harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Pavel Pervov (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HARMONY-3126) [drlvm] VMMagic: migration to new interfaces
Date Thu, 08 Feb 2007 13:23:05 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-3126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12471325

Pavel Pervov commented on HARMONY-3126:

1) VTable comression and reference compression are orthogonal. So, please, review your code
carefully to reflect this.
2) Field_Handle and Field* are autoconvertable inside VM Core code. So, no need to explicitly
3) Is it enough to compare annotations by name? Main concern here is that upon call to method_has_annotation,
all annotations for that method will be loaded.
4) Still if loading annotations is required it is better to use ClassLoader::LoadVerifyAndPrepareClass
instead of Type_Desc manipulation interfaces.
5) In is_magic_type_name please consider using memcmp where possible. Note, that you know
lengths of all strings participating in these operations. This will certainly increase performance.
6) Generally, set_magic_type() is excessive as you perform this operation from inside Field
method (namely, parse). If you prefer keeping this method, it should not be public.
7) I think, there exists only one PRAGMA_INLINE class. So, this field should be made static.
8) In case of some misconfiguration (absence of PRAGMA_INLINE class) OPT will just assert
(or die with SIGSEGV in release mode) instead of bringing out some useful diagnostics.
9) In md->hasAnnotation(inlinePragma) block please consider splitting logging into several
lines (Log, print, Log).
10) Optimization for Type::Boolean is not exactly related to magics, isn't it? There are also
several more changes/optimizations which are not related to magics. Would it be more correct
and clean to file them as separate JIRA?
11) Many string comparisons were introduced with magics support.
12) Changes in LHH: as both ifs return PROB_HEURISTIC_FAIL is it worth merging these ifs into

Phew!.. :)

> [drlvm] VMMagic: migration to new interfaces
> --------------------------------------------
>                 Key: HARMONY-3126
>                 URL: https://issues.apache.org/jira/browse/HARMONY-3126
>             Project: Harmony
>          Issue Type: Improvement
>          Components: DRLVM
>            Reporter: Mikhail Fursov
>         Assigned To: Alexey Varlamov
>         Attachments: vmmagic_1.zip, wrn.diff
> This patch contains:
> 1) all VMmagics methods updated to use annotations-like pragmas instead of obsolete exception-like
> 2) vmmagic.jar location is updated to point to the new version. Run 'build update' before
building drlvm
> 3) method_has_annotation was added to VM interfaces
> 4) magic types were removed from VM/GC enumeration
> 5) some refactoting was done to simplefy EM64T port (I hope it were all VM/GC changes
> 6) minor bugfixes and improvements

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message