[crashtracker] Enable client to configure which signals to trap #856
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
danielsn
commented
Feb 5, 2025
| SigHandler::Handler(f) => f(signum), | ||
| SigHandler::SigAction(f) => f(signum, sig_info, ucontext), | ||
| }; | ||
| eprintln!("Unexpected signal number {signum}, can't chain the handlers"); |
There was a problem hiding this comment.
What should we do here? abort?
There was a problem hiding this comment.
I guess this could happen if the user code changes the sigaction while we're in the middle of handling a signal? I'm not sure how it might arise.
danielsn
commented
Feb 5, 2025
| unsafe { create_alt_stack()? }; | ||
| } | ||
| let mut old_handlers = HashMap::new(); | ||
| // TODO: if this fails, we end in a situation where handlers have been registered with no chain |
danielsn
commented
Feb 5, 2025
| let boxed_ptr = Box::into_raw(Box::new(old_handlers)); | ||
|
|
||
| let res = OLD_HANDLERS.compare_exchange(ptr::null_mut(), boxed_ptr, SeqCst, SeqCst); | ||
| anyhow::ensure!( |
There was a problem hiding this comment.
This leaks the old_handlers, if we care?
There was a problem hiding this comment.
I don't think so--small one-time leak, right?
danielsn
commented
Feb 5, 2025
| @@ -674,9 +665,11 @@ pub fn restore_old_handlers(inside_signal_handler: bool) -> anyhow::Result<()> { | |||
| anyhow::ensure!(!prev.is_null(), "No crashtracking previous signal handlers"); | |||
| // Safety: The only nonnull pointer stored here comes from Box::into_raw() | |||
| let prev = unsafe { Box::from_raw(prev) }; | |||
There was a problem hiding this comment.
We could flip this around, and only rehydrate into a box when we are ready to drop it
BenchmarksComparisonBenchmark execution time: 2025-02-13 19:07:26 Comparing candidate commit 1aa8fb3 in PR branch Found 4 performance improvements and 0 performance regressions! Performance is the same for 48 metrics, 2 unstable metrics. scenario:credit_card/is_card_number/37828224631000521389798
scenario:credit_card/is_card_number_no_luhn/37828224631000521389798
CandidateCandidate benchmark detailsGroup 1
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
Group 8
Group 9
Group 10
Group 11
Group 12
Group 13
BaselineOmitted due to size. |
danielsn
commented
Feb 5, 2025
danielsn
commented
Feb 5, 2025
danielsn
force-pushed
the
dsn/crashtracker-many-signals
branch
from
8050f45 to
90442a0
Compare
danielsn
commented
Feb 6, 2025
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #856 +/- ##
==========================================
- Coverage 71.90% 71.62% -0.29%
==========================================
Files 324 324
Lines 48077 48226 +149
==========================================
- Hits 34572 34541 -31
- Misses 13505 13685 +180
|
One thing I strongly recommend is: let's put in some defaults! In past features "leaving it up to implementer" often meant unintended inconsistency as things get picked semi-randomly; I think in general it's nice to have a pattern of "if you don't want to choose, we pick a reasonable default, don't worry about it". |
https://github.com/DataDog/libdatadog/pull/856/files#diff-fded76796d1b1945fd14d3cdb6230e4692e94608fc4887d26e2530a004df451aR59 |
Ahaha my bad for scrolling too fast through the PR, and amazing work in getting this in :D :D :D |
sanchda
approved these changes
Feb 10, 2025
| SigHandler::Handler(f) => f(signum), | ||
| SigHandler::SigAction(f) => f(signum, sig_info, ucontext), | ||
| }; | ||
| eprintln!("Unexpected signal number {signum}, can't chain the handlers"); |
There was a problem hiding this comment.
I guess this could happen if the user code changes the sigaction while we're in the middle of handling a signal? I'm not sure how it might arise.
| let boxed_ptr = Box::into_raw(Box::new(old_handlers)); | ||
|
|
||
| let res = OLD_HANDLERS.compare_exchange(ptr::null_mut(), boxed_ptr, SeqCst, SeqCst); | ||
| anyhow::ensure!( |
There was a problem hiding this comment.
I don't think so--small one-time leak, right?
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.
What does this PR do?
Instead of hardcoding the signals to trap, allow the client to specify them
Motivation
We wanted to catch sigabort and sigill in addition to sigsegv and sigbus, and I figured lets just make it generic.
Additional Notes
Anything else we should know when reviewing?
How to test the change?