From e067310dce3fe017d4d6d482c15397916aeccc15 Mon Sep 17 00:00:00 2001
From: Christoph Urlacher <christoph.urlacher@protonmail.com>
Date: Wed, 5 Aug 2026 19:05:47 +0200
Subject: [PATCH] Add new filter mechanism: Exclude list of function symbols

---
 scripts/Modules/Filters.pm | 57 ++++++++++++++++++++++++++++++++++----
 scripts/Modules/Util.pm    | 48 ++++++++++++++++++++++++++++++++
 2 files changed, 100 insertions(+), 5 deletions(-)

diff --git a/scripts/Modules/Filters.pm b/scripts/Modules/Filters.pm
index 287ffe8..17ae6f5 100644
--- a/scripts/Modules/Filters.pm
+++ b/scripts/Modules/Filters.pm
@@ -94,6 +94,39 @@ my %CONFIGS = (
     },
 );
 
+# All entries are functions part of the native call chain
+my @NATIVE_CALL_FUNCS = (
+    'aot_invoke_native',               # imported-function dispatch (AOT)
+    'aot_call_indirect',               # call_indirect dispatch (AOT)
+    'wasm_interp_call_func_native',    # native dispatch (Interp)
+    'wasm_runtime_invoke_native',      # common
+    'invokeNative',                    # common
+
+    # Imported functions:
+    'host_fail_start_trace',
+    'host_fail_stop_trace',
+    'host_fail_marker_positive',
+    'host_fail_marker_negative',
+    'host_fail_marker_detected',
+    'host_print',
+);
+
+$CONFIGS{no_native_call_instr} = {
+    label   => "Exclude WAMR native call chain (instr)",
+    regions => [
+        map { { func => $_, col => 'p.injection_instr_absolute' } }
+          @NATIVE_CALL_FUNCS
+    ],
+};
+
+$CONFIGS{no_native_call_data} = {
+    label   => "Exclude WAMR native call chain (data)",
+    regions => [
+        map { { func => $_, col => 'g.data_physical_address' } }
+          @NATIVE_CALL_FUNCS
+    ],
+};
+
 # Those will be executed automatically by runner.pl (+ no filter at all)
 # my @DEFAULT_CONFIGS = ('no_aot_data');
 my @DEFAULT_CONFIGS = ();
@@ -117,12 +150,26 @@ sub build_filter_clause {
     return ""
       unless defined $experiment_dir && -f "$experiment_dir/system.elf";
 
+    my $elf = "$experiment_dir/system.elf";
+
     my @filters;
-    for my $pair (@$regions) {
-        my ( $start_sym, $end_sym, $col ) = @$pair;
-        my $start =
-          Util::elf_sym_addr( "$experiment_dir/system.elf", $start_sym );
-        my $end = Util::elf_sym_addr( "$experiment_dir/system.elf", $end_sym );
+    for my $region (@$regions) {
+        my ( $start, $end, $col );
+
+        if ( ref($region) eq 'HASH' ) {
+
+            # Single-function form: resolve [start, end) via symbol + size.
+            $col = $region->{col};
+            ( $start, $end ) = Util::elf_sym_range( $elf, $region->{func} );
+        }
+        else {
+            # Linker-range form: [start_sym, end_sym, col].
+            my ( $start_sym, $end_sym );
+            ( $start_sym, $end_sym, $col ) = @$region;
+            $start = Util::elf_sym_addr( $elf, $start_sym );
+            $end   = Util::elf_sym_addr( $elf, $end_sym );
+        }
+
         next unless defined $start && defined $end && $end > $start;
         push @filters, "$col NOT BETWEEN $start AND @{[$end - 1]}";
     }
diff --git a/scripts/Modules/Util.pm b/scripts/Modules/Util.pm
index a5299e1..b22c5d6 100644
--- a/scripts/Modules/Util.pm
+++ b/scripts/Modules/Util.pm
@@ -188,6 +188,54 @@ sub elf_sym_addr {
     return hex($1);
 }
 
+# Determine the address range of a function symbol
+sub elf_sym_range {
+    my ( $elffile, $sym ) = @_;
+
+    my @lines = qx{nm -nS "$elffile" 2>/dev/null};
+    return () unless @lines;
+
+    for my $i ( 0 .. $#lines ) {
+        my $line = $lines[$i];
+
+        # With size:    "0010eefe 000001fe T aot_invoke_native"
+        # Without size: "001298fe T invokeNative"
+        next
+          unless $line =~ /^([0-9a-f]+)(?:\s+([0-9a-f]+))?\s+\S\s+(\S+)\s*$/i;
+        my ( $addr, $size, $name ) = ( $1, $2, $3 );
+        next unless $name eq $sym;
+
+        my $start = hex($addr);
+
+        if ( defined $size ) {
+            return ( $start, $start + hex($size) );
+        }
+
+        # Those don't have sizes, skip them.
+        # They appear inside invokeNative_ia32.s
+        my %ELF_INTERNAL_LABELS = (
+            stack_aligned  => 1,
+            skip_push_args => 1,
+        );
+
+        # No size -> take next symbol whose address is > $start.
+        for my $j ( $i + 1 .. $#lines ) {
+            next
+              unless $lines[$j] =~
+              /^([0-9a-f]+)(?:\s+[0-9a-f]+)?\s+\S\s+(\S+)\s*$/i;
+            my ( $next_addr, $next_name ) = ( hex($1), $2 );
+            next if $next_addr <= $start;
+            next if $ELF_INTERNAL_LABELS{$next_name};
+            return ( $start, $next_addr );
+        }
+
+        # No next symbol, don't know size
+        return ();
+    }
+
+    return ();
+}
+
 sub elf_read_sections {
     my ($elffile) = @_;