RFR 8068687: Remove meta-index support and cleanup hotspot code for rt.jar etc in non-modular jdk image
harold.seigel at oracle.com
Tue Feb 17 13:50:48 UTC 2015
Thanks for the review.
On 2/16/2015 5:40 PM, Mandy Chung wrote:
> On 2/16/2015 12:49 PM, harold seigel wrote:
>> Please review this updated webrev for JDK-8068687:
> Thanks. Looks fine to me.
>> The only changes from the previous webrev are to classLoader.cpp
>> starting around line 1314, and removal of rt.jar comments from
>> cfgnode.cpp, library_call.cpp, and type.cpp.
>> These latest changes were tested with jtreg hotspot, java/lang,
>> java/io, and java/util, and with JCK lang, VM, and API.
>> Thanks, Harold
>> On 2/12/2015 3:21 PM, Mandy Chung wrote:
>>> On 2/12/15 6:49 AM, harold seigel wrote:
>>>> Please review this fix for bug 8068687. The fix removes the code
>>>> that implements meta-index support and rt.jar related processing,
>>>> because this code is no longer used. Removing meta-index support
>>>> may allow removal of LazyClassPathEntry support. I will look into
>>>> that in a separate bug.
>>> I was wondering that. It's fine to separate it from this bug.
>>>> Open webrev:
>>> classLoader.cpp - there are references to rt.jar that I think should
>>> be updated.
>>> 321 // Check whether zip name ends in "rt.jar"
>>> 1322 // This will match other archives named rt.jar as well, but this is
>>> 1323 // only used for debugging.
>>> 1324 return string_ends_with(zip->name, "rt.jar");
>>> 1352 // We stop at rt.jar, unless it is the first bootstrap path entry
>>> 1353 if (e->is_jrt() && e != _first_entry) break;
>>> Found rt.jar in comments - probably to take out those.
>>> 2601: // of safe & unsafe memory. Otherwise fails in a CTW of rt.jar
>>> 3125: // Happens in a CTW of rt.jar, 320-341, no extra flags
More information about the hotspot-runtime-dev