Handle execve's on Linux #177
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
When an execve is executed, the process's memory is replaced with the memory of a new program. When the new program is also a Ruby interpreter instance, rbspy should continue to gather stack traces. This is particularly important because some Ruby invocation methods, such as rvm, may do an execve to launch the script being invoked by the user. The strategy here is to check the result returned when getting the stack trace from the target process to detect the error that the pointer into the ruby thread state, specifically, has become invalid. If it has, we then reinitialize StackTraceGetter on the basis that if we've lost our thread the whole process is invalid. Currently this doesn't work on MacOS because the error-handling for copying addresses can't distinguish between an invalid read and process exit. Fixing that would require either changing the interface of read-process-memory to surface the mach return codes or just working around it in rbspy. It might also not work as reliably on 32-bit platforms, where there may be a greater chance of the thread state pointer being remapped into the memory of the new process.
Make StackTraceGetter take ProcessHandle instead of a generic, since doing anything meaningful with StackTraceGetter now requires it to support reinitialization. Add a type alias for the boxed stack trace type.
wabain
force-pushed
the
execve-support
branch
from
4a46622
to
3c83f39
Compare
|
Looks good to me and appears to fix #145 👍 |
Closed
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Outdated suggestions cannot be applied.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.
This is a fix for #145, handling
execcalls, for Linux. While the same approach should work for MacOS, extra work would be needed to distinguish between invalid reads and process exit and since I've already let this sit for many months, I'd prefer to land Linux support.The idea here is to check the result returned when getting the stack trace from the target process to detect the error that the pointer to the ruby thread state, specifically, has become invalid. If it has, we then reinitialize StackTraceGetter on the basis that if we've lost the thread the whole process is invalid.
In the test case I added I'm polling until the reinitialization happens. I did this because when I used a fixed timeout I was sometimes getting the trace before reinitialization (one every 500 test runs or so), and I didn't want to play the timeout tweaking game. With the current polling I still get the error "stack base and cfp address out of sync" once every few thousand test runs but I think that error should be transient and shouldn't impact real usage.