<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
<br>
On 12/23/2011 05:05 PM, Srinivas Ramakrishna wrote:
<blockquote
cite="mid:CABzyjym_kek+WH9Cug=-W1YK2KBhw4iY5cKvHy3ZS+S923qgaA@mail.gmail.com"
type="cite">Tony, So if I understand yr proposal, you would switch
the roles of the to and from copies' length fields (since you<br>
need both -- one to keep track of the original length and the
other to keep track of the scan finger in the array).<br>
</blockquote>
<br>
Correct.<br>
<br>
<blockquote
cite="mid:CABzyjym_kek+WH9Cug=-W1YK2KBhw4iY5cKvHy3ZS+S923qgaA@mail.gmail.com"
type="cite">
Of course there's a fwd'ing ptr from the the pre-image to the
post-image, but not vice-versa. So you'd still<br>
push the pre-image reference on the work queue, but use the length
field from the post-copy which you'd<br>
fetch and modify via the forwarding reference?</blockquote>
<br>
Correct.<br>
<br>
<blockquote
cite="mid:CABzyjym_kek+WH9Cug=-W1YK2KBhw4iY5cKvHy3ZS+S923qgaA@mail.gmail.com"
type="cite"> It is probably true that the post-image's length is
not used<br>
during GC once it's been copied, but it'd be good to check (I'm
especially wary of CMS... but of course<br>
this is limited to G1 -- does G1 ever need to scan or iterate over
regions that are subject to being copied<br>
into during an incremental pause?)<br>
</blockquote>
<br>
This is of course something I was also worried about. In G1 we
should not be scanning to-space objects that are being copied during
GC, not only because the length might be incorrect due to this
change but also because there are no guarantees that the objects are
well formed (another thread might be in the process of copying
them). For all regions we copy objects into we call save_marks() so
that we never go over saved_mark() during scanning.<br>
<br>
Tony<br>
<br>
<blockquote
cite="mid:CABzyjym_kek+WH9Cug=-W1YK2KBhw4iY5cKvHy3ZS+S923qgaA@mail.gmail.com"
type="cite">-- ramki<br>
<br>
<div class="gmail_quote">On Fri, Dec 23, 2011 at 1:52 PM, Srinivas
Ramakrishna <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:ysr1729@gmail.com">ysr1729@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">From sheer
gut instinct (and nothing rational), it seems better to try
and manage the race in the reading of the size in the
from-space field (you know<br>
when you've read a negative size that you've lost a race; add
appropriate asserts to see that the object is forwarded,<br>
use the size from the post-copy image, and bail out), than it
is to mess with the length of the to-space image. Just sheer
gut instinct -- the pre-image is<br>
something we can play with once we have copied it (or have
safely won the race to copy it and know we are the<br>
only ones who will be modifying it). The post-copy image is
the pristine thing that we typically don't want to mess<br>
with lest we screw up. OK, I am waving my hands wildly and
this is not always true. But still, see if you can manage the<br>
race in the reading of the pre-image without undue cost.<br>
<br>
I'll think about it too at bed-time today :-)<br>
<br>
Happy Holidays!<br>
-- ramki
<div class="HOEnZb">
<div class="h5"><br>
<br>
<div class="gmail_quote">On Fri, Dec 23, 2011 at 12:37 AM,
Tony Printezis <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:tony.printezis@oracle.com"
target="_blank">tony.printezis@oracle.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">Hi
all,<br>
<br>
I have some bad news, it turns out that there's a
subtle race in the code. I've been testing the marking
changes for quite a while now (which rely on this
patch) and I got a single failure. I'm pretty sure
it's the race I describe below. BTW, there's a
question at the end of the e-mail. :-)<br>
<br>
Let's assume two threads, A and B, are racing to
forward the same object array X. Consider the
following interleaving:<br>
<br>
Thread A:<br>
is X forwarded? no<br>
<br>
Thread B:<br>
is X forwarded? no<br>
size = X.size();<br>
X' = allocate size<br>
try to forward X to X'? yes<br>
copy X to X'<br>
chunk X, X.length is negative<br>
<br>
Thread A:<br>
size = X.size(); -> BOOM, as size() just read a
negative length<br>
<br>
Interesting this race exists today but it's benign due
to either careful design or sheer luck. When A reads
size = X.size() it will actually get a smaller size
than the actual size of X and will allocate a chunk
smaller than X (!!!). However, given that X is already
forwarded it won't need to copy it and will undo the
allocation. So, no harm done.<br>
<br>
Question: Is there a reason why we use the from-space
length field to encode the next index instead of the
to-space length field? If we used the latter we can
simplify the code by quite a lot. I can't immediately
think of any issues that this might cause.<span><font
color="#888888"><br>
<br>
Tony</font></span>
<div>
<div><br>
<br>
On 12/21/2011 5:25 PM, Tony Printezis wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0
0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
Hi all,<br>
<br>
I have two code reviews (thanks to John and
Ramki!). Can I push or is anybody else still
looking at this?<br>
<br>
Tony<br>
<br>
On 12/14/2011 05:17 PM, Tony Printezis wrote:<br>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
Hi all,<br>
<br>
Can I have a couple of code review for this
small change?<br>
<br>
<a moz-do-not-send="true"
href="http://cr.openjdk.java.net/%7Etonyp/7121623/webrev.0/"
target="_blank">http://cr.openjdk.java.net/~tonyp/7121623/webrev.0/</a><br>
<br>
The CR has a bit more explanation. The short
version is that I'm now encoding the "start
index of the next chunk" in the from-space
length field of a chunked array (say *that*
quickly!) as a negative value to always be
able to distinguish it from the real length.
This will simplify the code for the CR John is
currently working on (6484965) but it's a
small, self-contained change so it's good to
get it code reviewed and pushed separately.<br>
<br>
Tony<br>
<br>
<br>
</blockquote>
</blockquote>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</blockquote>
</div>
<br>
</blockquote>
</body>
</html>