> Although the original implementation of `JLI_Open()` borrowed from the
> code in src/hotspot/os/windows/os_windows.cpp, the improvements to
> os_windows.cpp don't seem to have been applied to java_md.c, causing
> some tests to fail when the path to JAR files is longer than `MAXPATH`
> (i.e. 260) characters on Windows (see associated JBS issue 8385024 for
> details). Since `JLI_Open()` is not just invoked inside tests, this is
> not a test-specific issue, so fixing the test is not the right solution.
>
> This patch applies the recent changes from os_windows.cpp to java_md.c
> so that `JLI_Open()` can correctly handle longer than `MAXPATH` paths.
> The new code is almost the same as that in `wide_abs_unc_path()` in
> os_windows.cpp, except that the code in java_md.c uses `JLI_MemAlloc()`
> and `JLI_MemFree()` for memory allocation and deallocation.
>
> Although it would be ideal to have just one implementation between
> HotSpot and the launcher, the dependencies of the two components
> prevents us from having a single implementation.
>
> ---------
> - [x] I confirm that I make this contribution in accordance with the [OpenJDK
> Interim AI Policy](https://openjdk.org/legal/ai).
Ashay Rane has updated the pull request with a new target base due to a merge
or a rebase. The incremental webrev excludes the unrelated changes brought in
by the merge/rebase. The pull request contains 10 additional commits since the
last revision:
- Merge branch 'microsoft:main' into JDK-8385024-jli-maxpath
- Do not rewrite paths beginning with \\.\ as UNC paths
- Add null check after `JLI_MemAlloc()`
- Address PR comments
1. Dropped prefix argument since the lone caller passed an empty value.
2. Added a block that I missed copying from `wide_abs_unc_path()` to
`convert_to_absolute_path()`.
- Address PR comments
1. Use `Path.of(System.getProperty("user.dir"))` instead of
`Path.of(".").toAbsolutePath().normalize()` for clarity.
2. Fix year in copyright notices.
- Remove statement that I had inserted for debugging
- Ensure hard failure when test doesn't pass
Previously, when the `testLongPathJarFile()` or the
`testLongResolvedPathJarFile()` test cases failed, the code in
TestHelper would append an error message (that was checked at the end of
the test) but the test cases themselves didn't throw an exception, so it
wasn't clear _which_ test case failed. More precisely, the output said:
```
Total: Passed: 7, Failed 0
Total of 2 failed
```
This patch updates these tests so that upon failure, they throw an
exception, causing the name of the failing test to be printed in the
error log.
- Add test to exercise the change
This patch adds a new test so that the relative path (which becomes the
argument to `JLI_Open()`) is shorter than `MAX_PATH` but that the
absolute path (that is resolved inside `_open()`) is longer than
`MAX_PATH`. To construct such a case, the test first probes the length
of the current working directory and then creates additional
subdirectories to meet the desired constraint.
- Address PR comments
1. Optimize for the common case when the JAR path is shorter than
`MAX_PATH` by using a stack buffer of size length `MAX_PATH`. If the
full path is longer than that, then we allocate a heap buffer of the
desired size before calling `GetFullPathNameW()` with a buffer of the
right size.
2. Normalize the path before called `set_path_prefix()`, just like in
os_windows.cpp. This ensures that prefixed paths with forward
slashes work correctly.
3. Renamed `create_unc_path()` to `convert_to_absolute_path()` to
reflect the common action performed by the function.
4. Dropped the accidental overwritting of `errno` after calling
`_wopen()`.
- Reapply changes from os_windows.cpp to `JLI_Open()` in java_md.c
Although the original implementation of `JLI_Open()` borrowed from the
code in src/hotspot/os/windows/os_windows.cpp, the improvements to
os_windows.cpp don't seem to have been applied to java_md.c, causing
some tests to fail when the path to JAR files is longer than `MAXPATH`
(i.e. 260) characters on Windows (see associated JBS issue 8385024 for
details). Since `JLI_Open()` is not just invoked inside tests, this is
not a test-specific issue, so fixing the test is not the right solution.
This patch applies the recent changes from os_windows.cpp to java_md.c
so that `JLI_Open()` can correctly handle longer than `MAXPATH` paths.
The new code is almost the same as that in `wide_abs_unc_path()` in
os_windows.cpp, except that the code in java_md.c uses `JLI_MemAlloc()`
and `JLI_MemFree()` for memory allocation and deallocation.
Although it would be ideal to have just one implementation between
HotSpot and the launcher, the dependencies of the two components
prevents us from having a single implementation.
-------------
Changes:
- all: https://git.openjdk.org/jdk/pull/31209/files
- new: https://git.openjdk.org/jdk/pull/31209/files/aa81d386..d5551343
Webrevs:
- full: https://webrevs.openjdk.org/?repo=jdk&pr=31209&range=09
- incr: https://webrevs.openjdk.org/?repo=jdk&pr=31209&range=08-09
Stats: 216837 lines in 2215 files changed: 101013 ins; 102255 del; 13569 mod
Patch: https://git.openjdk.org/jdk/pull/31209.diff
Fetch: git fetch https://git.openjdk.org/jdk.git pull/31209/head:pull/31209
PR: https://git.openjdk.org/jdk/pull/31209
