[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openjdk-2d-dev
Subject:    Re: [OpenJDK 2D-Dev] RFR: 6986863: ProfileDeferralMgr throwing ConcurrentModificationException [v6]
From:       Alexander Zuev <kizune () openjdk ! java ! net>
Date:       2021-01-28 12:00:50
Message-ID: nfjXHxW4A5DOiPGOso7rsh6uG-l50LCzctEP1-VrbKE=.220e1c31-5728-4c33-b7f3-0d4ff24ee40d () github ! com
[Download RAW message or body]

On Thu, 28 Jan 2021 03:44:00 GMT, Sergey Bylokhov <serb@openjdk.org> wrote:

> > This change intended to enhance the lazy initialization of the standard color \
> > profiles concurrently by different threads. 
> > We defer standard profiles loading because most of UI application uses a small \
> > amount of data from the profile like numComponents and colorSpaceType, and this \
> > data is known in advance. But any other profile-related activity (like a color \
> > conversion, profile data accesses, etc.) triggers profile activation when we load \
> > all profile data to the memory. 
> > Before the fix for JDK-6793818, we defer only one sRGB color profile, see: \
> > https://github.com/openjdk/jdk/commit/2726f2a3621dd2562d4fb660b4c3d376c65027aa 
> > Notes about the link above:
> > - The code in the ProfileDeferralMgr, which contain the Vector of profiles for \
> >                 activation does not use any synchronization
> > - The `activateDeferredProfile` and `activate` methods are implemented to throw \
> > `ProfileDataException`, but this exception is ignored during activation process:  \
> > https://github.com/openjdk/jdk/commit/2726f2a3621dd2562d4fb660b4c3d376c65027aa#diff-0839c25a6c999452be28b431c54d5daa91364d302cfda6efa5c56421c2f2bdcbR96
> >  
> > The fix:
> > - Drops the usage of ProfileDeferralMgr (which contained the Vector of profiles \
> > for activation) and ProfileActivator machinery. Instead, we will have just one \
> > `ICC_Profile.activate()` method which will activate and initialize the \
> >                 `ICC_Profile.cmmProfile` if it is null
> > - The `activate` method implementation mimics the old behavior when the \
> > CMMException and IOException were wrapped by the ProfileDataException, and the \
> > ProfileDataException itself was ignored during activation - > so an exception \
> > will not be thrown in the method itself, but only when the null profile will be \
> > used. 
> > See some comments inline.
> 
> Sergey Bylokhov has updated the pull request with a new target base due to a merge \
> or a rebase. The pull request now contains 17 commits: 
> - Merge branch 'JDK-6986863' of https://github.com/mrserb/jdk into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - update dates
> - Update ProfileActivationDuringPropertyAccess.java
> - Merge branch 'master' into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - Merge branch 'master' into JDK-6986863
> - ... and 7 more: https://git.openjdk.java.net/jdk/compare/7030d2e0...e0e59caf

Marked as reviewed by kizune (Reviewer).

-------------

PR: https://git.openjdk.java.net/jdk/pull/1613


[prev in list] [next in list] [prev in thread] [next in thread]