WIP: the tip of my filterx performance efforts #434
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
…tion_with_text() Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Instead of coding this individually in all FilterXExpr derivatives. Signed-off-by: Balazs Scheidler <[email protected]>
This new method can be used to rewrite the expression tree after parsing in order to execute faster. This is where we can check literal values and potentially evaluate literal expressions. Signed-off-by: Balazs Scheidler <[email protected]>
…mize() Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
We incorrectly called filterx_expr_init_method() directly instead of going through the virtual function pointer. Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Ubuntu has -fno-omit-frame-pointer in CFLAGS which Debian testing does not have, making profiling a lot more difficult on Debian. Signed-off-by: Balazs Scheidler <[email protected]>
Only exprs added as a stmt_expr had location information, meaning that conditions or exprs in ternaries did not have location information. Add those as well. Signed-off-by: Balazs Scheidler <[email protected]>
Sometimes (especially with location tracking bugs) a grammar location may be outside of the dimensions of the source text. Make sure we never index outside of the dimensions of the "lines" array. Signed-off-by: Balazs Scheidler <[email protected]>
… is enabled This is a performance optimization that should eliminate filter_object_repr() calls from the profile. Signed-off-by: Balazs Scheidler <[email protected]>
…lterXReMatchState Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
The special handling for $MSG is long gone, avoid handling $MSG as a macro as it requires extra copying in filterx and there's no functional requirement anymore. Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
bazsi
force-pushed
the
filterx-perf-project
branch
from
3d96664 to
c12a771
Compare
…mentation Instead of passing the arguments in a GPtrArray, let's use a simple stack allocated array, voiding a 2 mallocs and 2 frees in any simple function invocations. Signed-off-by: Balazs Scheidler <[email protected]>
Optimize the removal of multiple elements. Signed-off-by: Balazs Scheidler <[email protected]>
Instead of doing this in the hot path. Also eliminate the related helper function. Signed-off-by: Balazs Scheidler <[email protected]>
… non-zero To match how object-bytes/object-protobuf works. Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
It took me a while to realize why the output is not rendered in our light tests, once I removed $MSG as a macro. This message would have helped me a lot. Signed-off-by: Balazs Scheidler <[email protected]>
With the $MSG as a macro gone, $MSG became a typed value, so it does matter what we set it to. Since setting it to a bytes/protobuf value means that its not a string anymore, we can't render it as a part of a template unless we explicitly ask for the right type hint. This is what this patch changes in the broken testcases. Signed-off-by: Balazs Scheidler <[email protected]>
# Conflicts: # lib/filterx/filterx-grammar.ym
# Conflicts: # lib/filterx/expr-generator.c # lib/filterx/filterx-expr.c # lib/filterx/filterx-expr.h
…hod' into tip/axodepot
…nabled' into tip/axodepot
…es' into tip/axodepot
Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
Signed-off-by: Balazs Scheidler <[email protected]>
To make it easier to understand stackdumps. Signed-off-by: Balazs Scheidler <[email protected]>
bazsi
force-pushed
the
filterx-perf-project
branch
from
c12a771 to
64e6eb5
Compare
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.
Please ignore this PR as it is not intended to be merged. It contains work-in-progress local patches I am yet to extract into separate pull requests.
It is rebased on top of my local merge of all related PRs, so I can maintain those branches separately as they get merged.