Update xrdp-neutrinordp.c to support side button (forward/back) click

[naruhito]

Hello, thank you for the NeutrinoRDP proxy module feature of xrdp.

While using the feature, we found that so-called forward/back button won't work, and in this PR, we would like you to consider merging this patch to solve the issue. @matt335672

What are the buttons on the side of my mouse called?

[metalefty]

Just for reference, similar fix for the VNC module: #2426

[metalefty]

I don't think sending the mouse button as a keyboard event is a good way to implement this.

Is there any reason not to send mouse event as mouse event as-is?

[matt335672]

I don't have suitable mouse to look at this, but agreed with @metalefty. These should be send as mouse events somehow, if this is possible.

The event numbers you're using (115-118) correspond to the TS_POINTERX_EVENT in [MS-RDPBCGR] 2.2.8.1.1.3.1.1.4. On the input side, we map these events from the input PDU in xrdp_wm_process_input_mousex() if you want to look at that code.

This suggests to me the code you want should look more like this:-

--- a/neutrinordp/xrdp-neutrinordp.c
+++ b/neutrinordp/xrdp-neutrinordp.c
@@ -423,6 +423,26 @@ lxrdp_event(struct mod *mod, int msg, long param1, long param2,
         case 110:
             break;
 
+        case WM_BUTTON8UP:
+            flags = PTR_XFLAGS_BUTTON1;
+            mod->inst->input->ExtendedMouseEvent(mod->inst->input, flags, 0, 0);
+            break;
+
+        case WM_BUTTON8DOWN:
+            flags = PTR_XFLAGS_BUTTON1 | PTR_XFLAGS_DOWN;
+            mod->inst->input->ExtendedMouseEvent(mod->inst->input, flags, 0, 0);
+            break;
+
+        case WM_BUTTON9UP:
+            flags = PTR_XFLAGS_BUTTON2;
+            mod->inst->input->ExtendedMouseEvent(mod->inst->input, flags, 0, 0);
+            break;
+
+        case WM_BUTTON9DOWN:
+            flags = PTR_XFLAGS_BUTTON2 | PTR_XFLAGS_DOWN;
+            mod->inst->input->ExtendedMouseEvent(mod->inst->input, flags, 0, 0);
+            break;
+
         case 200:
             LOG_DEVEL(LOG_LEVEL_DEBUG, "Invalidate request sent from client");
             x = (param1 >> 16) & 0xffff;

I can compile this but I can't test it. Can you try it?

[naruhito]

I agree with you both; yes, mouse events should be treated as they are if possible. The code I made was a quick fix when I was asked by colleagues to address the issue.
I will talk to him and try your code. Please allow me some time, and thanks a lot for the review and advice. @metalefty @matt335672

[naruhito]

Hello @matt335672 @metalefty I tried the code using MX Ergo mouse device from my colleague.

The code almost worked as-is, except for that param1(=x) and param2(=y) were also needed to be passed, as forward/back works only when mouse pointer is over the window of the Web browser.

Because the commit history has become a little bit messy, I will create another PR if needed.

Also, in this PR, 115-118 magic numbers are used, following the other message types.

Constifying the existing magic numbers were a bit hard for me, as some of them are not in the xrdp_constants.h#L236-L258.

[matt335672]

@naruhito - thanks for that. The magic numbers are fine for now - we can sort that out later. I'm happy to merge this.

Regarding the commit history, you could create another PR, or you can squash the 4 commits into one, and we can merge that one.

Are you comfortable with squashing? I can give you a hand if you want to learn about it.

[naruhito]

I've now force-pushed one squashed commit (, was that what you meant?) Thank you again for the advice. @matt335672

For my note:

git rebase -i neutrinolabs/v0.9
pick/squash/squash
git push origin neutrinordp-proxy-sidebutton-click -f

[metalefty]

LGTM. Let's backport this to v0.9.

[matt335672]

@naruhito has made this request against v0.9

I'll merge this and handle the devel branch merge too.

@naruhito - thanks for raising this and testing it too