No samples generated on Windows #348
Comments
|
Can you attach the etl file? |
|
I've now generated a profile with a low sampling rate (to keep the etl as small as possible). |
|
I've done some debugging in etw_gecko.rs and no |
|
Doing |
|
How did you determine which events are read? |
|
I uncommented the |
|
@clemenswasser what do you get when you run the event-types example in the etw-reader crate on your etl file? which includes 2807 SampleProf events. |
|
Yeah. That's the problem. Thanks for your reference data! I get: The only difference is |
|
The problem seems to be that |
|
Could you check if this worked before #299? I don't have a full mental model of how it would have broken it, but we've seen other cases where merging changed the names of events. The example that I'm thinking of is that @vvuk found that, for some reason, for events which should have the name You can also turn your existing ETL into a merged ETL using |
|
I tested with a6704e9 (first commit before #299) |
|
I see, thanks. So it's not caused by a difference in unmerged/merged ETL files. "The handle is invalid" usually means you passed the wrong filename. (Yes, this needs a better error message.) The |
|
No, sorry, my fault the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
When recording any process with samply on Windows, the generated profile.json doesn't contain any samples, even though the etl file contains samples.
I used a matrix multiply sample application that came with Intel VTune compiled to a.exe as an example here, but could observe the same for every application I tested.
When using
--keep-etlI could confirm that the raw etl file does in fact contain samples:But the
samplessections in the profile.json are completely empty, which also causes the errorRangeError: minimumFractionDigits value is out of range..in the Firefox Profiler.The text was updated successfully, but these errors were encountered: