Support for browser extension(s) with Native Messaging #608

varjolintu · 2017-05-29T13:49:36Z

Added a new plugin "Browser extension (Native Messaging)" to the project. It listens the stdin asynchronously and transfers messages between a browser extension which supports Native Messaging API. This implementation doesn't restrict the development of another browser extension if anyone wants to do so.

Implements a new compilation option WITH_XC_BROWSER. The only dependency is libsodium which is used for the crypto.

Description

KeePassXC can be used with a safer browser extension (when compared to chromeIPass or passIFox) where no plain text keys or messages concerning login data are transferred. Also, the keepassxc-browser extension is WebExtension compatible (can be used with Firefox 57/Nightly when it's finally released).

There's also support for Unix domain sockets (named pipes on Windows) listening for supporting a proxy application between KeePassXC and the browser extension (disabled by default). This prevents browsers from communicating, launching and closing the KeePassXC instance directly.

Motivation and context

chromeIPass and passIFox are outdated with deprecated API calls
Fixes #28

How has this been tested?

The code was tested with the keepassxc-browser extension, with and without the compilation option. Environments used for are macOS 10.12.5 and Ubuntu Linux 17.04.
Changes are not affecting to other areas of the code. Browser extension plugin is not enabled by default even when the compilation option is being used (works similar to KeePassHTTP plugin).

Also, the code has been tested with multiple databases open, databases closed, and other various situations.

Types of changes

✅ New feature (non-breaking change which adds functionality)

Checklist:

✅ I have read the CONTRIBUTING document. [REQUIRED]
✅ My code follows the code style of this project. [REQUIRED]
✅ All new and existing tests passed. [REQUIRED]
✅ I have compiled and verified my code with -DWITH_ASAN=ON. [REQUIRED]
✅ My change requires a change to the documentation and I have updated it accordingly.

droidmonkey · 2017-06-15T01:07:37Z

This is great work, some of the commits should be squashed prior to merging in the future.

varjolintu · 2017-06-15T06:55:17Z

Thank you. The commits will be squashed in the near future when the final changes are ready. The feature now supports both direct Native Messaging with the browser extension and a proxy method through UDP sockets. The direct method requires option "Automatically save after every change" to be enabled. Without it the database loses any changes when browser exits. The latter method solves the issue but requires additional application (keepassxc-proxy, under development) between the KeePassXC and keepassxc-browser.

droidmonkey · 2017-06-15T13:29:14Z

You seriously rock! Thanks man

mauron85 · 2017-06-20T08:33:20Z

Awesome work. Hat off.

I've found this https://chrome.google.com/webstore/detail/keepassxc-browser/iopaggbpplllidnfmcghoonnokmjoicf/related

~~Is it replacement for keepassxc-browser (which I've never found/used) ?~~
Edit: I'm idiot, it's actually keepassxc-browser. :-)

phoerious

Thanks for updating the pull request.

I marked all issues I found during a first read-through. They are mostly formatting issues, but I also found a few gotyas.

phoerious · 2017-07-18T14:59:02Z

src/browser/BrowserAccessControlDialog.cpp

+#include "ui_BrowserAccessControlDialog.h"
+#include "core/Entry.h"
+
+BrowserAccessControlDialog::BrowserAccessControlDialog(QWidget *parent) :


Reference and pointer operators belong to the type, not the parameter name. No space before and one space after the operator.

Many more instances of this follow.

phoerious · 2017-07-18T15:02:02Z

src/browser/BrowserEntryConfig.cpp

+        return false;
+
+    QVariantMap map = doc.object().toVariantMap();
+    for (QVariantMap::iterator iter = map.begin(); iter != map.end(); ++iter) {


You should use const_iterator here.

phoerious · 2017-07-18T15:11:42Z

src/browser/BrowserOptionDialog.cpp

+    m_ui->bestMatchOnly->setChecked(settings.bestMatchOnly());
+    m_ui->unlockDatabase->setChecked(settings.unlockDatabase());
+    m_ui->matchUrlScheme->setChecked(settings.matchUrlScheme());
+    if (settings.sortByUsername())


Please use braces.

phoerious · 2017-07-18T15:14:06Z

src/browser/BrowserPasswordGeneratorWidget.cpp

@@ -0,0 +1,346 @@
+/*


You seem to duplicate the whole PasswordGeneratorWidget. Isn't there a way to integrate the existing widget?

Maybe. But the UI is different in the plugin than in the normal password generator. For example, there's no need for any generate buttons or the password text field etc. Of course I could give it a try.

I think we should use PasswordGeneratorWidget. About the generate buttons and the password field we can use something like the standalone mode of the current widget. Like an http mode that will hide useless UI controls

Those buttons could be made configurable via properties.

Also, all the config/preference should stay in one place (PasswordGeneratorWidget) and not being duplicated

phoerious · 2017-07-18T15:16:16Z

src/browser/BrowserService.cpp

+        switch(dbWidget->currentMode()) {
+        case DatabaseWidget::None:
+        case DatabaseWidget::LockedMode:
+            break;


Any reason not to use a return here, too?

The function returns false anyway after that?

Yes. But a return makes things clearer and we are less likely to introduce bugs in the future.

phoerious · 2017-07-18T15:52:57Z

src/browser/ChromeListener.cpp

+    const QByteArray ca = base64Decode(m_clientPublicKey);
+    const QByteArray sa = base64Decode(m_secretKey);
+
+    std::memcpy(m, ma.toStdString().data(), ma.length());


Missing bounds check?

crypto_box_open_easy() takes unsigned char arrays as parameters. Is there really a reliable way to convert QByteArray straight to unsigned char*? On solution to this is to wrap the QByteArrays to a std::vector and then use .data(). At least it would be safer than memcpy. What do you think?

What about QByteArray::data()? You can cast it to unsigned if needed.

It's possible yes, but doing a reinterpret_cast for every variable is not gonna look pretty.

It avoids duplicate allocation, though.
If you do it this way, though, you need to check at least if the char arrays have the capacity to hold the data.

phoerious · 2017-07-18T15:56:19Z

src/browser/ChromeListener.cpp

+    return result;
+}
+
+QByteArray ChromeListener::decrypt(const QString encrypted, const QString nonce) const


See comments for encrypt()

This has the same fix as encrypt().

phoerious · 2017-07-18T15:57:20Z

src/browser/Variant.cpp

+    QVariantMap result;
+    const QMetaObject *metaobject = object->metaObject();
+    int count = metaobject->propertyCount();
+    for (int i=0; i<count; ++i) {


Please use spaces between operators.

phoerious · 2017-07-18T15:58:19Z

src/gui/MainWindow.cpp

+{
+    public:
+        BrowserPlugin(DatabaseTabWidget * tabWidget) {
+            m_chromeListener = new ChromeListener(tabWidget);


This should be a smart pointer…

phoerious · 2017-07-18T15:58:31Z

src/gui/MainWindow.cpp

+        }
+
+        ~BrowserPlugin() {
+            delete m_chromeListener;


… and then this isn't needed.

phoerious · 2017-07-19T14:52:05Z

src/browser/ChromeListener.cpp

-    char buf[4] = {};
-    async_read(sd, buffer(buf,sizeof(buf)), transfer_at_least(1), [&](error_code ec, size_t br) {
+    std::array<char, 4> buf;
+    async_read(sd, buffer(buf), transfer_at_least(1), [&](error_code ec, size_t br) {
        if (!ec && br >= 1) {
            uint len = 0;
            for (int i = 0; i < 4; i++) {


Could you maybe use buf.size() here instead of a magic 4 or at least buf.at(i) in the next line instead of buf[i], so we have explicit bounds checks? This is just to be sure we won't introduce any errors in the future.

varjolintu · 2017-07-24T11:59:02Z

The final biggest feature is finally added to the branch. The Password Generator tab is completely removed from the Browser extension settings. It has been replaced with a Supported Browsers tab where user can install all the necessary JSON files for Native Messaging hosts. This way there is no need to use a separate installer script with the browser extension.

This tab can detect already installed host files, and also install/remove them when necessary. Under Windows the JSON file is created to the same directory with the executable, and the path to that JSON file is written to the registry with the correspoding browser.

duk3luk3 · 2017-08-08T12:42:07Z

Hi @varjolintu, as I already told you in https://github.com/varjolintu/keepassxc-browser/issues/13, I'm working on making some big changes to the KeepassHTTP interface (everything in src/http basically).

As far as I can see from a quick glance at your PR branch, you did a cp -r src/http src/browser and went from there. Did you make any big changes to the (user) interface other than replacing the http server with your own stdin / libsodium based server? I'm wondering about how much extra work I will have if I first make my changes to the http code and then have to redo them for the browser code.

I'm also wondering about the stdin listener - does that mean KeePassXC has to be forked from the browser so it can work? Otherwise, how could the extension attach to KeePassXC's stdin?

varjolintu · 2017-08-08T13:53:06Z

There's no big differences in the user interfaces. Just some code cleaning and minor changes. The biggest thing is that the password generator tab is removed from the settings dialog.

And yes, connecting to the stdin listener means the browser must start the KeePassXC process. There's also possibility to use a proxy app which is launced by the browser (handles the Native Messaging) and UDP is used to connect it to KeePassXC on-the-fly.

Keeping it short, any new changes to the UI side should't be hard to do because it's almost identical with the older HTTP side.

duk3luk3 · 2017-08-09T02:01:25Z

The proxy app sounds like a much better idea - e.g. if I want to run both chrome and firefox at the same time and have access to my keepass database from both, or other applications - Thunderbird comes to mind...

And thanks - I'll keep working on the KeePassHTTP code then and port it over once your code gets merged.

cryptomilk · 2017-08-23T06:28:06Z

And yes, connecting to the stdin listener means the browser must start the KeePassXC process. There's also possibility to use a proxy app which is launced by the browser (handles the Native Messaging) and UDP is used to connect it to KeePassXC on-the-fly.

It would be really nice if run on UNIX system it would use unix sockets for the connections. This allows to restrict access to only the user who started the keepassxc process. Not only per file permissions also using:

getsockopt(s, SOL_SOCKET, SO_PEERCRED, ...)

varjolintu · 2017-08-23T08:14:45Z

@cryptomilk, I will definitely look in to this.

cryptomilk · 2017-08-23T08:39:21Z

I'm a C programmer and C++ is like a different language for me, but I'm happy to review code in this area ;)

sedrubal · 2017-10-12T10:54:34Z

If this get's merged until November 14th users that used PassIFox (uses deprecated NPAPI) before, can change to keepassxc-browser (ready for firefox 57) ;)

jeroenev · 2017-10-12T13:52:28Z

@sedrubal i think most firefox users have already moved from passifox to KeePassHttp-Connector
since passifox is alreaddy pretty broken on FF55 and 56, and incompatible with multiprocess

ConorIA · 2017-10-12T13:54:49Z

FWIW, I didn't even find KeePassHttp-Connector, though I didn't look too far once I found keepassxc-browser on the add on portal.

TheZ3ro · 2017-10-18T19:49:37Z

@sedrubal @ConorIA now KeePassHTTP-Connector is suggested on the website instead of PassIFox and ChromeIPass https://keepassxc.org/project

mbrockman1 · 2017-10-22T13:52:11Z

I would love to see this merged, native messaging is so much more secure and better to use!

ahmedelz · 2017-10-27T18:29:01Z

Are these changes going to make it to v2.3.0?

ishitatsuyuki · 2017-11-01T08:07:52Z

All inline reviews seem to be resolved. What's the status of this PR? Merge conflicts only?

varjolintu · 2017-11-01T08:22:36Z

@ishitatsuyuki There's still some work left before merging. Mainly I have been removing boost/asio dependencies and replacing them with Qt only code. Also for proxy feature, UDP is removed and replaced with Unix domain sockets, using named pipes under Windows. When these are solved the merging of the PR could be started.

droidmonkey · 2017-12-30T18:53:35Z

src/browser/CMakeLists.txt

+        set(pkg_config_libs_private "${pkg_config_libs_private} -lsodium")
+    else()
+        message(FATAL_ERROR "libsodium is not installed. Install it, then run CMake again")
+    endif()


Remove everything from if(SODIUM_FOUND) to endif(), it is simply not needed at all. The find_package method handles all of this for you.

droidmonkey · 2017-12-30T18:54:34Z

src/browser/CMakeLists.txt

+    )
+
+    add_library(keepassxcbrowser STATIC ${keepassxcbrowser_SOURCES})
+    target_link_libraries(keepassxcbrowser Qt5::Core Qt5::Concurrent Qt5::Widgets Qt5::Network ${sodium_LIBRARY_RELEASE})


Change ${sodium_LIBRARY_RELEASE} to simply sodium. All the details are handled in the find_package method.

phoerious

Found a couple more issues, partially caused by conflict resolution.

phoerious · 2017-12-31T11:41:50Z

share/linux/org.keepassxc.KeePassXC.appdata.xml

@@ -1,15 +1,16 @@
 <?xml version="1.0" encoding="UTF-8"?>


You messed up conflict resolution here. There should be no diff for this file.

phoerious · 2017-12-31T11:55:36Z

src/browser/HostInstaller.h

+
+private:
+    QString     getTargetPath(SupportedBrowsers browser) const;
+    QString		getBrowserName(SupportedBrowsers browser) const;


Indentation messed up.

phoerious · 2017-12-31T11:56:17Z

src/browser/HostInstaller.h

+public:
+    HostInstaller();
+    bool checkIfInstalled(SupportedBrowsers browser);
+    void installBrowser(SupportedBrowsers browser, const bool enabled, const bool proxy = false, const QString location = QString());


Must have missed that before: Use const& instead of const value for non-integral parameters. Same for other methods in this class. Also don't default-initialize with QString(). Do that with "".

phoerious · 2017-12-31T12:01:16Z

src/proxy/NativeMessagingHost.cpp

+    quint32 length = 0;
+    std::cin.read(reinterpret_cast<char*>(&length), 4);
+    if (!std::cin.eof() && length > 0)
+    {


{ on same line as if.

yan12125 · 2018-01-01T09:23:10Z

src/CMakeLists.txt

+
+    add_custom_command(TARGET ${PROGNAME}
+                       POST_BUILD
+                       COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change /usr/local/opt/qt/lib/QtCore.framework/Versions/5/QtCore "@executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore" ${PROXY_APP_DIR}


This might break non-Homebrew Qt installations (e.g. MacPorts).

Yes. It's possible. AFAIK there's no way to get the exact framework location with cmake. So this is the current workaround. Maybe doing symbolic links to that location might work with MacPorts. I will do some testing. Of course this should work without any additional operations.

I think we should fix that in a separate PR.
The usual way to handle this is to use file() with several hints for possible locations.

yan12125 · 2018-01-03T12:27:48Z

Hello, I'd like to test this patch against MacPorts's Qt, but I got an error before the "changing linking" command. Here are the last few lines of my log:

Scanning dependencies of target KeePassXC
[ 50%] Building CXX object src/CMakeFiles/KeePassXC.dir/main.cpp.o
[ 50%] Building CXX object src/CMakeFiles/KeePassXC.dir/KeePassXC_autogen/mocs_compilation.cpp.o
[ 50%] Linking CXX executable KeePassXC.app/Contents/MacOS/KeePassXC
Deploying app bundle
Copying keepassxc-proxy inside the application
Error copying file "/Users/yen/Projects/keepassxc/build/src/proxy/keepassxc-proxy" to "KeePassXC.app/Contents/MacOS/keepassxc-proxy".
make[2]: *** [src/CMakeFiles/KeePassXC.dir/build.make:139: src/KeePassXC.app/Contents/MacOS/KeePassXC] Error 1
make[2]: *** Deleting file 'src/KeePassXC.app/Contents/MacOS/KeePassXC'
make[1]: *** [CMakeFiles/Makefile2:155: src/CMakeFiles/KeePassXC.dir/all] Error 2
make: *** [Makefile:163: all] Error 2

My build steps are:

git clone https://github.com/keepassxreboot/keepassxc && cd keepassxc
wget "https://github.com/keepassxreboot/keepassxc/pull/608.patch"
patch -Np1 -i 608.patch
mkdir build && cd build
cmake .. -DWITH_XC_BROWSER=ON
make

If I looked into /Users/yen/Projects/keepassxc/build/src/proxy and there's no keepassxc-proxy. If I run make keepassxc-proxy first, the whole build process, including "changing linking" commands, runs fine. I guess add_custom_command commands need more explicit dependency declarations?

Environment: MacPorts, cmake 3.10.1, Qt 5.10.0

varjolintu · 2018-01-03T13:52:34Z

@yan12125 Just running make builds the keepassxc-proxy also. It's not possible separately (at this point). If the copying won't succeed the build must have failed.

Doing the symbolic links to the correct locations with QtCore and QtNetwork frameworks should work. But let's not fill this PR thread with issues. Please do an issue to my fork with a full log. Thanks.

phoerious · 2018-01-03T13:55:49Z

@varjolintu AUTOUIC problems have been fixed in develop, please rebase.

phoerious

🎉

…Qt version

phoerious · 2018-01-04T22:11:11Z

So, no more angry unicorns. Follow-up fixes in fresh PRs, please. 😉

jeroenev · 2018-01-05T15:24:15Z

🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉

phoerious · 2018-01-05T15:47:59Z

Don't be too excited yet. We still need to figure out the store listing before this is working. 😉

TheZ3ro · 2018-01-05T16:19:53Z

Please stop commenting here. I see unicorns every time I load this page 🦄 🦄 🦄 🦄

apiraino · 2018-01-22T16:45:33Z

Hello,

I wanted to give a sneak peek at this new feature, so I cloned the repo and built the develop branch.
However the keepassxc is not connecting to its FF browser plugin.

Also tried to compile with -DCMAKE_BUILD_TYPE=Debug but I could't get anything useful out.

My test env:

Ubuntu 17.10
Firefox 59b2
I'm on 9c641ddf6b5b888994df2ef2ced2541fd86042bb
no firejail

Any hint about that? Thanks!

@TheZ3ro really sorry to comment on this PR - I see the github timeouts, but I didn't really know where else to ask :-)

TheZ3ro · 2018-01-22T17:54:33Z

@apiraino check out this issue: keepassxreboot/keepassxc-browser#3
If it doesn't fit you, please open a new issue as this feature is now effectively part of keepassxc

I'm locking conversation on this. If someone read this message, please open a new issue.
Thanks and sorry for the inconvenience

varjolintu changed the title ~~Feature/chrome kee pass xc~~ Feature/chromeKeePassXC May 29, 2017

varjolintu changed the title ~~Feature/chromeKeePassXC~~ Support for browser extension(s) with Native Messaging May 29, 2017

phoerious added this to the v2.3.0 milestone May 29, 2017

phoerious added the new feature label May 29, 2017

mauron85 mentioned this pull request Jun 22, 2017

Add DBus support #283

Merged

phoerious requested changes Jul 18, 2017

View reviewed changes

phoerious added the feature: Browser label Jul 18, 2017

phoerious reviewed Jul 19, 2017

View reviewed changes

droidmonkey mentioned this pull request Aug 6, 2017

[WIP] Create new KeepassHTTP KeyAcceptDialog #850

Closed

varjolintu mentioned this pull request Sep 6, 2017

Merge changes back to keepassxc varjolintu/keepassxc#6

Closed

RealOrangeOne mentioned this pull request Oct 10, 2017

Support Native Messaging #1054

Closed

droidmonkey requested changes Dec 30, 2017

View reviewed changes

phoerious requested changes Dec 31, 2017

View reviewed changes

yan12125 reviewed Jan 1, 2018

View reviewed changes

phoerious approved these changes Jan 3, 2018

View reviewed changes

yan12125 mentioned this pull request Jan 3, 2018

Failed to build from source with MacPorts varjolintu/keepassxc#21

Closed

varjolintu force-pushed the feature/chromeKeePassXC branch from 90a4201 to 99e20d2 Compare January 3, 2018 15:07

droidmonkey approved these changes Jan 4, 2018

View reviewed changes

keepassxc-browser

06518c5

phoerious force-pushed the feature/chromeKeePassXC branch 5 times, most recently from 8945246 to d427eb1 Compare January 4, 2018 21:49

Update CI Dockerfile to include libsodium and bump minimum supported …

d80c6b4

…Qt version

phoerious force-pushed the feature/chromeKeePassXC branch from d427eb1 to d80c6b4 Compare January 4, 2018 21:56

phoerious merged commit a5a5c67 into keepassxreboot:develop Jan 4, 2018

varjolintu deleted the feature/chromeKeePassXC branch January 11, 2018 06:49

soredake mentioned this pull request Jan 21, 2018

fixes for the keepassxc 2.2.5 version netblue30/firejail#1744

Merged

keepassxreboot locked as too heated and limited conversation to collaborators Jan 22, 2018

phoerious added pr: new feature Pull request that adds a new feature and removed new feature labels Nov 22, 2024

Support for browser extension(s) with Native Messaging #608

Support for browser extension(s) with Native Messaging #608

Conversation

varjolintu commented May 29, 2017 • edited Loading

Description

Motivation and context

How has this been tested?

Types of changes

Checklist:

droidmonkey commented Jun 15, 2017

varjolintu commented Jun 15, 2017

droidmonkey commented Jun 15, 2017

mauron85 commented Jun 20, 2017 • edited Loading

phoerious left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

varjolintu Jul 18, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

varjolintu commented Jul 24, 2017

duk3luk3 commented Aug 8, 2017

varjolintu commented Aug 8, 2017

duk3luk3 commented Aug 9, 2017

cryptomilk commented Aug 23, 2017

varjolintu commented Aug 23, 2017

cryptomilk commented Aug 23, 2017

sedrubal commented Oct 12, 2017

jeroenev commented Oct 12, 2017

ConorIA commented Oct 12, 2017

TheZ3ro commented Oct 18, 2017

mbrockman1 commented Oct 22, 2017

ahmedelz commented Oct 27, 2017

ishitatsuyuki commented Nov 1, 2017

varjolintu commented Nov 1, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

phoerious left a comment • edited Loading

Choose a reason for hiding this comment

phoerious Dec 31, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

phoerious Dec 31, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

varjolintu Jan 1, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yan12125 commented Jan 3, 2018

varjolintu commented Jan 3, 2018

phoerious commented Jan 3, 2018

phoerious left a comment

Choose a reason for hiding this comment

phoerious commented Jan 4, 2018

jeroenev commented Jan 5, 2018

phoerious commented Jan 5, 2018

TheZ3ro commented Jan 5, 2018

apiraino commented Jan 22, 2018

TheZ3ro commented Jan 22, 2018

varjolintu commented May 29, 2017 •

edited

Loading

mauron85 commented Jun 20, 2017 •

edited

Loading

varjolintu Jul 18, 2017 •

edited

Loading

phoerious left a comment •

edited

Loading

phoerious Dec 31, 2017 •

edited

Loading

phoerious Dec 31, 2017 •

edited

Loading

varjolintu Jan 1, 2018 •

edited

Loading