Original repo broken service handle missing#11
Open
SkybuckFlying wants to merge 4 commits intotmcdos:masterfrom
Open
Original repo broken service handle missing#11SkybuckFlying wants to merge 4 commits intotmcdos:masterfrom
SkybuckFlying wants to merge 4 commits intotmcdos:masterfrom
Conversation
…k detachment. The RamService was observed to cause hangs during Windows 11 shutdown. The hypothesized cause is a race condition with the Service Control Manager (SCM). The `DetachRamDisk` function, called directly during `ServiceStop` and `ServiceShutdown` events, can be a long-running operation. The SCM imposes a strict timeout for services to respond to control signals. If the service fails to respond promptly, the SCM forcibly terminates it. This premature termination likely left the RAM disk in an unstable state, causing the OS shutdown process to hang while waiting for a resource that was not cleanly released. This commit refactors the service's shutdown logic to prevent this timeout and ensure a graceful exit: - **Introduced `TShutdownThread`**: A dedicated thread is now responsible for executing the `DetachRamDisk` function. - **Asynchronous Shutdown**: On `ServiceStop` or `ServiceShutdown`, instead of detaching the disk directly, the service now creates and starts the `TShutdownThread`. This allows the main service thread to remain responsive and acknowledge the SCM request immediately. - **Graceful Wait**: The main `ServiceExecute` loop, upon termination, now safely waits for the `TShutdownThread` to complete its execution. This guarantees that the RAM disk is fully detached and cleaned up before the service process exits. - **State Management**: The `FShutdownThread` handle is initialized to `nil` on service start to ensure clean state management. This change ensures the service signals its compliance with the shutdown request in a timely manner while allowing the potentially lengthy disk detachment process to complete safely in the background, resolving the system hang. - (Note: RamService.dpr assumes a .\bin folder exists for outputting the binary, if this bin folder does not exist, the build might fail, either user must create bin folder or adjust project settings and remove bin from project settings.) - First initial improvement, however, the system could still hang during long operations. The solution would be to periodically inform the system the shutdown service is still busy. These improvements will be attempted in a next commit.
The RamDisk service could hang or be terminated prematurely by the Service Control Manager (SCM) during startup or shutdown. This was because the service did not report its status while performing long-running operations like creating or removing the RAM disk. The SCM would interpret this lack of communication as the service being unresponsive and would terminate it. This commit addresses the issue by implementing proper status reporting in `SrvMain.pas`. The `ServiceStart` and `ServiceStop` procedures now: 1. Set a `WaitHint` of 60 seconds, providing a generous timeout for the operations. 2. Periodically call `ReportStatus` with an incrementing `CheckPoint` value. This ensures the SCM is kept informed of the service's progress (SERVICE_START_PENDING, SERVICE_STOP_PENDING) and prevents it from incorrectly flagging the service as hung, thereby improving service stability and reliability.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
I've decided to upload the broken code, (To a seperate branch called: "original_repo_broken_ServiceHandleMissing")
As far as I can tell the broken code is not really that big of an issue.
To fix it, it would probably require storing the ServiceHandle into the object.
h_sv is created somewhere, can't remember clearly, maybe storing that in ServiceHandle member field could fix this code.
For now I ran out of steam trying to fix this.
But perhaps you will find this code interesting to see what it does... there are some other fixes in here too.. in create ram and such in case for whatever reason ram disk creation fails, I could imagine this might happen if out of memory or something...