<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    Hi all,<br>
    <br>
    Here's a new patch, which addresses StefanJ's issues:<br>
    <br>
    <a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~stefank/8152491/webrev.02.delta">http://cr.openjdk.java.net/~stefank/8152491/webrev.02.delta</a><br>
    <a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~stefank/8152491/webrev.02">http://cr.openjdk.java.net/~stefank/8152491/webrev.02</a><br>
    <br>
    Thanks,<br>
    StefanK<br>
    <br>
    <div class="moz-cite-prefix">On 2016-04-08 10:57, Stefan Johansson
      wrote:<br>
    </div>
    <blockquote cite="mid:5707727E.6050202@oracle.com" type="cite">
      <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
      Hi StefanK,<br>
      <br>
      <div class="moz-cite-prefix">On 2016-04-06 20:16, Stefan Karlsson
        wrote:<br>
      </div>
      <blockquote cite="mid:57055295.8030902@oracle.com" type="cite">
        <meta http-equiv="content-type" content="text/html;
          charset=utf-8">
        Hi all,<br>
        <br>
        Please review to convert TracePageSizes to use UL.<br>
        <br>
        <a moz-do-not-send="true" class="moz-txt-link-freetext"
          href="http://cr.openjdk.java.net/%7Estefank/8152491/webrev.01">http://cr.openjdk.java.net/~stefank/8152491/webrev.01</a><br>
        <a moz-do-not-send="true" class="moz-txt-link-freetext"
          href="https://bugs.openjdk.java.net/browse/JDK-8152491">https://bugs.openjdk.java.net/browse/JDK-8152491</a><br>
        <br>
        TracePageSizes used to be a develop flag, but I've decided to
        create a UL tag accessible from release builds. We've recently
        needed to debug some large pages code and wanted this
        information in release builds.<br>
        <br>
      </blockquote>
      <tt>Looks good in general and I like that we make it available in
        product mode. Just a few comments.</tt><br>
      <pre><span class="changed">os.cpp:
1534 void os::trace_page_sizes_for_requested_size(const char* str,</span>
<span class="changed">1535                                              const size_t requested_size,
</span><span class="changed">...</span>
<span class="changed">1541   log_info(pagesizes)("%s:"</span>
<span class="changed">1542                       " req_size=" SIZE_FORMAT "%s"</span>
<span class="changed">1543                       " base=" PTR_FORMAT</span>
<span class="changed">...</span>
<span class="changed">1547                       str,</span>
<span class="changed">1548                       trace_page_size_params(size),</span>
<span class="changed">1549                       p2i(base),</span>
<tt>
On line 1548 we should use requested_size instead of size.</tt><tt>
---</tt>
</pre>
      <tt>genCollectedHeap.cpp:<br>
        170 <br>
        171   os::trace_page_sizes("Heap",
        collector_policy()->min_heap_byte_size(),<br>
        172                        total_reserved,<br>
        173                        alignment,<br>
        174                        heap_rs->base(),<br>
        175                        heap_rs->size());<br>
      </tt><br>
      <tt>Move min_heap_byte_size() to its own line.<br>
        ---<br>
        <br>
        I also think that it would be better to use 'pagesize' instead
        of 'pagesizes' as the tag, even though the old flag said
        pagesizes. But there are a lot of other flags that also have an
        ending s, exceptions for example, so I'm not sure how we should
        handle it. It would be good to make the tags feel "unified".<br>
        <br>
        Thanks,<br>
        Stefan<br>
        <br>
      </tt>
      <blockquote cite="mid:57055295.8030902@oracle.com" type="cite">
        Changes in the patch:<br>
        <br>
        1) Replaced TracePageSizes with the 'pagesizes' UL tag<br>
        2) Cleaned up the memory region description strings<br>
        3) Used B/K/M/G suffixes when the size is exactly divisible by
        those<br>
        4) Removed
        <meta http-equiv="content-type" content="text/html;
          charset=utf-8">
        trace_gen_sizes, since the output was too obscure to be that
        helpful<br>
        5) Added tracing for Serial and CMS<br>
        6) Fixed a jtreg test that verified the output from
        TracePageSizes<br>
        <br>
        Example output from 'java -Xlog:pagesizes -Xmx512m
        -XX:+UseLargePages -version':<br>
        <br>
        [0.018s][info][pagesizes] CodeHeap 'non-nmethods':  min=2496K
        max=8M base=0x00007fd735000000 page_size=4K size=8M<br>
        [0.018s][info][pagesizes] CodeHeap 'profiled nmethods': 
        min=2496K max=116M base=0x00007fd735800000 page_size=4K
        size=116M<br>
        [0.018s][info][pagesizes] CodeHeap 'non-profiled nmethods': 
        min=2496K max=116M base=0x00007fd73cc00000 page_size=4K
        size=116M<br>
        [0.056s][info][pagesizes] Heap:  min=8M max=512M
        base=0x00000000e0000000 page_size=2M size=512M<br>
        [0.056s][info][pagesizes] Block Offset Table: req_size=1M
        base=0x00007fd6d5fe1000 page_size=4K alignment=4K size=1M<br>
        [0.056s][info][pagesizes] Card Table: req_size=1M
        base=0x00007fd6d5de1000 page_size=4K alignment=4K size=1M<br>
        [0.056s][info][pagesizes] Card Counts Table: req_size=1M
        base=0x00007fd6d5ce1000 page_size=4K alignment=4K size=1M<br>
        [0.056s][info][pagesizes] Prev Bitmap: req_size=8M
        base=0x00007fd6d5400000 page_size=2M alignment=2M size=8M<br>
        [0.056s][info][pagesizes] Next Bitmap: req_size=8M
        base=0x00007fd6d4c00000 page_size=2M alignment=2M size=8M<br>
        <br>
        The patch uses the LogStreamCHeap from the following patch that
        is out for review:<br>
        <a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://mail.openjdk.java.net/pipermail/hotspot-dev/2016-April/022477.html">http://mail.openjdk.java.net/pipermail/hotspot-dev/2016-April/022477.html</a><br>
        <br>
        Tested with JPRT and new internal VM test for the suffixes.<br>
        <br>
        Thanks,<br>
        StefanK<br>
      </blockquote>
      <br>
    </blockquote>
    <br>
  </body>
</html>