i#5538 memtrace seek, part 10: Top-level headers #5739
Conversation
Adds cached values of the 5 top-level headers to the metrace_stream_t interface and implements this for the readers. Adds checks of these values to invariant_checker. Adds a test with skipped instructions using the skip_unit_tests checked-in trace. Reverses the order of the initial 2 markers and the tid,pid pair sent to the reader to avoid a 0 tid in tools. I am surprised this hasn't caused more problems and I thought it was already this fixed way. Issue: #5538
…online bundles; Disable ordinal checks for serial multithread mode
|
x32 failure is api.startstop assert about sysenter #4604 |
| @@ -221,6 +221,7 @@ template <typename T> class file_reader_t : public reader_t { | |||
| return false; | |||
| } | |||
| // Read the meta entries until we hit the pid. | |||
| std::queue<trace_entry_t> marker_queue; | |||
There was a problem hiding this comment.
What does this intermediate marker_queue accomplish? That the markers are stored after the tid and pid? Add comment.
There was a problem hiding this comment.
Yes, the file has marker,marker,tid,pid but the reader has to fill in the tid+pid fields of memref_t so it is best for it to see the tid+pid first (o/w tools doing per-thread data in serial mode have problems). Added a comment.
There was a problem hiding this comment.
We could consider reordering the original order of the fields in raw2trace.
There was a problem hiding this comment.
That would avoid the reordering multiple pieces of code are having to do right now. But it makes it hard to change things with a version bump b/c the version is not in the first record so we'd be locked into that format? There are tradeoffs.
There was a problem hiding this comment.
Seems fine to skip trace format change for now. Looks like it needs more thought.
Adds a reader subclass to raw2trace for computing the memref_t record count for each chunk. A new marker is inserted in the chunk header which the zipfile skip code uses to obtain the correct ref count when skipping over chunks. Adds a count suppression feature where the view tool prints 0 for the record count for the synthetic timestamp+cpu added after a seek. Updates the seek test with a new trace and new expected output. Issue: #5538
…); update skip test output for prior PR with 0 for inserted time,cpu
|
run arm tests |
|
Hitting the same attach timeouts it seems on Jenkins: #5740 |
We probably should add it to the ignore-list? |
|
Re-run is still hitting the attach ones: someone will have to put effort into fixing or adding to ignore list. Also hit the inv checker on the re-run (not on prior runs): #5724. |
Adds cached values of the 5 top-level headers to the metrace_stream_t interface and implements this for the readers.
Adds checks of these values to invariant_checker.
Adds a test with skipped instructions using the skip_unit_tests checked-in trace.
Reverses the order of the initial 2 markers and the tid,pid pair sent to the reader to avoid a 0 tid in tools. I am surprised this hasn't caused more problems and I thought it was already this fixed way.
Issue: #5538