On Tue, 23 Jun 2026 23:13:41 GMT, Ashay Rane <[email protected]> wrote:
> Before this patch, the test in `testExtractToReadOnlyDir()`, which > validates whether making jimage extract to a read-only directory results > in a failure, denied creating files but it did not deny creating > subdirectories. Although this does make jimage fail (and consequently, > make the test pass), this isn't completely correct for two reasons. > > First, the test invocation ends up writing a series of empty > subdirectories, when in reality we don't want the destination directory > to be modified at all. Second, and more importantly, this leaves these > directories in a state where Cygwin cannot remove them (see JBS issue > for more details). > > This patch fixes the problem by adding `APPEND_DATA` to the deny list, > thus prohibiting the principal running the test from creating both files > as well as directories. This effectively sidesteps the issue of leaving > broken directories on disk. > > See > https://learn.microsoft.com/en-us/windows/win32/fileio/file-access-rights-constants > for details about the file permissions, reproduced in part below: > >> FILE_WRITE_DATA: For a directory object, the right to create a file in > the directory. >> >> FILE_APPEND_DATA: For a directory object, the right to create a > subdirectory. > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). I think this is okay but would be good if @dbalek could comment. ------------- Marked as reviewed by alanb (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/31646#pullrequestreview-4625266818
