On Mon, 3 Oct 2022 13:59:25 GMT, Jayathirth D V <j...@openjdk.org> wrote:
> In https://bugs.openjdk.org/browse/JDK-8270915 we have a valid GIF image > which was generated using a third party app, but it contains large amount of > Metadata(numerous Application extension blocks). > > Also GIFImageReader doesn't use ignoreMetadata flag set by user while reading > Metadata which causes heavy memory usage in this scenario. We need to provide > support for ignoring metadata in GIFImageReader, so readMetadata() is updated > to use the "ignoreMetadata" flag and ignore all blocks except "Image > Descriptor" and "Graphics Control Extension". > > I have updated readMedata() to also use ReaderUtil.staggeredReadByteStream() > where we are allocating upfront memory based on parameters from metadata. > Also this PR has small refactoring of import statements. > > To replicate the bug we need byte stream with large amount of valid metadata, > so i am not able to add regression test case. I have run all javax/imageio > jtreg tests and there are no issues. This pull request has now been integrated. Changeset: 8c15f77a Author: Jayathirth D V <j...@openjdk.org> URL: https://git.openjdk.org/jdk/commit/8c15f77abac0beb7f39a90fdfc5efb245b09ca32 Stats: 88 lines in 1 file changed: 39 ins; 10 del; 39 mod 8270915: GIFImageReader disregards ignoreMetadata flag which causes memory exhaustion Reviewed-by: prr ------------- PR: https://git.openjdk.org/jdk/pull/10536
