starred/terminal
1
0
Fork 0
mirror of https://github.com/microsoft/terminal.git synced 2026-02-04 05:35:20 +00:00
Code Issues 19.0k Packages Projects Releases 20 Wiki Activity

[DefTerm] Some invocations can cause deadlocks in terminal + another GUI application in Attach mode #18603

New Issue
Closed
opened 2026-01-31 06:19:02 +00:00 by claunia · 4 comments
claunia commented 2026-01-31 06:19:02 +00:00
Owner
Copy Link

Originally created by @DHowett on GitHub (Oct 4, 2022).

The following conditions must be true:

  1. Terminal is set as the default
  2. Terminal is set to "Attach" new instances "to the most recently used window"
  3. Terminal is running

If so, the following application will deadlock:

#define WIN32_LEAN_AND_MEAN
#include <windows.h>
#include <stdlib.h>
#include <string_view>

int APIENTRY wWinMain(HINSTANCE hInstance, HINSTANCE, LPWSTR, int)
{
	static constexpr auto szWindowClass{ L"BlahWindowClass" };
	static constexpr std::wstring_view szMessage{ L"Launch Windows Terminal w/ \"Attach to most recent...\" set\nClick anywhere to deadlock" };

	auto WndProc = [](HWND hWnd, UINT message, WPARAM wParam, LPARAM lParam) -> LRESULT {
		switch (message)
		{
		case WM_LBUTTONUP:
			system("cmd /s /c echo hi");
			break;
		case WM_PAINT:
		{
			PAINTSTRUCT ps;
			HDC hdc = BeginPaint(hWnd, &ps);
			RECT r{ 0, 0, 500, 200 };
			DrawTextW(hdc, szMessage.data(), szMessage.size(), &r, DT_LEFT);
			EndPaint(hWnd, &ps);
		}
		break;
		case WM_DESTROY:
			PostQuitMessage(0);
			break;
		default:
			return DefWindowProcW(hWnd, message, wParam, lParam);
		}
		return 0;
	};

	WNDCLASSEXW wcex{};
	wcex.cbSize = sizeof(WNDCLASSEX);
	wcex.lpfnWndProc = static_cast<LRESULT(CALLBACK*)(HWND, UINT, WPARAM, LPARAM)>(WndProc);
	wcex.hInstance = hInstance;
	wcex.hCursor = LoadCursorW(nullptr, (LPCWSTR)IDC_ARROW);
	wcex.hbrBackground = (HBRUSH)(COLOR_WINDOW + 1);
	wcex.lpszClassName = szWindowClass;
	RegisterClassExW(&wcex);

	HWND hWnd = CreateWindowExW(0, szWindowClass, L"Title", WS_OVERLAPPEDWINDOW, CW_USEDEFAULT, 0, 500, 200, nullptr, nullptr, hInstance, nullptr);

	ShowWindow(hWnd, SW_SHOWDEFAULT);
	UpdateWindow(hWnd);

	MSG msg;
	while (GetMessage(&msg, nullptr, 0, 0))
	{
		TranslateMessage(&msg);
		DispatchMessage(&msg);
	}

	return (int)msg.wParam;
}

Save it as foo.cpp and compile:

cl /std:c++17 foo.cpp /link onecoreuap.lib

Launch foo.exe and follow the instruction.

Windows Terminal attempts to bring itself to the foreground by calling AttachThreadInput, BringToForeground and AttachThreadInput in rapid succession.
When the application that is currently in the foreground is calling out to system() to spawn a commandline application, BringToForeground will fail: nobody
is running a message loop, and therefore no win32 messages are being drained. That application is waiting for system() to return before processing further
messages.

This may seem like a theoretical problem, but I encountered it while trying to play The Battle for Wesnoth while on vacation 😀

Originally created by @DHowett on GitHub (Oct 4, 2022). The following conditions must be true: 1. Terminal is set as the default 2. Terminal is set to "Attach" new instances "to the most recently used window" 3. Terminal is running If so, the following application will deadlock: ```c++ #define WIN32_LEAN_AND_MEAN #include <windows.h> #include <stdlib.h> #include <string_view> int APIENTRY wWinMain(HINSTANCE hInstance, HINSTANCE, LPWSTR, int) { static constexpr auto szWindowClass{ L"BlahWindowClass" }; static constexpr std::wstring_view szMessage{ L"Launch Windows Terminal w/ \"Attach to most recent...\" set\nClick anywhere to deadlock" }; auto WndProc = [](HWND hWnd, UINT message, WPARAM wParam, LPARAM lParam) -> LRESULT { switch (message) { case WM_LBUTTONUP: system("cmd /s /c echo hi"); break; case WM_PAINT: { PAINTSTRUCT ps; HDC hdc = BeginPaint(hWnd, &ps); RECT r{ 0, 0, 500, 200 }; DrawTextW(hdc, szMessage.data(), szMessage.size(), &r, DT_LEFT); EndPaint(hWnd, &ps); } break; case WM_DESTROY: PostQuitMessage(0); break; default: return DefWindowProcW(hWnd, message, wParam, lParam); } return 0; }; WNDCLASSEXW wcex{}; wcex.cbSize = sizeof(WNDCLASSEX); wcex.lpfnWndProc = static_cast<LRESULT(CALLBACK*)(HWND, UINT, WPARAM, LPARAM)>(WndProc); wcex.hInstance = hInstance; wcex.hCursor = LoadCursorW(nullptr, (LPCWSTR)IDC_ARROW); wcex.hbrBackground = (HBRUSH)(COLOR_WINDOW + 1); wcex.lpszClassName = szWindowClass; RegisterClassExW(&wcex); HWND hWnd = CreateWindowExW(0, szWindowClass, L"Title", WS_OVERLAPPEDWINDOW, CW_USEDEFAULT, 0, 500, 200, nullptr, nullptr, hInstance, nullptr); ShowWindow(hWnd, SW_SHOWDEFAULT); UpdateWindow(hWnd); MSG msg; while (GetMessage(&msg, nullptr, 0, 0)) { TranslateMessage(&msg); DispatchMessage(&msg); } return (int)msg.wParam; } ``` Save it as `foo.cpp` and compile: ``` cl /std:c++17 foo.cpp /link onecoreuap.lib ``` Launch `foo.exe` and follow the instruction. --- Windows Terminal attempts to bring itself to the foreground by calling `AttachThreadInput`, `BringToForeground` and `AttachThreadInput` in rapid succession. When the application that is currently in the foreground is calling out to `system()` to spawn a commandline application, `BringToForeground` will fail: nobody is running a message loop, and therefore no win32 messages are being drained. That application is waiting for `system()` to return before processing further messages. This may seem like a theoretical problem, but I encountered it while trying to play [The Battle for Wesnoth](https://www.wesnoth.org/) while on vacation :grinning:
claunia commented 2026-01-31 06:19:03 +00:00
Author
Owner
Copy Link

@ghost commented on GitHub (Oct 18, 2022):

:tada:This issue was addressed in #14195, which has now been successfully released as Windows Terminal v1.15.2874.🎉

Handy links:

@ghost commented on GitHub (Oct 18, 2022): :tada:This issue was addressed in #14195, which has now been successfully released as `Windows Terminal v1.15.2874`.:tada: Handy links: * [Release Notes](https://github.com/microsoft/terminal/releases/tag/v1.15.2874) * [Store Download](https://www.microsoft.com/store/apps/9n8g5rfz9xk3?cid=storebadge&ocid=badge)
claunia commented 2026-01-31 06:19:03 +00:00
Author
Owner
Copy Link

@ghost commented on GitHub (Dec 14, 2022):

:tada:This issue was addressed in #14195, which has now been successfully released as Windows Terminal Preview v1.16.3463.0 and v1.16.3464.0.🎉

Handy links:

@ghost commented on GitHub (Dec 14, 2022): :tada:This issue was addressed in #14195, which has now been successfully released as `Windows Terminal Preview v1.16.3463.0 and v1.16.3464.0`.:tada: Handy links: * [Release Notes](https://github.com/microsoft/terminal/releases/tag/v1.16.3464.0) * [Store Download](https://www.microsoft.com/store/apps/9n8g5rfz9xk3?cid=storebadge&ocid=badge)
claunia commented 2026-01-31 06:19:04 +00:00
Author
Owner
Copy Link

@FuPeiJiang commented on GitHub (Jan 10, 2026):

@DHowett

#14195 does fix the deadlock, but now it's a delay of 5000ms, which is a bit long

SendMessageTimeoutW(oldForegroundWindow, WM_NULL, 0, 0, SMTO_NOTIMEOUTIFNOTHUNG | SMTO_BLOCK | SMTO_ABORTIFHUNG, 1000, nullptr)

using SMTO_NOTIMEOUTIFNOTHUNG ignores the 1000ms timeout, and it takes the internal timeout of 5 seconds for it to be detected as hung (since it was newly/recently hung)

https://learn.microsoft.com/en-us/windows/win32/api/winuser/nf-winuser-ishungappwindow

I really have no solution though, only decreasing the timeout to 50ms

SendMessageTimeoutW(oldForegroundWindow, WM_NULL, 0, 0, SMTO_BLOCK | SMTO_ABORTIFHUNG, 50, nullptr)

A potential solution:
Since newly spawned processes can always call SetForegroundWindow (or equivalent) (for reasons I'm not sure of)
WindowsTerminal.exe could just spawn SetForegroundWindow.exe passing the HWND as command line arg, instead of calling a function, the exe could be made incredibly small using https://github.com/ayaka14732/TinyPE-on-Win10
Since there are already other .exe in the distributed folder, SetForegroundWindow.exe wouldn't be too out of place

The real "theoretical" problem would be a hang happening between the two AttachThreadInput, after the thread has been detected as "not hung"

@FuPeiJiang commented on GitHub (Jan 10, 2026): @DHowett #14195 does fix the deadlock, but now it's a delay of 5000ms, which is a bit long ```cpp SendMessageTimeoutW(oldForegroundWindow, WM_NULL, 0, 0, SMTO_NOTIMEOUTIFNOTHUNG | SMTO_BLOCK | SMTO_ABORTIFHUNG, 1000, nullptr) ``` using `SMTO_NOTIMEOUTIFNOTHUNG` ignores the 1000ms timeout, and it takes the internal timeout of 5 seconds for it to be detected as hung (since it was newly/recently hung) <https://learn.microsoft.com/en-us/windows/win32/api/winuser/nf-winuser-ishungappwindow> I really have no solution though, only decreasing the timeout to 50ms ```cpp SendMessageTimeoutW(oldForegroundWindow, WM_NULL, 0, 0, SMTO_BLOCK | SMTO_ABORTIFHUNG, 50, nullptr) ``` A potential solution: Since newly spawned processes can always call `SetForegroundWindow` (or equivalent) (for reasons I'm not sure of) WindowsTerminal.exe could just spawn `SetForegroundWindow.exe` passing the HWND as command line arg, instead of calling a function, the exe could be made incredibly small using https://github.com/ayaka14732/TinyPE-on-Win10 Since there are already other `.exe` in the distributed folder, `SetForegroundWindow.exe` wouldn't be too out of place The real "theoretical" problem would be a hang happening between the two `AttachThreadInput`, after the thread has been detected as "not hung"
claunia commented 2026-01-31 06:19:04 +00:00
Author
Owner
Copy Link

@FuPeiJiang commented on GitHub (Jan 10, 2026):

some documentation about this behavior

If the windows use the same input queue, the message is sent synchronously, first to the window procedure of the top-level window being deactivated, then to the window procedure of the top-level window being activated.

https://learn.microsoft.com/en-us/windows/win32/inputdev/wm-activate

Five Things Every Win32 Developer Should Know

Synchronous Focus

  • A has focus, B calls SetFocus(hwndB)
  • B must wait for A to release focus since
    input is now synchronous
  • If A is hung, then B hangs too

Synchronous Input

  • In the queue are two input messages, first
    a message for A, then a message for B.
  • B calls GetMessage – sees message for A
    and says “I can’t process my input until A
    processes its input first”
  • If A is hung, then B never processes input
@FuPeiJiang commented on GitHub (Jan 10, 2026): some documentation about this behavior > If the windows use the same input queue, the message is sent synchronously, first to the window procedure of the top-level window being deactivated, then to the window procedure of the top-level window being activated. <https://learn.microsoft.com/en-us/windows/win32/inputdev/wm-activate> > Five Things Every Win32 Developer Should Know Synchronous Focus * A has focus, B calls SetFocus(hwndB) * B must wait for A to release focus since input is now synchronous * If A is hung, then B hangs too Synchronous Input * In the queue are two input messages, first a message for A, then a message for B. * B calls GetMessage – sees message for A and says “I can’t process my input until A processes its input first” * If A is hung, then B never processes input
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
⛺ Reserved
A11yCO
A11yMAS
A11ySev1
A11ySev2
A11ySev3
A11yTTValidated
A11yUsable
A11yVoiceAccess
A11yWCAG
Area-Accessibility
Area-AtlasEngine
Area-AzureShell
Area-Build
Area-Build
Area-Chat
Area-CmdPal
Area-CodeHealth
Area-Commandline
Area-CookedRead
Area-DefApp
Area-Extensibility
Area-Fonts
Area-GroupPolicy
Area-i18n
Area-Input
Area-Interaction
Area-Interop
Area-Localization
Area-Output
Area-Performance
Area-Portable
Area-Quality
Area-Remoting
Area-Rendering
Area-Schema
Area-Server
Area-Settings
Area-SettingsUI
Area-ShellExtension
Area-ShellExtension
Area-ShellExtension
Area-Suggestions
Area-Suggestions
Area-TerminalConnection
Area-TerminalControl
Area-Theming
Area-UserInterface
Area-VT
Area-Windowing
Area-WPFControl
AutoMerge
Blocking-Ingestion
Culprit-Centennial
Culprit-WinUI
Disability-All
Disability-Blind
Disability-LowVision
Disability-Mobility
External-Blocked-WinUI3
Fixed
Gathering-Data
good first issue
HCL-E+D
HCL-WindowsTerminal
Help Wanted
Impact-Compatibility
Impact-Compliance
Impact-Correctness
Impact-Visual
In-PR
InclusionBacklog
InclusionBacklog-Windows TerminalWin32
InclusionCommitted-202206
Issue-Bug
Issue-Docs
Issue-Feature
Issue-Feature
Issue-Question
Issue-Samples
Issue-Scenario
Issue-Task
Needs-Attention
Needs-Author-Feedback
Needs-Bisect
Needs-Discussion
Needs-Repro
Needs-Tag-Fix
Needs-Tag-Fix
Needs-Triage
No-Recent-Activity
Priority-0
Priority-1
Priority-2
Priority-3
Product-Cmd.exe
Product-Colortool
Product-Colortool
Product-Colortool
Product-Conhost
Product-Conpty
Product-Meta
Product-Powershell
Product-Terminal
Product-WSL
pull-request
Mirrored from GitHub Pull Request
 Resolution-Answered
Resolution-By-Design
Resolution-Duplicate
Resolution-External
Resolution-Fix-Available
Resolution-Fix-Committed
Resolution-No-Repro
Resolution-Won't-Fix
Severity-Blocking
Severity-Crash
Severity-DataLoss
spam
this-will-be-a-breaking-change
Tracking-External
WindowsTerminal_Win32
Work-Item
zAskModeBug
zInbox-Bug
No Label Area-DefApp Needs-Tag-Fix Priority-1 Product-Terminal Resolution-Fix-Committed Severity-Blocking
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
claunia
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/terminal#18603
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?