects which
> have actually been removed?
No, it just makes sure that all dirty pages have been sent back to the
pager before clearing the flag.
>> @@ -851,7 +851,10 @@ drop_pager_softrefs (struct node *node)
>> spin_unlock (&node_to_page_lock);
>>
>> if (MAY_CACHE && pager)
>> - pager_change_attributes (pager, 0, MEMORY_OBJECT_COPY_DELAY, 0);
>> + {
>> + pager_sync (pager, 1);
>
> Won't using wait=1 reduce performance due to waiting for completion?
Clearing the flag will, implicitily, generate a full object
synchronization, but with a higher cost, since pages are sent one by
one. By synchronizing the object explicitily (and waiting for
completion), all data is sent through m_o_lock_request, which is able
to send to the pager up to 32 pages at a time (DATA_WRITE_MAX).
