DMD currently uses a very hacky form of "sampling" by default to avoid
recording stack traces for all blocks. This makes DMD run faster than when it
records all stack traces.
This patch changes the sampling method used; in fact, it avoids "sampling" at
all. The existence of all heap blocks is now recorded exactly, but by default
we only record an allocation stack for each heap block if a Bernoulli trial
succeeds. This choice works well because getting the stack trace is ~100x
slower than recording the block's existence.
Overall, this approach is simpler and it also gives better output -- the choice
of which blocks to record allocation stacks for is mathematically sound, no
stack trace gets blamed for allocations it didn't do, and block counts and
sizes are now always exact.
Other specific things changed.
- All notion of sampling is removed from the various data structures.
- The --sample-below option is removed in favour of --stacks={partial,full}.
- The format of the JSON output file has changed.
- The names of various test files have changed to reflect concept changes.
The new "num" property lets identical blocks be aggregated in the output. This
patch only uses the "num" property for dead blocks, because that's where the
greatest potential benefit lies, but it could be used for live blocks as well.
On one test case (a complex PDF file) running with --mode=cumulative
--sample-below=1 this patch had the following effects.
- Change in running speed was negligible.
- Compressed output file size dropped from 8.8 to 5.0 MB.
- Compressed output file size dropped from 297 to 50 MB.
- dmd.py runtime (without stack fixing) dropped from 30 to 8 seconds.
Now that defining $DMD is no longer necessary to run DMD, this patch does the
following.
- Removes all the places where we set DMD=1 (test harnesses, etc.)
- Still handles DMD=1, for backwards compatibility.
- Prints "$DMD is undefined" at DMD start-up if appropriate.
- Writes a |null| value for |dmdEnvVar| in the JSON if $DMD is undefined. Bumps
the DMD output version number accordingly.
- Changes a bunch of the test files accordingly, including changing the mode of
script-ignore-alloc-fns.json in order to test a case where $DMD is undefined.
This patch moves profiling mode selection from post-processing (in dmd.py) to
DMD start-up. This will make it easier to add new kinds of profiling, such as
cumulative heap profiling.
Specifically:
- There's a new --mode option. |LiveWithReports| is the default, as it is
currently.
- dmd.py's --ignore-reports option is gone.
- There's a new |mode| field in the JSON output.
- Reports-related operations are now no-ops if DMD isn't in LiveWithReports
mode.
- Diffs are only allowed for output files that have the same mode.
- A new function ResetEverything() replaces the SetSampleBelowSize() and
ClearBlocks(), which were used by the test to change DMD options.
- The tests in SmokeDMD.cpp are split up so they can be run multiple times, in
different modes. The exact combinations of tests and modes has been changed a
bit.