First possible bug in OpenJDK 11 port

[Michael Osipov]

Am 2019-03-16 um 07:14 schrieb Greg Lewis:
> On Sat, Mar 16, 2019 at 12:20:27AM +0100, Michael Osipov wrote:
>> Folks,
>>
>> here is a stripped down example from an issue with the new port:
>>
>>> $ cat ../src/exec/Test.java
>>> package exec;
>>>
>>> import java.io.IOException;
>>>
>>> public class Test {
>>>
>>>      public static void main(String[] args) throws IOException, InterruptedException {
>>>          ProcessBuilder builder = new ProcessBuilder("/bin/sh", "-c", "cd && env");
>>>          builder.inheritIO();
>>>          Process p = builder.start();
>>>          System.out.println("exit code: " + p.waitFor());
>>>      }
>>>
>>> }
>>
>> Fails:
>>> osipovmi at deblndw011x:~/exec/bin
>>> $ /usr/local/openjdk11/bin/java exec.Test
>>> Exception in thread "main" java.io.IOException: Cannot run program "/bin/sh": error=13, Permission denied
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1128)
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1071)
>>>      at exec.Test.main(Test.java:12)
>>> Caused by: java.io.IOException: error=13, Permission denied
>>>      at java.base/java.lang.ProcessImpl.forkAndExec(Native Method)
>>>      at java.base/java.lang.ProcessImpl.<init>(ProcessImpl.java:341)
>>>      at java.base/java.lang.ProcessImpl.start(ProcessImpl.java:272)
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1107)
>>>      ... 2 more
>>
>>
>> as well as
>>> osipovmi at deblndw011x:~/exec/bin
>>> $ /usr/local/bootstrap-openjdk11/bin/java exec.Test
>>> Exception in thread "main" java.io.IOException: Cannot run program "/bin/sh": error=13, Permission denied
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1128)
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1071)
>>>      at exec.Test.main(Test.java:12)
>>> Caused by: java.io.IOException: error=13, Permission denied
>>>      at java.base/java.lang.ProcessImpl.forkAndExec(Native Method)
>>>      at java.base/java.lang.ProcessImpl.<init>(ProcessImpl.java:341)
>>>      at java.base/java.lang.ProcessImpl.start(ProcessImpl.java:272)
>>>      at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1107)
>>>      ... 2 more
>>
>> and
>>
>>> osipovmi at deblndw011x:~/exec/bin
>>> $ /usr/local/bootstrap-openjdk8/bin/java exec.Test
>>> Exception in thread "main" java.io.IOException: Cannot run program "/bin/sh": error=13, Permission denied
>>>      at java.lang.ProcessBuilder.start(ProcessBuilder.java:1048)
>>>      at exec.Test.main(Test.java:12)
>>> Caused by: java.io.IOException: error=13, Permission denied
>>>      at java.lang.UNIXProcess.forkAndExec(Native Method)
>>>      at java.lang.UNIXProcess.<init>(UNIXProcess.java:251)
>>>      at java.lang.ProcessImpl.start(ProcessImpl.java:134)
>>>      at java.lang.ProcessBuilder.start(ProcessBuilder.java:1029)
>>>      ... 1 more
>>
>> but successfully runs with:
>>
>>> osipovmi at deblndw011x:~/exec/bin
>>> $ /usr/local/openjdk8/bin/java exec.Test
>>> SSH_CLIENT=147.54.65.49 55876 22
>>> LOGNAME=osipovmi
>>> LSCOLORS=ExGxFxdxCxDxDxhbadExEx
>>> HTTP_PROXY=http://194.145.60.1:9400
>>> LANG=de_DE.UTF-8
>>> JAVA_HOME=/usr/local/openjdk11
>>> NO_PROXY=localhost .siemens.net .siemens.com .siemens.de
>>> CLICOLOR=YES
>>> MAIL=/var/mail/osipovmi
>>> NCURSES_NO_UTF8_ACS=1
>>> PATH=/sbin:/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin
>>> EDITOR=vim
>>> OLDPWD=/net/home/osipovmi/exec/bin
>>> DISPLAY=localhost:10.0
>>> PWD=/net/home/osipovmi
>>> _=/usr/local/openjdk8/bin/java
>>> TERM=xterm
>>> SSH_TTY=/dev/pts/0
>>> USER=osipovmi
>>> HOME=/net/home/osipovmi
>>> SSH_CONNECTION=147.54.65.49 55876 147.54.64.17 22
>>> PS1=\[\033[1;32m\]\u@\h\[\033[0m\]:\[\033[1;34m\]\w\[\033[0m\]\n$
>>> SHELL=/bin/sh
>>> MM_CHARSET=UTF-8
>>> HTTPS_PROXY=http://194.145.60.1:9400
>>> BLOCKSIZE=K
>>> FTP_PROXY=http://194.145.60.1:9400
>>> SHLVL=1
>>> exit code: 0
>>
>> What is causing this issue?
>>
>> For those who'd like to reproduce the bug: Clone Maven GitHub, try to to
>> build from master and see the Maven Surefire Plugin failing when it
>> tried to launch the Surefire Booter.
>>
>> Java 8 Update 202, Java 11 newest from ports. OS is
>>> FreeBSD deblndw011x.ad001.siemens.net 12.0-STABLE FreeBSD 12.0-STABLE r344655 DEBLNDW011X  amd64
> 
> I can reproduce this with the current port in the FreeBSD ports tree.
> 
> It is better behaved with the current code in Github since there has been
> some work done on process handling.  With that code the command executes,
> but afterwards it throws an exception:
> 
> Exception in thread "main" java.lang.RuntimeException: No such process
> 	at java.base/java.lang.ProcessHandleImpl.isAlive0(Native Method)
> 	at java.base/java.lang.ProcessHandleImpl.getInternal(ProcessHandleImpl.java:250)
> 	at java.base/java.lang.ProcessImpl.<init>(ProcessImpl.java:349)
> 	at java.base/java.lang.ProcessImpl.start(ProcessImpl.java:272)
> 	at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1107)
> 	at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1071)
> 	at TestExec.main(TestExec.java:9)
> 
> So there is still some work to do there.  The relevant commit to the github
> repo is likely this one:
> 
> https://github.com/battleblow/openjdk-jdk11u/commit/4cfeb8650f3f90119aa1e52020ac6ad6620a34ff
> 
> This probably means there are still some problems with the
> os_getParentPidAndTimings code given that is what isAlive0 is calling
> (look at ProcessHandleImpl_unix.c).

Hi Greg,

that for the confirmation. Looks like a showstopper for a developer like 
me. Even my webapps won't run because they rely on ProcessBuilder.

Is there anything I can help to get this fixed? That type of C code is, 
unfortunately, out of my knowledge.

I want to verify your (and others) great work against all Maven 
subprojects as well as CodeHaus to find subtile bugsin the port.

Michael