Discussion:
[Open64-devel] review request for changes to libhugetlbfs
Gilmore, Doug
2012-08-23 22:13:38 UTC
Permalink
I have two changes to libhugetlbfs that I would like to have reviewed.

The first is just a fix to eliminate a compiler warning.

The second is a change to the library to handle situations where a
huge page mapping would exhaust the huge page limit. Previously no
huge pages would be allocated (when mapping ELF segments this failure
would cause program termination). With these changes, huge pages are
allocated up to consume the huge page limit, but small pages are used
to satisfy the rest of the allocation.

Could a gatekeeper review these changes when they have the chance?

Thanks,

Doug
C. Bergström
2012-08-23 22:14:54 UTC
Permalink
Post by Gilmore, Doug
I have two changes to libhugetlbfs that I would like to have reviewed.
Do you have any intention to push this upstream as well? Is there any
plans to sync with upstream or at least cherry-pick some of their changes?

http://sourceforge.net/p/libhugetlbfs/code/commit_browser
Gilmore, Doug
2012-08-24 01:00:59 UTC
Permalink
-----Original Message-----
Sent: Thursday, August 23, 2012 3:15 PM
To: Gilmore, Doug
Cc: open64-devel
Subject: Re: [Open64-devel] review request for changes to libhugetlbfs
Post by Gilmore, Doug
I have two changes to libhugetlbfs that I would like to have reviewed.
Do you have any intention to push this upstream as well?
Don't know.

I have been hoping that the performance issues with Transparent Huge
Pages (THP) will be addressed in the near future thus there is no need
for libhugetlbfs (the performance gains for THP are more appreciable
running under a VM rather than on the hardware).

The development of libhugetlbfs has been pretty quiet lately,
so it may be others feel the same way.

If that is the case then there isn't much of concern about merging
changes into the upstream library that may soon become obsolete.
plans to sync with upstream or at least cherry-pick some of their changes?
http://sourceforge.net/p/libhugetlbfs/code/commit_browser
We have in the past, r3436 and r3445 are such fixes.

Doug
C. Bergström
2012-08-24 08:19:08 UTC
Permalink
Post by Gilmore, Doug
-----Original Message-----
Sent: Thursday, August 23, 2012 3:15 PM
To: Gilmore, Doug
Cc: open64-devel
Subject: Re: [Open64-devel] review request for changes to libhugetlbfs
Post by Gilmore, Doug
I have two changes to libhugetlbfs that I would like to have reviewed.
Do you have any intention to push this upstream as well?
Don't know.
I have been hoping that the performance issues with Transparent Huge
Pages (THP) will be addressed in the near future thus there is no need
for libhugetlbfs (the performance gains for THP are more appreciable
running under a VM rather than on the hardware).
The development of libhugetlbfs has been pretty quiet lately,
so it may be others feel the same way.
If that is the case then there isn't much of concern about merging
changes into the upstream library that may soon become obsolete.
I'm not a mind reader and don't want to guess what others are thinking.
Quiet projects can be for any sort of reasoning - I'd recommend to
submit your patches upstream for review since someone there probably
knows the code better. If accepted you make the sync easier in the
future and upstream gets a little activity.
David Coakley
2012-08-27 03:36:44 UTC
Permalink
Hi Doug,

I reviewed the changes. Here are my comments/questions:

The formatting (spaces vs. tabs) seems inconsistent, but it's already
that way in the project. It would be nice to clean it up in a
separate update.

Do you run the tests in the 'tests' subdirectory? Are there any new
tests that should be added for the changes that you made in the second
patch?

-David
Post by Gilmore, Doug
I have two changes to libhugetlbfs that I would like to have reviewed.
The first is just a fix to eliminate a compiler warning.
The second is a change to the library to handle situations where a
huge page mapping would exhaust the huge page limit. Previously no
huge pages would be allocated (when mapping ELF segments this failure
would cause program termination). With these changes, huge pages are
allocated up to consume the huge page limit, but small pages are used
to satisfy the rest of the allocation.
Could a gatekeeper review these changes when they have the chance?
Thanks,
Doug
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Open64-devel mailing list
https://lists.sourceforge.net/lists/listinfo/open64-devel
Gilmore, Doug
2012-08-27 05:40:29 UTC
Permalink
-----Original Message-----
Sent: Sunday, August 26, 2012 8:37 PM
To: Gilmore, Doug
Cc: open64-devel
Subject: Re: [Open64-devel] review request for changes to libhugetlbfs
Hi Doug,
The formatting (spaces vs. tabs) seems inconsistent, but it's already
that way in the project. It would be nice to clean it up in a
separate update.
I can do that.
Do you run the tests in the 'tests' subdirectory?
tests that should be added for the changes that you made in the second
patch?
-David
We included the tests directory with the original import, but we have
not reworked the tests to work with the Open64 driver linking
mechanism.

We'll need to investigate that.

Thanks,

Doug
Post by Gilmore, Doug
I have two changes to libhugetlbfs that I would like to have
reviewed.
Post by Gilmore, Doug
The first is just a fix to eliminate a compiler warning.
The second is a change to the library to handle situations where a
huge page mapping would exhaust the huge page limit. Previously no
huge pages would be allocated (when mapping ELF segments this failure
would cause program termination). With these changes, huge pages are
allocated up to consume the huge page limit, but small pages are used
to satisfy the rest of the allocation.
Could a gatekeeper review these changes when they have the chance?
Thanks,
Doug
---------------------------------------------------------------------
---------
Post by Gilmore, Doug
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond.
Discussions
Post by Gilmore, Doug
will include endpoint security, mobile security and the latest in
malware
Post by Gilmore, Doug
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Open64-devel mailing list
https://lists.sourceforge.net/lists/listinfo/open64-devel
Loading...