8245194: Unix domain socket channel implementation #52

Michael-Mc-Mahon · 2020-09-07T12:05:07Z

Continuing this review as a PR on github with the comments below incorporated. I expect there will be a few more iterations before integrating.

On 06/09/2020 19:47, Alan Bateman wrote:

On 26/08/2020 15:24, Michael McMahon wrote:

As I mentioned the other day, I wasn't able to use the suggested method on Windows which returns an absolute path. So, I added a method to WindowsPath which returns the path in the expected UTF-8 encoding as a byte[]. Let me know what you think of that.

There is a fair bit of other refactoring and simplification done also. Next version is at:

http://cr.openjdk.java.net/~michaelm/8245194/impl.webrev/webrev.9/

Adding a method to WindowsPath to encode the path using UTF-8 is okay but I don't think we should be caching it as the encoding for sun_path is an outlier on Windows. I'm also a bit dubious about encoding a relative path when the resolved path (before encoding) is > 247 chars. The documentation on the MS site isn't very completely and I think there are a number points that require clarification to fully understand how this will work with relative, directly relative and drive relative paths.

Maybe it would be better to just use the path returned from toString() and do the conversion to UTF-8 externally. That would leave WindowsPath unchanged.

In the same area, the new PathUtil is a bit inconsistent with the existing provider code. One suggestion is to add a method to AbstractFileSystemProvider instead. That is the base class the platform providers and would be a better place to get the file path in bytes.

Okay, I gave the method a name that is specific to Unix domain sockets because this UTF-8 strangeness is not likely to be useful by other components.

One other comment on the changes to the file system provider it should be okay to change UnixUserPrinipals ad its fromUid and fromGid method to be public. This would mean that the patch shouldn't need to add UnixUserGroup (the main issue is that class is that it means we end up with two classes with static factory methods doing the same thing).

Okay, that does simplify it a bit.

Thanks,
Michael.

-Alan.

Progress

Change must not contain extraneous whitespace
Commit message must refer to an issue
Change must be properly reviewed

Testing

	Linux x32	Linux x64	Windows x64	macOS x64
Build	✔️ (1/1 passed)	✔️ (5/5 passed)	✔️ (2/2 passed)	✔️ (2/2 passed)
Test (tier1)		✔️ (9/9 passed)	✔️ (9/9 passed)	✔️ (9/9 passed)

Issue

JDK-8245194: Unix domain socket channel implementation

Reviewers

Erik Joelsson (@erikj79 - Reviewer) ⚠️ Review applies to 0096645
Daniel Fuchs (@dfuch - Reviewer) ⚠️ Review applies to 368fa8b
Alan Bateman (@AlanBateman - Reviewer) ⚠️ Review applies to 36623a4
Chris Hegarty (@ChrisHegarty - Reviewer) ⚠️ Review applies to c01c3c9

Download

$ git fetch https://git.openjdk.java.net/jdk pull/52/head:pull/52
$ git checkout pull/52

…Alan's email 06/09/2020

… comment this morning

bridgekeeper · 2020-09-07T12:08:13Z

👋 Welcome back michaelm! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2020-09-07T12:08:59Z

@Michael-Mc-Mahon The following labels will be automatically applied to this pull request: core-libs hotspot-jfr net nio.

When this pull request is ready to be reviewed, an RFR email will be sent to the corresponding mailing lists. If you would like to change these labels, use the /label (add|remove) "label" command.

mlbridge · 2020-09-07T12:14:36Z

Webrevs

Michael-Mc-Mahon · 2020-09-09T07:19:05Z

/csr needed

openjdk · 2020-09-09T07:19:33Z

@Michael-Mc-Mahon the issue for this pull request, JDK-8245194, already has an approved CSR request: JDK-8247942

…automatically bound server sockets are created. After this change it is no longer necessary to specify the location in individual tests.

mlbridge · 2020-09-11T13:04:29Z

Mailing list message from Alan Bateman on nio-dev:

On 07/09/2020 13:14, Michael McMahon wrote:

In the same area, the new PathUtil is a bit inconsistent with the existing provider code. One suggestion is to add a
method to AbstractFileSystemProvider instead. That is the base class the platform providers and would be a better place
to get the file path in bytes.

Okay, I gave the method a name that is specific to Unix domain sockets because this UTF-8 strangeness is not likely to
be useful by other components.

The file system provider shouldn't know anything about Unix domain
sockets so I prefer not to have "UnixDomain" in the name of this method.
I also would prefer if null is an exception so that it is consistent
with the other methods.

I think it would be useful to also summarize how the bind/connect works
on Windows. For example, suppose the working directory is 240 characters
and the UnixDomainSocketAddress is to a simple relative path of 20
characters. If I understand correctly, the proposal will encode the
simple relative path (not the resolved absolute path) to bytes using
UTF-8 so it will probably be 20+ bytes in this case. This hould be okay
but inconsistent with the rest of? file system implementation which will
use the long form. Also if the name is long then it won't use the long
path prefix (\\?\) but instead rely on failure because the resulting
sequence of bytes when encoded is > 100, is that correct?

-Alan.

Michael-Mc-Mahon · 2020-09-11T14:41:32Z

Mailing list message from Alan Bateman on nio-dev:

On 07/09/2020 13:14, Michael McMahon wrote:

In the same area, the new PathUtil is a bit inconsistent with the existing provider code. One suggestion is to add a
method to AbstractFileSystemProvider instead. That is the base class the platform providers and would be a better place
to get the file path in bytes.

Okay, I gave the method a name that is specific to Unix domain sockets because this UTF-8 strangeness is not likely to
be useful by other components.

The file system provider shouldn't know anything about Unix domain
sockets so I prefer not to have "UnixDomain" in the name of this method.
I also would prefer if null is an exception so that it is consistent
with the other methods.

I'm not sure what to call the method then. It returns a UTF-8 string
converted to bytes on Windows and the output of getByteArrayForSysCalls on Unix.
It could be called getByteArrayUTF8 on Windows, but that would not
be right on Unix.

Since sockets are file system entities, why not have socket in the method name?

What about getByteArrayForSocket() ?

I think it would be useful to also summarize how the bind/connect works
on Windows. For example, suppose the working directory is 240 characters
and the UnixDomainSocketAddress is to a simple relative path of 20
characters. If I understand correctly, the proposal will encode the
simple relative path (not the resolved absolute path) to bytes using
UTF-8 so it will probably be 20+ bytes in this case.

Right

This hould be okay
but inconsistent with the rest of? file system implementation which will
use the long form. Also if the name is long then it won't use the long
path prefix (\?) but instead rely on failure because the resulting
sequence of bytes when encoded is > 100, is that correct?

Yes, that is how it is working currently. We check the length in native code
just before calling bind(). Whether the long path prefix is present or not
won't matter because the path will not be used once it is longer than ~100 bytes.

Michael.

-Alan.

…to 10/2019

mlbridge · 2020-09-27T15:56:18Z

Mailing list message from Alan Bateman on nio-dev:

On 24/09/2020 13:17, Michael McMahon wrote:
I've pulled the latest commits from jdk/pull/55 and have another set of
comments.

src/java.base/share/classes/sun/nio/ch/InetSocketChannelImpl.java
src/java.base/share/classes/sun/nio/ch/InetServerSocketChannelImpl.java
- class level comment is out dated after the split so should be updated
to make it clear that it for Inet implementations.
- the comment "End of field protected by stateLock" has been copied down
from ServerSocketChannelImpl.java and is confusing here so I think
should be removed.
- The existing convention is to declare the final fields at the top so
best to keep consistent after the split.

src/java.base/share/classes/sun/net/ext/ExtendedSocketOptions.java
- the change to isStreamOption would be cleaned if the "SO_FLOW_SLA"
case is removed (I think it was missed by JDK-8244582).

src/java.base/share/classes/sun/net/util/SocketExceptions.java
- it's a bit inconsistent for of(IOException, SocketAddress) to handle
InetSocketAddress types itself, it would be cleaner to handle both
address types or use two helper methods.
- Is the null check in ofUnixDomain left over from a previous iteration?

src/java.base/share/classes/sun/nio/ch/SelectorProviderImpl.java
- it would be help to split the really long lines to keep it consistent
with the existing code

src/java.base/share/classes/sun/nio/fs/AbstractFileSystemProvider.java
- it might be cleaner to rename the method to getSunPathForSocketFile.
It doesn't access the file so would be better not to throw IOException.
- for the description it might be clearer to say that it "Returns a path
name as bytes for ..."

src/java.base/share/native/libnet/net_util.h
- are you sure that struct sockaddr_un is defined on all platforms,
including slightly older VC++ releases.
- the lines are a bit long here, inconsistent with the existing code.

src/java.base/unix/classes/sun/nio/ch/UnixDomainHelper.java
- I think this should be removed and the method included in
UnixDomainSockets
- The hardcoding of /tmp and /var/tmp needs to be dropped too. We also
need to find a name name for jdk.nio.channels.tmpdir as this is specific
to Unix domain sockets.
- you can avoid the ugly cast by creating a PrivilegedAction and then
return doPrivileged(pa).

src/java.base/unix/classes/sun/nio/fs/UnixFileSystemProvider.java
- if you rename the parameter to obj and use UnixPath file =
UnixPath.toUnixPath(obj) then it will be consistent with the existing
code. It will also ensure that ProviderMismatchException is thrown.

src/java.base/windows/classes/sun/nio/fs/WindowsFileSystemProvider.java
- this can use WindowsPath file = WindowsPath.toWindowsPathobj)

src/java.base/windows/classes/sun/nio/ch/PipeImpl.java
src/java.base/windows/classes/sun/nio/ch/SinkChannelImpl.java
- I think the change to SinkChannelImpl.java should be dropped. Instead,
move the setting TCP_NODELAY to the PipeImpl constructor and it should
be a lot cleaner. It also means the "buffered" field is not needed.
- if you rename tryUnixDomain noUnixDomainSockets you avoid the
volatile-write at initialization time
- createListener would be a clean if the flag is checked before the
try/catch.

src/java.base/windows/native/libnet/net_util_md.h
- what is the include of afunix.h needed, left over?

src/java.base/windows/native/libnio/ch/Net.c
- are these changed needed? I would only checked localInetAddress and
remoteInetAddress to be called on file descriptor to inet sockets.

src/jdk.net/linux/classes/jdk/net/LinuxSocketOptions.java
src/jdk.net/macosx/classes/jdk/net/MacOSXSocketOptions.java
- IOUtil.makePipe return a long that encodes the two file descriptors
and may be simpler to avoid populating the int[] in JNI code.

test/jdk/com/sun/nio/sctp/SctpServerChannel/NonBlockingAccept.java
- did you mean to change this test?

test/jdk/java/net/httpclient/whitebox/java.net.http/jdk/internal/net/http/ConnectionPoolTest.java
- is this needed?

I'm still working through new Unix* code and the tests so more comments
soon.

-Alan

Michael-Mc-Mahon · 2020-09-29T07:10:57Z

Hi Alan,

Thanks for the comments. Assume all are accepted, except as queried below

Mailing list message from Alan Bateman on nio-dev:

On 24/09/2020 13:17, Michael McMahon wrote:
I've pulled the latest commits from jdk/pull/55 and have another set of
comments.

src/java.base/share/classes/sun/nio/ch/InetSocketChannelImpl.java
src/java.base/share/classes/sun/nio/ch/InetServerSocketChannelImpl.java

class level comment is out dated after the split so should be updated
to make it clear that it for Inet implementations.

the comment "End of field protected by stateLock" has been copied down
from ServerSocketChannelImpl.java and is confusing here so I think
should be removed.

The existing convention is to declare the final fields at the top so
best to keep consistent after the split.

src/java.base/share/classes/sun/net/ext/ExtendedSocketOptions.java

the change to isStreamOption would be cleaned if the "SO_FLOW_SLA"
case is removed (I think it was missed by JDK-8244582).

src/java.base/share/classes/sun/net/util/SocketExceptions.java

it's a bit inconsistent for of(IOException, SocketAddress) to handle
InetSocketAddress types itself, it would be cleaner to handle both
address types or use two helper methods.

Is the null check in ofUnixDomain left over from a previous iteration?

src/java.base/share/classes/sun/nio/ch/SelectorProviderImpl.java

it would be help to split the really long lines to keep it consistent
with the existing code

src/java.base/share/classes/sun/nio/fs/AbstractFileSystemProvider.java

it might be cleaner to rename the method to getSunPathForSocketFile.
It doesn't access the file so would be better not to throw IOException.

for the description it might be clearer to say that it "Returns a path
name as bytes for ..."

src/java.base/share/native/libnet/net_util.h

are you sure that struct sockaddr_un is defined on all platforms,
including slightly older VC++ releases.

struct sockaddr_un is declared in <afunix.h> on Windows and <sys/un.h>
on Unix. So, it is included via net_util_md.h

I suspect the definition is only in fairly recent SDKs on Windows, and I'm not sure
what I can do about that. If <afunix.h> is present
it will have the definition. If it's not present, then there will be a compile
error, but I don't know how to tell if it is present..

the lines are a bit long here, inconsistent with the existing code.

src/java.base/unix/classes/sun/nio/ch/UnixDomainHelper.java

I think this should be removed and the method included in
UnixDomainSockets

The hardcoding of /tmp and /var/tmp needs to be dropped too. We also
need to find a name name for jdk.nio.channels.tmpdir as this is specific
to Unix domain sockets.

The problem here is that the Java impl of UnixDomainHelper is platform
specific, but UnixDomainSockets is shared, but with platform specific
native methods. If I make UnixDomainSockets.java platform specific then
there will be a lot of duplication of code. Maybe we could pick a better name
than UnixDomainHelper?

As regards hard-coding of /tmp and /var/tmp, I think we need to be very
careful to ensure that the directory name used to hold these automatically
bound sockets is as short as possible. ${java.io.tmpdir} is sometimes
long enough on our test systems to push the names over the ~100 byte limit.
I don't see a good alternative to using /tmp or /var/tmp on Unix
On Windows %TEMP% seems to be always defined and seems to always
refer to a directory with a short name.

What about "jdk.nio.unixdomain.tmpdir" for the property name?

you can avoid the ugly cast by creating a PrivilegedAction and then
return doPrivileged(pa).

src/java.base/unix/classes/sun/nio/fs/UnixFileSystemProvider.java

if you rename the parameter to obj and use UnixPath file =
UnixPath.toUnixPath(obj) then it will be consistent with the existing
code. It will also ensure that ProviderMismatchException is thrown.

src/java.base/windows/classes/sun/nio/fs/WindowsFileSystemProvider.java

this can use WindowsPath file = WindowsPath.toWindowsPathobj)

src/java.base/windows/classes/sun/nio/ch/PipeImpl.java
src/java.base/windows/classes/sun/nio/ch/SinkChannelImpl.java

I think the change to SinkChannelImpl.java should be dropped. Instead,
move the setting TCP_NODELAY to the PipeImpl constructor and it should
be a lot cleaner. It also means the "buffered" field is not needed.

if you rename tryUnixDomain noUnixDomainSockets you avoid the
volatile-write at initialization time

createListener would be a clean if the flag is checked before the
try/catch.

src/java.base/windows/native/libnet/net_util_md.h

what is the include of afunix.h needed, left over?

Explained above <afunix.h> is Windows only.

src/java.base/windows/native/libnio/ch/Net.c

are these changed needed? I would only checked localInetAddress and
remoteInetAddress to be called on file descriptor to inet sockets.

I had changed this file, but reverted all changes in the last version. I don't
know why the webrev is still seeing changes. After this round, hopefully
they will be gone.

src/jdk.net/linux/classes/jdk/net/LinuxSocketOptions.java
src/jdk.net/macosx/classes/jdk/net/MacOSXSocketOptions.java

IOUtil.makePipe return a long that encodes the two file descriptors
and may be simpler to avoid populating the int[] in JNI code.

test/jdk/com/sun/nio/sctp/SctpServerChannel/NonBlockingAccept.java

did you mean to change this test?

test/jdk/java/net/httpclient/whitebox/java.net.http/jdk/internal/net/http/ConnectionPoolTest.java

is this needed?

I'm still working through new Unix* code and the tests so more comments
soon.

-Alan

Thanks for the review. I'm making the changes as described, except where noted
above for further discussion. This should appear as a new webrev revision today.

Michael

(1) rename UnixDomainHelper to UnixDomainSocketsUtil (2) remove hardcoded /tmp and /var/tmp paths from UnixDomainSocketsUtil (3) replace (2) with documented system/networking properties (4) Small update to UnixDomainSocketAddress API (5) CSR for (3) and (4) submitted at JDK-8253930 (6) Update build to generate net.properties from shared net.properties.common plus platform specific additions.

…e/Copy.gmk

test/jdk/java/net/UnixDomainSocketAddress/AddressTest.java

test/jdk/java/nio/channels/spi/SelectorProvider/inheritedChannel/Launcher.java

test/jdk/java/net/UnixDomainSocketAddress/LengthTest.java

test/jdk/java/nio/channels/unixdomain/Bind.java

test/jdk/java/nio/channels/unixdomain/policy1

test/jdk/java/nio/channels/unixdomain/NonBlockingAccept.java

test/jdk/java/nio/channels/unixdomain/IOExchanges.java

- test updates from Chris

src/java.base/windows/classes/sun/nio/ch/WindowsSelectorImpl.java

Michael-Mc-Mahon · 2020-10-28T17:24:47Z

/integrate

openjdk · 2020-10-28T17:27:07Z

@Michael-Mc-Mahon Since your change was applied there have been 29 commits pushed to the master branch:

8bde2f4: 8255013: implement Record Classes as a standard feature in Java, follow-up
0425889: 8255429: Remove C2-based profiling
aaf4f69: 8255233: InterpreterRuntime::at_unwind should be a JRT_LEAF
bbf0a31: 8255397: x86: coalesce reference and int entry points into vtos bytecodes
3bd5b80: 8243583: Change 'final' error checks to throw ICCE
1f00c3b: 8255527: Shenandoah: Let ShenadoahGCStateResetter disable barriers
3c4fc79: 8255299: Drop explicit zeroing at instantiation of Atomic* objects
6b2d11b: 8255246: AArch64: Implement BigInteger shiftRight and shiftLeft accelerator/intrinsic
591e7e2: 8255378: [Vector API] Remove redundant vector length check after JDK-8254814 and JDK-8255210
2c9dfc7: 8255234: ZGC: Bulk allocate forwarding data structures
... and 19 more: https://git.openjdk.java.net/jdk/compare/dccfd2b3e0636488e255da1fa5b1b58e9bc1b606...master

Your commit was automatically rebased without conflicts.

Pushed as commit 6bb7e45.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

* Add a PEAParanoid mode. In this mode, C2 aborts when it is expanding an object which has been materialized at least once. This is not supposed to happen and will have allocation duplication on some paths. There are two reasons for this. 1) because we skip passive materialization, an object PHI may render the object NSR and hinders it from eliminating. if it's a reducible phi, try to use Cesar's patch to mitigate it. 2) we don't support recursive materialization like Graal PEA. I added a new testcase 'CompositeObjects' to demonstrate it. * fix formatting issue --------- Co-authored-by: Xin Liu <xxinliu@amazon.com>

Michael-Mc-Mahon added 2 commits September 7, 2020 10:47

unixdomainchannels: initial commit from hg sandbox with changes from …

e2eb987

…Alan's email 06/09/2020

unixdomainchannels: fixing compile error on Windows and Alan's review…

6d164c9

… comment this morning

openjdk bot added the rfr Pull request is ready for review label Sep 7, 2020

openjdk bot added nio nio-dev@openjdk.org core-libs core-libs-dev@openjdk.org net net-dev@openjdk.org hotspot-jfr hotspot-jfr-dev@openjdk.org labels Sep 7, 2020

Made some changes relating to selection of the local directory where …

4d08c15

…automatically bound server sockets are created. After this change it is no longer necessary to specify the location in individual tests.

Michael-Mc-Mahon added 5 commits September 14, 2020 12:55

Merge branch 'master' into unixdomainchannels

110b45c

unixdomainchannels: updates from review of Sept 11 2020

22f37a8

unixdomainchannels: updates after comments sent by Alan 14 Sept

3669509

unixdomainchannels: some tests were failing on Windows version prior …

d29e8cc

…to 10/2019

Merge branch 'master' into unixdomainchannels

a08186d

Michael-Mc-Mahon added 6 commits September 29, 2020 09:53

unixdomainchannels: Update after Alan's review of Sept 29

c669e2d

unixdomainchannels: typo in WindowsFileSystemProvider.java

4dc90c7

unixdomainchannels: change to Net.c was missed after all

3fe6baa

unixdomainchannels: remove more cruft from Windows Net.c

e78228b

Merge branch 'master' into unixdomainchannels

c40d205

openjdk bot added the build build-dev@openjdk.org label Oct 2, 2020

unixdomainchannels: error in the last commit in make/modules/java.bas…

dc70cd6

…e/Copy.gmk

openjdk bot removed the rfr Pull request is ready for review label Oct 20, 2020

forgot to stage updated test files in last commit

3594e4b

openjdk bot added ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 20, 2020

dfuch reviewed Oct 20, 2020

View reviewed changes

test/jdk/java/net/UnixDomainSocketAddress/AddressTest.java Show resolved Hide resolved

test/jdk/java/nio/channels/spi/SelectorProvider/inheritedChannel/Launcher.java Outdated Show resolved Hide resolved

further test update from Daniel

e07522b

dfuch approved these changes Oct 20, 2020

View reviewed changes

ChrisHegarty reviewed Oct 21, 2020

View reviewed changes

test/jdk/java/net/UnixDomainSocketAddress/LengthTest.java Outdated Show resolved Hide resolved

updated one error in review

368fa8b

dfuch approved these changes Oct 21, 2020

View reviewed changes

Merge branch 'master' into unixdomainchannels

01d7cf5

ChrisHegarty suggested changes Oct 21, 2020

View reviewed changes

test/jdk/java/nio/channels/unixdomain/Bind.java Outdated Show resolved Hide resolved

test/jdk/java/nio/channels/unixdomain/Bind.java Outdated Show resolved Hide resolved

ChrisHegarty suggested changes Oct 21, 2020

View reviewed changes

test/jdk/java/nio/channels/unixdomain/policy1 Show resolved Hide resolved

test/jdk/java/nio/channels/unixdomain/NonBlockingAccept.java Show resolved Hide resolved

test/jdk/java/nio/channels/unixdomain/IOExchanges.java Outdated Show resolved Hide resolved

- fixed 32 bit Linux build error

c01c3c9

- test updates from Chris

ChrisHegarty approved these changes Oct 22, 2020

View reviewed changes

AlanBateman reviewed Oct 24, 2020

View reviewed changes

src/java.base/windows/classes/sun/nio/ch/WindowsSelectorImpl.java Outdated Show resolved Hide resolved

Michael-Mc-Mahon added 2 commits October 27, 2020 09:44

tidy up WindowsSelectorImpl

bb6c729

Merge branch 'master' into unixdomainchannels

1a8b304

openjdk bot added ready Pull request is ready to be integrated and removed ready Pull request is ready to be integrated labels Oct 27, 2020

openjdk bot closed this Oct 28, 2020

openjdk bot added integrated Pull request has been integrated and removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 28, 2020

Michael-Mc-Mahon deleted the unixdomainchannels branch October 28, 2020 18:11

calvinccheung mentioned this pull request Feb 12, 2021

8261608: Move common CDS archive building code to archiveBuilder.cpp #2536

Closed

3 tasks

cushon pushed a commit to cushon/jdk that referenced this pull request Apr 2, 2021

Add aliasing annotation (openjdk#52)

ef55b48

8245194: Unix domain socket channel implementation #52

8245194: Unix domain socket channel implementation #52

Uh oh!

Conversation

Michael-Mc-Mahon commented Sep 7, 2020 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Testing

Issue

Reviewers

Download

Uh oh!

bridgekeeper bot commented Sep 7, 2020

Uh oh!

openjdk bot commented Sep 7, 2020

Uh oh!

mlbridge bot commented Sep 7, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

Michael-Mc-Mahon commented Sep 9, 2020

Uh oh!

openjdk bot commented Sep 9, 2020

Uh oh!

mlbridge bot commented Sep 11, 2020

Uh oh!

Michael-Mc-Mahon commented Sep 11, 2020

Uh oh!

mlbridge bot commented Sep 27, 2020

Uh oh!

Michael-Mc-Mahon commented Sep 29, 2020

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Michael-Mc-Mahon commented Oct 28, 2020

Uh oh!

openjdk bot commented Oct 28, 2020

Uh oh!

Uh oh!

Michael-Mc-Mahon commented Sep 7, 2020 •

edited by openjdk bot

Loading

mlbridge bot commented Sep 7, 2020 •

edited

Loading