<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Hi Jon,<br>
<br>
<div class="moz-cite-prefix">On 2015-01-09 17:48, Jon Masamitsu
wrote:<br>
</div>
<blockquote cite="mid:54B00674.7020302@oracle.com" type="cite">
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Esjohanss/8062063/hotspot.02/src/share/vm/gc_implementation/g1/g1PageBasedVirtualSpace.cpp.frames.html">http://cr.openjdk.java.net/~sjohanss/8062063/hotspot.02/src/share/vm/gc_implementation/g1/g1PageBasedVirtualSpace.cpp.frames.html</a><br>
<br>
Can this be put under the <b>else </b>branch for the <b>if
(_special)</b><br>
<pre> 147 if (AlwaysPreTouch) {
<span class="changed"> 148 os::pretouch_memory(page_start(start), page_start(end));</span>
149 }</pre>
</blockquote>
No, it needs to be outside since the first time we commit "special"
memory we need to pretouch it the same way as "normal" memory.<br>
<blockquote cite="mid:54B00674.7020302@oracle.com" type="cite"> The
BitMap<br>
<br>
<pre><span class="new"> 52 // Bitmap used to keep track of which pages are dirty or not for _special</span>
<span class="new"> 53 // spaces. This is needed because for those spaces the underlying memory</span>
<span class="new"> 54 // will only be zero filled the first time it is committed. Calls to commit</span>
<span class="new"> 55 // will use this bitmap and return whether or not the memory is zero filled.</span>
<span class="new"> 56 BitMap _dirty;
</span></pre>
is used to decide what parts of the memory being brought back (the
equivalent of<br>
the commit() for _special) to zero. The commit() is passed a
start address and size. What<br>
is the situation where you need the information in _dirty?
Meaning you could zero<br>
the range passed into commit() i.e., (start, start+size), yes?
When would that differ<br>
from using _dirty?<br>
</blockquote>
Again the first time is the special case. Since the memory is zero
the first time we don't what to clear it. If doing this we risk
getting regressions, and this is the whole reason the zero_filled
boolean exist for the on_commit method in the
G1MappingChangedListener. So basically we use the information in
_dirty to determine if it is the first time memory is committed or
not. If it is not the first time commit will return the the memory
is not zero filled and this is then passed on to the on_commit
method.<br>
<br>
Also, we don't want to zero the range in commit (the reason for
doing this re-spin), since that would affect all users of
G1PageBasedVirtualSpace. <br>
<br>
Thanks,<br>
Stefan<br>
<blockquote cite="mid:54B00674.7020302@oracle.com" type="cite"> <br>
Jon <br>
<pre><span class="new"></span></pre>
<div class="moz-cite-prefix">On 1/9/2015 6:31 AM, Stefan Johansson
wrote:<br>
</div>
<blockquote cite="mid:54AFE624.8030104@oracle.com" type="cite">Updated
links: <br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Esjohanss/8062063/hotspot.01-02/">http://cr.openjdk.java.net/~sjohanss/8062063/hotspot.01-02/</a>
<br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Esjohanss/8062063/hotspot.02/">http://cr.openjdk.java.net/~sjohanss/8062063/hotspot.02/</a>
<br>
<br>
Cheers, <br>
Stefan <br>
<br>
On 2015-01-09 12:31, Stefan Johansson wrote: <br>
<blockquote type="cite">Hi Thomas and Kim, <br>
<br>
On 2015-01-08 23:02, Thomas Schatzl wrote: <br>
<blockquote type="cite">Hi Stefan, <br>
<br>
On Thu, 2015-01-08 at 14:05 -0500, Kim Barrett wrote: <br>
<blockquote type="cite">On Jan 8, 2015, at 12:59 PM, Stefan
Johahttp://cr.openjdk.java.net/~sjohanss/8062063/hotspot.01-02/nsson
<a moz-do-not-send="true" class="moz-txt-link-rfc2396E"
href="mailto:stefan.johansson@oracle.com"><stefan.johansson@oracle.com></a>
wrote: <br>
<blockquote type="cite">Sorry for doing a re-spin on this,
but since this is targeted to go into 8 as well I want
to minimize the risk of introducing a regression. <br>
<br>
After yesterdays comments I started thinking more about
what regressions this fix might cause and today I've had
good discussions with Thomas and Mikael. I've also did
some quick measurements that shows additional time for
the YCs expanding the heap after a shrink. Since we
don't really need the heap regions to be cleared I think
we need to avoid this regression, by going with another
solution and I don't think having this time added to the
full GC shrinking the heap is wanted either. <br>
<br>
The first proposal that is explained in the bug-report
would avoid clearing memory that don't have to be
cleared, but just doing the simple solution explained
there might cause startup regressions due to touching
memory during startup that isn't needed. Mixing that
approach with the one proposed yesterday will allow us
to only clear memory when absolutely needed. See new
webrev here: <br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Esjohanss/8062063/hotspot.01">http://cr.openjdk.java.net/~sjohanss/8062063/hotspot.01</a>
<br>
<br>
This approach leaves the clearing to the listener
registered with each mapper and for the bitmaps this
will make sure that they are cleared, but for the heap
we won't do anything (because the heap has no
requirement of having zeroed backing memory). <br>
</blockquote>
I think I might prefer having the new bitmap called
_zero_filled and <br>
flip the sense of it. The present name, _needs_zeroing,
is mildly <br>
confusing to me, since whether zeroing is needed is
caller-dependent. <br>
</blockquote>
I somewhat tend to agree with Kim here. <br>
<br>
Maybe one of _needs_clear_on_commit, _not_zero_on_commit, <br>
_is_clear_on_commit, or _pages_dirty_after_uncommit, but
ymmv. :) <br>
</blockquote>
I too agree, but I changed it to simply _dirty and added a big
comment for it. This lets me keep the current logic where a 1
in the bitmap means it is not filled with zeros (dirty). <br>
<br>
New webrev and incremental one: <br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://wikifiles.se.oracle.com/dev/sjohanss/8062063/hotspot.02/">http://wikifiles.se.oracle.com/dev/sjohanss/8062063/hotspot.02/</a>
<br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://wikifiles.se.oracle.com/dev/sjohanss/8062063/hotspot.01-02/">http://wikifiles.se.oracle.com/dev/sjohanss/8062063/hotspot.01-02/</a>
<br>
<br>
Thanks, <br>
Stefan <br>
<br>
<blockquote type="cite">
<blockquote type="cite">It seems to me this new version can
result in unnecessary page <br>
clearing; commit will return true if any page in the range
is not <br>
zeroed. This can lead to a caller that needs zeroed pages
clearing <br>
the entire requested range, even if only some (perhaps
small) subset <br>
of the range is actually dirty. <br>
<br>
Of course, the previous attempted fix also had unnecessary
page <br>
clearing, since dirty pages were being cleared even if the
caller <br>
doesn't care. The new code seems likely to be an
improvement overall. <br>
<br>
In the context of fixing the bug at hand, I think this
change looks <br>
good, up to the naming and sense of the new bit map. <br>
<br>
But it looks like the API provided by
G1PageBasedVirtualSpace is less <br>
than ideal in this area, and could perhaps use further
work. Though <br>
it might not be worth worrying about, as the cases where
it matters <br>
may be rare and not especially important. <br>
</blockquote>
It simply assumes that a commit() zero-fills the page
lazily. <br>
<br>
I do not think it is worth worrying a lot about it. There
need to be a <br>
lot of circumstances involved, and the new change at least
always avoids the <br>
clearing of the Java heap space. <br>
<br>
The best solution would simply be doing away with the
pre-commit hack <br>
when using large pages :) <br>
<br>
Thomas <br>
<br>
<br>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>