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

conhost can take over keyboard keys when launched from .exe symbolic link #10647

New Issue
Closed
opened 2026-01-31 02:26:28 +00:00 by claunia · 12 comments
claunia commented 2026-01-31 02:26:28 +00:00
Owner
Copy Link

Originally created by @DHowett on GitHub (Sep 17, 2020).

Originally assigned to: @miniksa on GitHub.

You're going to think this is crazy talk, but hear me out.

Using Vb (19041) conhost, do this:

cmd

mklink %USERPROFILE%\cmd_link.exe c:\windows\system32\cmd.exe

Open up File Explorer to %USERPROFILE% and double-click cmd_link.exe.

If you have a repro, any time you press L that console window will come to the foreground.

Wild, right?

It looks like the shell invokes cmd_link.exe with STARTF_TITLEISLINKNAME and an lpTitle of ...\cmd_link.exe.

STARTF_TITLEISLINKNAME: The lpTitle member contains the path of the shortcut file (.lnk) that the user invoked to start this process. This is typically set by the shell when a .lnk file pointing to the launched application is invoked. Most applications will not need to set this value.

It turns out that this excerpt from the documentation is a fabrication, and lpTitle might point to an .exe file.

Anyway, when the title is the "link name", conhost uses the title to find the .lnk file. It gets in here:

f91b53d5fd/src/interactivity/win32/SystemConfigurationProvider.cpp (L94-L106)

This fails. s_GetLinkValues attempts to get an IShellLink... for what is clearly not a .lnk file. However, check out that last parameter.

It's never cleared. It's filled with stack garbage.

Moments later, we do this:

f91b53d5fd/src/interactivity/win32/SystemConfigurationProvider.cpp (L114-L115)

...which sets up the hotkey that will later be used to activate the window, down here:

f91b53d5fd/src/interactivity/win32/window.cpp (L350-L353)

We should probably zero-initialize wHotKey.

Originally created by @DHowett on GitHub (Sep 17, 2020). Originally assigned to: @miniksa on GitHub. You're going to think this is crazy talk, but hear me out. Using Vb (19041) conhost, do this: ### cmd ``` mklink %USERPROFILE%\cmd_link.exe c:\windows\system32\cmd.exe ``` Open up File Explorer to `%USERPROFILE%` and _double-click `cmd_link.exe`_. If you have a repro, any time you press <kbd>L</kbd> that console window will come to the foreground. Wild, right? It looks like the shell invokes `cmd_link.exe` with `STARTF_TITLEISLINKNAME` and an `lpTitle` of `...\cmd_link.exe`. > **`STARTF_TITLEISLINKNAME`**: The `lpTitle` member contains the path of the shortcut file (.lnk) that the user invoked to start this process. This is typically set by the shell when a .lnk file pointing to the launched application is invoked. Most applications will not need to set this value. It turns out that this [excerpt from the documentation](https://docs.microsoft.com/windows/win32/api/processthreadsapi/ns-processthreadsapi-startupinfoa) is a fabrication, and `lpTitle` might point to an `.exe` file. Anyway, when the title is the "link name", conhost uses the title to find the `.lnk` file. It gets in here: https://github.com/microsoft/terminal/blob/f91b53d5fdc5b20387e297357668f5f14a795d4c/src/interactivity/win32/SystemConfigurationProvider.cpp#L94-L106 This fails. `s_GetLinkValues` attempts to get an `IShellLink...` for what is clearly not a `.lnk` file. However, check out that last parameter. It's never cleared. It's filled with stack garbage. Moments later, we do this: https://github.com/microsoft/terminal/blob/f91b53d5fdc5b20387e297357668f5f14a795d4c/src/interactivity/win32/SystemConfigurationProvider.cpp#L114-L115 ...which sets up the hotkey that will later be used to activate the window, down here: https://github.com/microsoft/terminal/blob/f91b53d5fdc5b20387e297357668f5f14a795d4c/src/interactivity/win32/window.cpp#L350-L353 We should probably zero-initialize `wHotKey`.
claunia commented 2026-01-31 02:26:29 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Sep 17, 2020):

It hits randomly because, well, it's stack memory. Sometimes it lines up and happens to be L, and sometimes it stays \0 and everything is fine.

@DHowett commented on GitHub (Sep 17, 2020): It hits randomly because, well, it's stack memory. Sometimes it lines up and happens to be `L`, and sometimes it stays `\0` and everything is fine.
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Sep 17, 2020):

Notes for future reproducers: It doesn't reproduce at all in a debug build. You will likely figure out why.

@DHowett commented on GitHub (Sep 17, 2020): Notes for future reproducers: It doesn't reproduce at all in a debug build. You will likely figure out why.
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@dmex commented on GitHub (Sep 17, 2020):

We should probably zero-initialize wHotKey.
It doesn't reproduce at all in a debug build.

You can reproduce uninitialized variables more frequently by disabling ASLR for the debug binaries. Add to the debug vcxproj:

<Link>
    <RandomizedBaseAddress>false</RandomizedBaseAddress>
</Link>
@dmex commented on GitHub (Sep 17, 2020): > We should probably zero-initialize wHotKey. > It doesn't reproduce at all in a debug build. You can reproduce uninitialized variables more frequently by disabling ASLR for the debug binaries. Add to the debug vcxproj: ``` <Link> <RandomizedBaseAddress>false</RandomizedBaseAddress> </Link> ```
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@eryksun commented on GitHub (Sep 17, 2020):

It turns out that this excerpt from the documentation is a fabrication, and lpTitle might point to an .exe file.

The shell team should be notified about their bug. The shell API (e.g. ShellExecuteExW) shouldn't use STARTF_TITLEISLINKNAME for a filesystem symlink. This flag means the title is the path of a shell link (.lnk), not any other kind of link.

mklink %USERPROFILE%\cmd_link.exe c:\windows\system32\cmd.exe

Off topic: a symlink to an executable usually has to be in the same directory as the executable, even though it happens to work from another directory in a simple case such as cmd.exe. The system uses the directory of the symlink itself as the application directory instead of resolving its final path. The real application directory may be required for finding private DLLs and assemblies, among other resources.

@eryksun commented on GitHub (Sep 17, 2020): > It turns out that this excerpt from the documentation is a fabrication, and lpTitle might point to an .exe file. The shell team should be notified about their bug. The shell API (e.g. `ShellExecuteExW`) shouldn't use `STARTF_TITLEISLINKNAME` for a filesystem symlink. This flag means the title is the path of a shell link (.lnk), not any other kind of link. --- > mklink %USERPROFILE%\cmd_link.exe c:\windows\system32\cmd.exe Off topic: a symlink to an executable usually has to be in the same directory as the executable, even though it happens to work from another directory in a simple case such as cmd.exe. The system uses the directory of the symlink itself as the application directory instead of resolving its final path. The real application directory may be required for finding private DLLs and assemblies, among other resources.
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@zadjii-msft commented on GitHub (Sep 17, 2020):

This is gonna sound crazy, but we've totally seen this bug once before. I totally forgot what the resolution was, but maybe @miniksa does.

@zadjii-msft commented on GitHub (Sep 17, 2020): This is gonna sound crazy, but we've totally seen this bug once before. I totally forgot what the resolution was, but maybe @miniksa does.
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Sep 22, 2020):

Filed MSFT-29529566 for the folks who own ShellExecute. 😄

@DHowett commented on GitHub (Sep 22, 2020): Filed MSFT-29529566 for the folks who own ShellExecute. :smile:
claunia commented 2026-01-31 02:26:30 +00:00
Author
Owner
Copy Link

@RJohnsonPGH commented on GitHub (Sep 23, 2020):

I am experiencing this in Powershell as well, so doesn't seem to be limited to just Terminal

@RJohnsonPGH commented on GitHub (Sep 23, 2020): I am experiencing this in Powershell as well, so doesn't seem to be limited to just Terminal
claunia commented 2026-01-31 02:26:31 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Sep 23, 2020):

Conhost, the windows console host, also lives in this repository. That is the application powershell uses to host itself.

@DHowett commented on GitHub (Sep 23, 2020): Conhost, the windows console host, also lives in this repository. That is the application powershell uses to host itself.
claunia commented 2026-01-31 02:26:31 +00:00
Author
Owner
Copy Link

@RJohnsonPGH commented on GitHub (Sep 23, 2020):

Ah, that explains it.

Additionally, my repro doesn't involve symlinks at all, but rather shortcuts to powershell.exe which obviously is a .lnk.

Example target:

C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe -File \\fileserver\Script.ps1

will reliably take focus when lower case 'L' is pressed.

Edit: After re-reading the whole thread, I see now that the issue with shortcuts triggering this is what is primarily described in the first comment and that the issue with symlinks is a combination of a shell bug and this bug. Sorry for any confusion on my part.

@RJohnsonPGH commented on GitHub (Sep 23, 2020): Ah, that explains it. Additionally, my repro doesn't involve symlinks at all, but rather shortcuts to powershell.exe which obviously is a `.lnk`. Example target: `C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe -File \\fileserver\Script.ps1` will reliably take focus when lower case 'L' is pressed. Edit: After re-reading the whole thread, I see now that the issue with shortcuts triggering this is what is primarily described in the first comment and that the issue with symlinks is a combination of a shell bug and this bug. Sorry for any confusion on my part.
claunia commented 2026-01-31 02:26:31 +00:00
Author
Owner
Copy Link

@miniksa commented on GitHub (Jan 4, 2021):

This was MSFT-14659706 and my resolution there was literally to initialize the variables so this didn't happen. So I'm going to go figure out what happened to that fix.

@miniksa commented on GitHub (Jan 4, 2021): This was MSFT-14659706 and my resolution there was literally to initialize the variables so this didn't happen. So I'm going to go figure out what happened to that fix.
claunia commented 2026-01-31 02:26:31 +00:00
Author
Owner
Copy Link

@miniksa commented on GitHub (Jan 4, 2021):

Apparently I made this fix on the OS side on the same day and approximate time as MSFT-14612110. We were working in two repositories back then, the main OS one and our own "OpenConsole" one in preparation to one day do open engineering... like we do now! We had a replication bot "Git2Git" to move things back and forth. Somehow I dispatched both replications at the same time and the one for this fix looked like a revert of the one of the other fix. But it actually packed in this change too at the same time. I abandoned the replication and the fix was then lost to the sands of time. Human error... mine this time. Not a systemic issue, as far as I can see.

I will go dig the fix out of the sand and bring it in.

@miniksa commented on GitHub (Jan 4, 2021): Apparently I made this fix on the OS side on the same day and approximate time as MSFT-14612110. We were working in two repositories back then, the main OS one and our own "OpenConsole" one in preparation to one day do open engineering... like we do now! We had a replication bot "Git2Git" to move things back and forth. Somehow I dispatched both replications at the same time and the one for this fix looked like a revert of the one of the other fix. But it actually packed in this change too at the same time. I abandoned the replication and the fix was then lost to the sands of time. Human error... mine this time. Not a systemic issue, as far as I can see. I will go dig the fix out of the sand and bring it in.
claunia commented 2026-01-31 02:26:32 +00:00
Author
Owner
Copy Link

@gerardog commented on GitHub (Oct 20, 2022):

Today I had deja vu between this one and created PowerShell/PowerShell#18333

@gerardog commented on GitHub (Oct 20, 2022): Today I had deja vu between this one and created PowerShell/PowerShell#18333
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-Server Issue-Bug Priority-2 Product-Conhost Resolution-Fix-Committed
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#10647
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?