Improve shared memory protection #1464
Merged
Conversation
This file contains 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
Signed-off-by: DL6ER <[email protected]>
…aware of a possible DEBUG_SHMEM flag when creating the shared memory segments Signed-off-by: DL6ER <[email protected]>
… objects. The scenario we are resolving here is the following: FTL is already running and someone(/-thing) deletes all the shared memory objects. If another instance of FTL is then started, it doesn't know a previous process is running and creates new SHM objects for itself. Both processes can - in theory - run just fine without touching each other as they are both actually using *different* shared memory objects. You are right when you think this is bizarre. The reason is that the first one has the shared memory objects still mapped into memory, i.e., "deleting" them from disk only removes the visible file handles. The next FTL instance creates files with the very same name, however, they are also new and distict files, i.e., their memory will point elsewhere. Both instances can now run in parallel just fine *until* one of them needs to resize a shared memory objects. If one of the shared memory events now gets resized to a size larger than it was before BUT smaller than what the other FTL instance is expecting, the other instance will instantaneously crash with a SIGSEGV (Bus error). This commit resolves this by storing the PID of the SHM object creator in the settings object. Each FTL instance reloads this shared memory instance now before performing any potentially dangerous operation and checks if the shared memory files on disk are still owned by this process. If this is not the case, we are in serious trouble and exit immediately. This should allow the second instance (you could call it the "rightful owner" of the current existing SHM objects) a fairly good chance to never even notice this and continue to operate just fine. Signed-off-by: DL6ER <[email protected]>
…TERM) when trying to restart FTL really early Signed-off-by: DL6ER <[email protected]>
…imes. The PID in this file will later be overwritten after forking Signed-off-by: DL6ER <[email protected]>
Signed-off-by: DL6ER <[email protected]>
This was
linked to
issues
Nov 3, 2022
yubiuser
reviewed
Nov 5, 2022
yubiuser
approved these changes
Nov 6, 2022
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
By submitting this pull request, I confirm the following:
How familiar are you with the codebase?:
10
Store and check shared memory ownership before resizing shared memory objects.
The scenario we are resolving here is the following: FTL is already running and someone(/-thing) deletes all the shared memory objects. If another instance of FTL is then started, it doesn't know a previous process is running and creates new SHM objects for itself. Both processes can - in theory - run just fine without touching each other as they are both actually using different shared memory objects. You are right when you think this is bizarre. The reason is that the first one has the shared memory objects still mapped into memory, i.e., "deleting" them from disk only removes the visible file handles. The next FTL instance creates files with the very same name, however, they are also new and distinct files, i.e., their memory will point elsewhere. Both instances can now run in parallel just fine until one of them needs to resize a shared memory objects. If one of the shared memory events now gets resized to a size larger than it was before BUT smaller than what the other FTL instance is expecting, the other instance will instantaneously crash with a
SIGSEGV(a bus error to be precise).This PR resolves this by storing the PID of the SHM object creator in the settings object.
Each FTL instance reloads this shared memory instance now before performing any potentially dangerous operation and checks if the shared memory files on disk are still owned by this process. If this is not the case, we are in serious trouble and exit immediately. This should allow the second instance (you could call it the "rightful owner" of the current existing SHM objects) a fairly good chance to never even notice this and continue to operate just fine.
Resolves the root causes behind #1463, #1448 and possibly other similar issues in the past by preventing a certain "mis"configuration to have the bad effect it had for these users.