I'm injecting some code to hook apis in processes but I have some issues in some applications like chrome.exe
My test app launches a suspended process, do injection and api hooking and then resumes it.
CreateProcessW is hooked in order to be able to hook child processes. If CreateProcessW is called, it is forced to be created suspended, hook the child and resume it.
The injected code only depends on ntdll api's so, although hooked processes are not fully initialized yet, ntdll.dll is always present.
Code is injected using a helper thread using CreateRemoteThread or NtCreateThreadEx with the CREATE_SUSPENDED flag. (No matter which one, the issue still there)
After this intro, the problem is that in some processes like some chrome childs, CreateRemoteThread returns TRUE but when I resume the injector thread, it exits with code 0xC0000022 and the process exits too.
If I attach WinDbg to a chrome.exe child process that is suspended, before I do anything, it fails too and chrome.exe ends with the same behavior.
Seems O.S. code executed before RtlUserThreadStart, generates the error but I don't know how to debug it.
How can I debug code that runs before RtlUserThreadStart? Is there a debugger or a windbg option that allows me to do that?
EDIT:
Following the last post from here, I could retrieve this info:
0a88:0814 # 02688302 - LdrpInitializeProcess - INFO: Beginning execution of chrome.exe (c:\Program Files (x86)\Google\Chrome\Application\chrome.exe)
Current directory: C:\Windows
Search path: C:\Windows\SYSTEM32 0a88:0814 # 02688318 - LdrpInitializeProcess - ERROR: Initializing the current directory to "C:\Windows" failed with status 0xc0000022
0a88:0814 # 02688334 - LdrLoadDll - ENTER: DLL name: C:\Windows\SYSTEM32\wow64.dll DLL path: NULL 0a88:0814 # 02688349 - LdrpLoadDll - ENTER: DLL name: C:\Windows\SYSTEM32\wow64.dll DLL path: C:\Windows\SYSTEM32
0a88:0814 # 02688365 - LdrpLoadDll - INFO: Loading DLL C:\Windows\SYSTEM32\wow64.dll from path C:\Windows\SYSTEM32 0a88:0814 # 02688380 - LdrpFindOrMapDll - ENTER: DLL name: C:\Windows\SYSTEM32\wow64.dll DLL path: C:\Windows\SYSTEM32
0a88:0814 # 02688396 - LdrpSearchPath - ENTER: DLL name: C:\Windows\SYSTEM32\wow64.dll DLL path: C:\Windows\SYSTEM32
0a88:0814 # 02688412 - LdrpResolveFileName - ENTER: DLL name: C:\Windows\SYSTEM32\wow64.dll
0a88:0814 # 02688427 - LdrpResolveFileName - RETURN: Status: 0xc0000022
0a88:0814 # 02688443 - LdrpSearchPath - RETURN: Status: 0xc0000022
0a88:0814 # 02688458 - LdrpFindOrMapDll - RETURN: Status: 0xc0000022
0a88:0814 # 02688474 - LdrpLoadDll - RETURN: Status: 0xc0000022
0a88:0814 # 02688490 - LdrLoadDll - RETURN: Status: 0xc0000022
0a88:0814 # 02688505 - LdrpInitializeProcess - ERROR: Loading WOW64 image management DLL "C:\Windows\SYSTEM32\wow64.dll" failed with status 0xc0000022
0a88:0814 # 02688521 - _LdrpInitialize - ERROR: Process initialization failed with status 0xc0000022
0a88:0814 # 02688536 - LdrpInitializationFailure - ERROR: Process initialization failed with status 0xc0000022
The process is created with a restricted token, the main thread inherits it but my injector thread isn't restricted because it is created by my app.
I can assume ntdll's apis are not hooked yet by chrome (in this case) because injection takes place before CreateProcess returns to chrome.
May the non-restricted token in my thread conflicts with process token in some way?
Take a look at Debugging WinLogon in the windbg help (debugger.chm). Simply substitute "chrome.exe" for "winlogon.exe". This technique controls a user mode debugger (ntsd) from the kernel mode debugger. I believe this will allow you debug chrome.exe's process initialization much earlier than using a user mode debugger alone.
The issue in chrome was the following:
Chrome launches child processes with very limited privileges (because of the sandbox) but before resuming the main thread it impersonates the main thread with a token with more privileges in order to let the process initialize.
My injector thread was not impersonating so the limited process token raised the 0xC0000022 exit code when the LdrpInitializeProcess routine was executed.
Related
My unit tests pass when run locally. When executed by a GitHub Action when a branch is committed, this action
on:
push:
branches:
- '*' # matches every branch that doesn't contain a '/'
- '*/*' # matches every branch containing a single '/'
- '**' # matches every branch
- '!master' # excludes master
workflow_dispatch:
inputs:
build:
description: Unit tests
type: boolean
required: true
default: "true"
jobs:
build:
strategy:
matrix:
os: [windows-latest]
# os: [ubuntu-latest, windows-latest, macos-latest]
runs-on: ${{ matrix.os }}
steps:
- name: Checkout
uses: actions/checkout#v2
- name: Install Node.js
uses: actions/setup-node#v1
with:
node-version: 16.x
- run: npm install
- run: npm i gulp gulp-cli -g
if: runner.os == 'MacOS'
- run: gulp
- run: xvfb-run -a npm test
if: runner.os == 'Linux'
- run: npm test
if: runner.os != 'Linux'
basically runs npm test.
When I do that locally, it works fine:
npm test
> vscode-print#0.9.25 test
> node ./out/test/runTest.js
Debugger listening on ws://127.0.0.1:63067/252bd7f9-c748-45b4-b9e6-8ba21e9fc909
For help, see: https://nodejs.org/en/docs/inspector
Debugger attached.
Found existing install in C:\Users\Peter\source\vsc-print\.vscode-test\vscode-win32-x64-archive-1.68.1. Skipping download
[16792:0704/000228.095:ERROR:node_bindings.cc(286)] Most NODE_OPTIONs are not supported in packaged apps. See documentation for more details.
[16792:0704/000228.095:ERROR:node_bindings.cc(286)] Most NODE_OPTIONs are not supported in packaged apps. See documentation for more details.
[main 2022-07-03T14:02:28.455Z] update#ctor - updates are disabled by the environment
[main 2022-07-03T14:02:29.679Z] Starting extension host with pid 32 (fork() took 15 ms).
Loading development extension at c:\Users\Peter\source\vsc-print
Decryption failed
Print Extension Test Suite
✔ Check platform browser launch command on win32 (200ms)
✔ Check platform alternate browser launch command on win32 (112ms)
✔ Print active editor (304ms)
Could not find the language 'jsonc', did you forget to load/include a language module?
✔ Print folder (812ms)
4 passing (2s)
[main 2022-07-03T14:02:32.887Z] Extension host with pid 32 exited with code: 0, signal: null.
Exit code: 0
Done
Waiting for the debugger to disconnect...
When the Action runs it, this is the output.
Downloaded VS Code into D:\a\vsc-print\vsc-print\.vscode-test\vscode-win32-x64-archive-1.68.1
[main 2022-07-03T13:06:32.496Z] update#ctor - updates are disabled by the environment
workbench#open()
[main 2022-07-03T13:06:37.868Z] Calling fork to start extension host...
[main 2022-07-03T13:06:37.879Z] Starting extension host with pid 776 (fork() took 11 ms).
IExtensionHostStarter.start() took 42 ms.
Loading development extension at d:\a\vsc-print\vsc-print
Print Extension Test Suite
1) Check platform browser launch command on win32
Error: 4 tests failed.
2) Check platform alternate browser launch command on win32
3) Print active editor
at d:\a\vsc-print\vsc-print\out\test\suite\index.js:25:27
4) Print folder
at done (d:\a\vsc-print\vsc-print\node_modules\mocha\lib\mocha.js:1009:7)
0 passing (518ms)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
4 failing
1) Print Extension Test Suite
Check platform browser launch command on win32:
Error: command 'extension.test.browserLaunchCommand' not found
at v._tryExecuteCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3532)
at v.executeCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3414)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
2) Print Extension Test Suite
Check platform alternate browser launch command on win32:
Error: command 'extension.test.browserLaunchCommand' not found
at v._tryExecuteCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3532)
at v.executeCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3414)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
3) Print Extension Test Suite
Print active editor:
Error: command 'extension.test.flags' not found
at v._tryExecuteCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3532)
at v.executeCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3414)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
4) Print Extension Test Suite
Print folder:
Error: command 'extension.test.flags' not found
at v._tryExecuteCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3532)
at v.executeCommand (vscode-file://vscode-app/d:/a/vsc-print/vsc-print/.vscode-test/vscode-win32-x64-archive-1.68.1/resources/app/out/vs/workbench/workbench.desktop.main.js:1706:3414)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
[main 2022-07-03T13:06:41.294Z] Waiting for extension host with pid 776 to exit.
[main 2022-07-03T13:06:41.330Z] Extension host with pid 776 exited with code: 1, signal: null.
Exit code: 1
Failed
Failed to run tests
Error: Process completed with exit code 1.
The unrecognised commands are registered near the start of the extension's activation event.
export function activate(context: vscode.ExtensionContext) {
let ecmPrint = vscode.workspace.getConfiguration("print", null).editorContextMenuItemPosition,
etmButton = vscode.workspace.getConfiguration("print", null).editorTitleMenuButton,
disposable: vscode.Disposable;
vscode.commands.executeCommand("setContext", "ecmPrint", ecmPrint);
vscode.commands.executeCommand("setContext", "etmButton", etmButton);
context.subscriptions.push(vscode.workspace.onDidChangeConfiguration(checkConfigurationChange));
context.subscriptions.push(vscode.commands.registerCommand("extension.print", printCommand));
context.subscriptions.push(vscode.commands.registerCommand("extension.printFolder", printFolderCommand));
context.subscriptions.push(vscode.commands.registerCommand("extension.test.flags", () => testFlags));
context.subscriptions.push(vscode.commands.registerCommand("extension.test.browserLaunchCommand", PrintSession.getLaunchBrowserCommand));
This suggests that the activation method hasn't been called. Is this something I must do explicitly?
I just noticed this up the top of extension.test.ts
// You can import and use all API from the 'vscode' module
// as well as import your extension to test it
import * as vscode from 'vscode';
// import * as myExtension from '../../extension';
That's not how my tests are written. They don't import the code and call it directly, they assume the extension is installed and active and they use registered commands. Some of these commands were added for no other reason than to allow fine grained unit tests.
The problem I describe in the question is due to the fact that the extension is not installed.
Discovering I can import from the extension doesn't solve all the problems. Some of my tests are strictly speaking integration tests. They need the extension to be installed and active.
To meet that requirement, gulp package rather than gulp build will also produce the VSIX. The only real trick is obtaining the VSIX name to install it in the test environment.
const vsixName = fs.readdirSync(extensionDevelopmentPath)
.filter(p => path.extname(p) === ".vsix")
.sort((a, b) => a < b ? 1 : a > b ? -1 : 0)[0];
const launchArgs = [
path.resolve(__dirname, '../../src/test/test-docs'),
"--install-extension",
vsixName
];
A hybrid approach is possible; some of the registered commands exist only for use by unit tests. These could be junked in favour of direct calls.
I have installed ActiveMQ Artemis on my office laptop. I created a broker, and from the broker's bin folder I ran artemis-exe install.
Then after that I ran the start command. It shows it has started but, when I check status it shows stopped. I cannot access localhost:8161. Here is my log file:
2020-09-07 12:30:45,458 DEBUG - Starting WinSW in the CLI mode
2020-09-07 12:30:46,033 INFO - Starting the service with id 'artemis-btBroker-0.0.0.0'
2020-09-07 12:30:46,737 DEBUG - Starting WinSW in the service mode
2020-09-07 12:30:46,837 DEBUG - Completed. Exit code is 0
2020-09-07 12:30:46,881 INFO - Starting %JAVA_HOME%\bin\java.exe -Xbootclasspath/a:C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\lib\jboss-logmanager-2.1.10.Final.jar;C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\lib\wildfly-common-1.5.2.Final.jar -XX:+UseParallelGC -Xms512M -Xmx1024M -classpath C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\lib\artemis-boot.jar -Dartemis.home=C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0 -Dartemis.instance=C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\bin\btBroker -Ddata.dir=C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\bin\btBroker\data -Dartemis.instance.etc=C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\bin\btBroker\etc -Djava.util.logging.manager=org.jboss.logmanager.LogManager -Dlogging.configuration=file:/C:/Users/613379463/Downloads/apache-artemis-2.15.0-bin/apache-artemis-2.15.0/bin/btBroker/etc//logging.properties -Djava.security.auth.login.config=C:\Users\613379463\Downloads\apache-artemis-2.15.0-bin\apache-artemis-2.15.0\bin\btBroker\etc\login.config -Dhawtio.realm=activemq -Dhawtio.offline="true" -Dhawtio.role=amq -Dhawtio.rolePrincipalClasses=org.apache.activemq.artemis.spi.core.security.jaas.RolePrincipal -Djolokia.policyLocation=file:/C:/Users/613379463/Downloads/apache-artemis-2.15.0-bin/apache-artemis-2.15.0/bin/btBroker/etc//jolokia-access.xml org.apache.activemq.artemis.boot.Artemis run
2020-09-07 12:30:47,081 DEBUG - Completed. Exit code is 0
The Artemis ActiveMQ service is known to start and immediately stop (without any errors shown on command line) if %JAVA_HOME% is not set, as Apurv Adarsh's comment mentions.
The solution is to set the env var and then run the service as normal, as follows:
(1) Set JAVA_HOME
Verify %JAVA_HOME% system environment variable exists:
echo $Env:JAVA_HOME # from PowerShell
echo %JAVA_HOME% # from cmd
# e.g. 'C:\Program Files\Java\jdk-10.0.2\'
# or blank if not set.
if not, find java.exe (download/install, if necessary):
> where.exe java.exe # from PowerShell
> where java.exe # from cmd
# C:\Program Files (x86)\Common Files\Oracle\Java\javapath\java.exe
# C:\Program Files\Java\jdk-10.0.2\bin\java.exe
Set JAVA_HOME env var, I'll choose the jdk-10.0.2 version. /M will set as system variable, instead of user var. btw, no need to include '/bin/java.exe'
> setx /M JAVA_HOME "C:\Program Files\Java\"
SUCCESS: Specified value was saved.
(Restart the prompt to load the newly set env variable.)
(2) Install & Start Artemis Windows Service
cd into Artemis ActiveMQ root directory. Let my_broker be the name of the broker instance already created.
Ensure service is installed, then start:
> "my_broker/bin/artemis-service" install
...
> "my_broker/bin/artemis-service" start
2021-01-01 12:00:00,000 INFO - Starting the service with id 'artemis-my_broker-0.0.0.0'
(3) Check Running Status:
Verify service status is running:
> "my_broker/bin/artemis-service" status
Started
I'd like to ask since I am setting up the Eclipse IDE for Yocto app development and I got stuck to start the QEMU from within Eclipse.
I got working QEMU image OK such as
ubuntu#ubuntu:~/work/community/build-x11$ runqemu qemuarm
tmp/deploy/images/qemuarm/zImage-qemuarm.bin
tmp/deploy/images/qemuarm/fsl-image-multimedia-full-qemuarm.ext4
Within the Eclipse I follow
https://www.yoctoproject.org/docs/2.5/sdk-manual/sdk-manual.html#oxygen-starting-qemu-in-user-space-nfs-mode
But by configuring "External Tools" and try to run QEMU I got following
runqemu - INFO - Running MACHINE=qemuarm bitbake -e...
ERROR: Unable to find conf/bblayers.conf or conf/bitbake.conf. BBAPTH is unset and/or not in a build directory?
runqemu - WARNING - Couldn't run 'bitbake -e' to gather environment information:
runqemu - WARNING - Can't find qemuboot conf file, DEPLOY_DIR_IMAGE is NULL!
runqemu - INFO - Running MACHINE=qemuarm bitbake -e...
ERROR: Unable to find conf/bblayers.conf or conf/bitbake.conf. BBAPTH is unset and/or not in a build directory?
runqemu - WARNING - Couldn't run 'bitbake -e' to gather environment information:
runqemu - INFO - Setting STAGING_DIR_NATIVE to OECORE_NATIVE_SYSROOT (/home/ubuntu/work/community/build-x11/tmp/work/armv5e-fslc-linux-gnueabi/meta-ide-support/1.0-r3/recipe-sysroot-native)
runqemu - INFO - Setting STAGING_BINDIR_NATIVE to /home/ubuntu/work/community/build-x11/tmp/work/armv5e-fslc-linux-gnueabi/meta-ide-support/1.0-r3/recipe-sysroot-native/usr/bin
runqemu - INFO - QB_MEM is not set, use 512M by default
runqemu - INFO - Continuing with the following parameters:
KERNEL: [/home/ubuntu/work/community/build-x11/tmp/deploy/images/qemuarm/zImage-qemuarm.bin]
MACHINE: [qemuarm]
FSTYPE: [nfs]
NFS_DIR: [/home/ubuntu/work/community/build-x11/MY_QEMU_ROOTFS]
CONFFILE: []
/bin/sh: 1: stty: not found
Traceback (most recent call last):
File "/home/ubuntu/work/community/sources/poky/scripts/runqemu", line 1270, in main
config.setup_network()
File "/home/ubuntu/work/community/sources/poky/scripts/runqemu", line 997, in setup_network
self.saved_stty = subprocess.check_output("stty -g", shell=True).decode('utf-8')
File "/usr/lib/python3.5/subprocess.py", line 626, in check_output
**kwargs).stdout
File "/usr/lib/python3.5/subprocess.py", line 708, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command 'stty -g' returned non-zero exit status 127
Cleanup
Command 'lesspipe' is available in the following places
* /bin/lesspipe
* /usr/bin/lesspipe
The command could not be located because '/bin:/usr/bin' is not included in the PATH environment variable.
lesspipe: command not found
Command 'dircolors' is available in '/usr/bin/dircolors'
The command could not be located because '/usr/bin' is not included in the PATH environment variable.
dircolors: command not found
ubuntu#ubuntu:~/eclipse/cpp-oxygen/eclipse$
I wonder if any has experience such a problem on setting it up "External Tools" with Eclipse?
Thank you
Navigate to the build path and then trigger this command will work.
OR
Run source oe-init-build-env build path.
VSCode Version: 1.16.1
OS Version: Windows Server 2012 R2 Standard
Is it possible to run the PowerShell terminal using PowerShell v2.0? I have an older SharePoint farm that requires earlier version of PowerShell and would like to run that version instead of more current versions. When I open VS Code on my server the PowerShell terminal crashes.
Here is the log of the crash.
9/18/2017 10:09:24 AM [NORMAL] - Visual Studio Code v1.16.1 64-bit
9/18/2017 10:09:24 AM [NORMAL] - PowerShell Extension v1.4.3
9/18/2017 10:09:24 AM [NORMAL] - Operating System: Windows 64-bit
9/18/2017 10:09:24 AM [NORMAL] - Language server starting --
9/18/2017 10:09:24 AM [NORMAL] - exe: C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe
9/18/2017 10:09:24 AM [NORMAL] - args: C:\Users\me\.vscode\extensions\ms-vscode.powershell-1.4.3\scripts\Start-EditorServices.ps1 -EditorServicesVersion '1.4.1' -HostName 'Visual Studio Code Host' -HostProfileId 'Microsoft.VSCode' -HostVersion '1.4.3' -AdditionalModules #('PowerShellEditorServices.VSCode') -BundledModulesPath 'C:\Users\me\.vscode\extensions\ms-vscode.powershell-1.4.3\modules' -EnableConsoleRepl -LogLevel 'Normal' -LogPath 'C:\Users\me\.vscode\extensions\ms-vscode.powershell-1.4.3\logs\1505754564-f1d3cfae-50af-4f74-9503-061208469b8d1505754561535\EditorServices.log' -SessionDetailsPath 'C:\Users\me\.vscode\extensions\ms-vscode.powershell-1.4.3\sessions\PSES-VSCode-10764-479503' -FeatureFlags #()
9/18/2017 10:09:25 AM [NORMAL] - powershell.exe started, pid: 6412
9/18/2017 10:09:25 AM [NORMAL] - powershell.exe terminated or terminal UI was closed
9/18/2017 10:10:25 AM [NORMAL] - Language server startup failed.
9/18/2017 10:10:25 AM [ERROR] - The language service could not be started:
9/18/2017 10:10:25 AM [ERROR] - Timed out waiting for session file to appear.
I experienced the same symptoms after updating VS Code (1.25.1) and PowerShell extension (1.8.2). Discovered that C:\Users\USERNAME.vscode\extensions\ms-vscode.powershell-1.8.2\modules\PowerShellEditorServices\Start-EditorServices.ps1 was failing to launch by looking at the logs and trying to launch it manually from a separate PowerShell instance with the same arguments. It returned the following prompt: Do you want to run software from this untrusted publisher? [V] Never run [D] Do not run [R] Run once [A] Always run [?] Help (default is "D"): indicating that the digital signature was not trusted.
To remedy this I performed the following:
Browse to the file noted in the error message above. (ex. C:\Users\USERNAME.vscode\extensions\ms-vscode.powershell-1.8.2\modules\PowerShellEditorServices\Start-EditorServices.ps1)
Right click on the file and choose Properties | Digital Signatures
Select the signature and click Details | View Certificate
Choose the Details tab and click Copy to File...
Save the file with a .CER extension.
Using the Certificates MMC Snapin, import this file into the Trusted Publishers Certificate Store on the Local Machine.
(original solution post in https://github.com/PowerShell/vscode-powershell/issues/1360 )
Every build has failed as of Tuesday. I'm not exactly sure what happened. The Phing targets (clean/prepare) are being executed properly. Additionally, the unit tests are passing with flying colors, with only a warning for duplicate code (not a reason for a fail). I tried removing the phpDoc target to see if that was causing the error, but the build still failed.
Started by user chris Updating
file://localhost/projects/svn/ips-com/trunk
At revision 234 no change for
file://localhost/projects/svn/ips-com/trunk
since the previous build [trunk] $
/opt/phing/bin/phing clean prepare
-logger phing.listener.NoBannerLogger Buildfile:
/var/lib/hudson/.hudson/jobs/IPS/workspace/trunk/build.xml
IPS > clean:
[echo] Clean... [delete] Deleting directory
/var/lib/hudson/.hudson/jobs/IPS/workspace/build
IPS > prepare:
[echo] Prepare...
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build/logs
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build/logs/coverage
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build/logs/coverage-html
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build/docs
[mkdir] Created dir: /var/lib/hudson/.hudson/jobs/IPS/workspace/build/app
BUILD FINISHED
Total time: 1.0244 second
[workspace] $ /bin/bash -xe
/tmp/hudson3259012225710915845.sh
+ cd trunk/tests
+ /usr/local/bin/phpunit --verbose -d memory_limit=512M --log-junit
../../build/logs/phpunit.xml
--coverage-clover ../../build/logs/coverage/clover.xml
--coverage-html ../../build/logs/coverage-html/
PHPUnit 3.5.0 by Sebastian Bergmann.
IPS Default_IndexControllerTest .
Default_AuthControllerTest ......
Manage_UsersControllerTest .....
testDeleteInvalidUserId ..
testGetPermissionsForInvalidUserId .. Audit_OverviewControllerTest
............
Time: 14 seconds, Memory: 61.00Mb
[30;42m[2KOK (28 tests, 198
assertions) [0m[2K Writing code
coverage data to XML file, this may
take a moment.
Generating code coverage report, this
may take a moment.
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
Warning: Unknown: Error occured while
closing statement in Unknown on line 0
[workspace] $ /bin/bash -xe
/tmp/hudson1439023061736436000.sh
+ /usr/local/bin/phpcpd --log-pmd ./build/logs/cpd.xml ./trunk phpcpd
1.3.2 by Sebastian Bergmann.
Found 1 exact clones with 6 duplicated
lines in 2 files:
library/Ips/Form/Decorator/SplitInput.php:8-14
library/Ips/Form/Decorator/FeetInches.php:10-16
0.04% duplicated lines out of 16585 total lines of code.
Time: 4 seconds, Memory: 19.50Mb [DRY]
Skipping publisher since build result
is FAILURE Publishing Javadoc [xUnit]
[INFO] - Starting to record. [xUnit]
[WARNING] - Can't create the path
/var/lib/hudson/.hudson/jobs/IPS/workspace/generatedJUnitFiles.
Maybe the directory already exists.
[xUnit] [INFO] - Processing
PHPUnit-3.4 (default) [xUnit] [INFO] -
[PHPUnit-3.4 (default)] - 1 test
report file(s) were found with the
pattern 'build/logs/phpunit.xml'
relative to
'/var/lib/hudson/.hudson/jobs/IPS/workspace'
for the testing framework 'PHPUnit-3.4
(default)'. [xUnit] [INFO] -
Converting
'/var/lib/hudson/.hudson/jobs/IPS/workspace/build/logs/phpunit.xml'
. [xUnit] [INFO] - Stopping recording.
Publishing Clover coverage report...
Publishing Clover XML report...
Publishing Clover coverage results...
Finished: FAILURE
What changed since Tuesday? Try to manually run exactly the same commands that Hudson tries to run from the same directory that Hudson starts it from (usually the jobs workspace directory). Of course with the user account that Hudson is started under.
There are several possibilities. ranging from standard groups for a directory, to permission, or other things outside of Hudson. Was Hudson upgraded? Was a plugin upgraded? Was the OS or php upgraded? Was there a change in the default or user .profile or .env (or the equivalent files)? Does another process accesses the workspace? ......
Once I had the problem that all of the sudden my deployment scripts did not run anymore. The mystery was, that I could still run the script from command line with the Hudson user account. The reason was simple but took a while to uncover. There was a java upgrade from 5 to 6. Both versions were available. After comparing the environment variables, there was a difference in the path. The problem was that the new path was set in the global .profile. But Hudson does not open an interactive shell, therefore the .profile will not be executed. If you have a problem like this, you can put the initialization in the .env file (or whatever the filename is for your system), because this will be run regardless if it is a interactive shell or not. Alternatively you can configure Hudson to set it on master or node/slave level.
if you want a command to not break the 'build' as a failure you have to add #! in front of the command to prevent the flags -xe which produce this behaviour.