Re: [PATCH] mm: make PR_SET_THP_DISABLE immediately active

On June 2, 2017 10:50:59 PM GMT+03:00, Andrew Morton [off-list ref] wrote:

On Fri,  2 Jun 2017 18:03:22 +0300 "Mike Rapoport"
[off-list ref] wrote:

quoted

PR_SET_THP_DISABLE has a rather subtle semantic. It doesn't affect

any

quoted

existing mapping because it only updated mm->def_flags which is a

template

quoted

for new mappings. The mappings created after

prctl(PR_SET_THP_DISABLE) have

quoted

VM_NOHUGEPAGE flag set.  This can be quite surprising for all those
applications which do not do prctl(); fork() & exec() and want to

control

quoted

their own THP behavior.

Another usecase when the immediate semantic of the prctl might be

useful is

quoted

a combination of pre- and post-copy migration of containers with

CRIU.  In

quoted

this case CRIU populates a part of a memory region with data that was

saved

quoted

during the pre-copy stage. Afterwards, the region is registered with
userfaultfd and CRIU expects to get page faults for the parts of the

region

quoted

that were not yet populated. However, khugepaged collapses the pages

and

quoted

the expected page faults do not occur.

In more general case, the prctl(PR_SET_THP_DISABLE) could be used as

a

quoted

temporary mechanism for enabling/disabling THP process wide.

Implementation wise, a new MMF_DISABLE_THP flag is added. This flag

is

quoted

tested when decision whether to use huge pages is taken either during

page

quoted

fault of at the time of THP collapse.

It should be noted, that the new implementation makes

PR_SET_THP_DISABLE

quoted

master override to any per-VMA setting, which was not the case

previously.

quoted

Fixes: a0715cc22601 ("mm, thp: add VM_INIT_DEF_MASK and

PRCTL_THP_DISABLE")

"Fixes" is a bit strong.  I'd say "alters".  And significantly altering
the runtime behaviour of a three-year-old interface is rather a worry,
no?

Well, there are people that consider current behavior as bug :)
One can argue we alter the implementation​details and users should not rely on that...

Perhaps we should be adding new prctl modes to select this new
behaviour and leave the existing PR_SET_THP_DISABLE behaviour as-is?

-- 
Sincerely yours,
Mike.

--
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>