[core] Attribute placement new storage symbols to components (#15092)

esphome/analyze_memory/__init__.py

@@ -56,6 +56,10 @@ _COMPONENT_PREFIX_LIB = "[lib]"
 _COMPONENT_CORE = f"{_COMPONENT_PREFIX_ESPHOME}core"
 _COMPONENT_API = f"{_COMPONENT_PREFIX_ESPHOME}api"
 
+# Placement new storage suffix (generated by codegen Pvariable)
+_PSTORAGE_SUFFIX = "__pstorage"
+
+
 # C++ namespace prefixes
 _NAMESPACE_ESPHOME = "esphome::"
 _NAMESPACE_STD = "std::"
@@ -332,6 +336,13 @@ class MemoryAnalyzer:
         # Demangle C++ names if needed
         demangled = self._demangle_symbol(symbol_name)
 
+        # Check for placement new storage symbols (generated by codegen)
+        # Format: {component}__{id}__pstorage
+        if demangled.endswith(_PSTORAGE_SUFFIX) and (
+            component := self._match_pstorage_component(demangled)
+        ):
+            return component
+
         # Check for special component classes first (before namespace pattern)
         # This handles cases like esphome::ESPHomeOTAComponent which should map to ota
         if _NAMESPACE_ESPHOME in demangled:
@@ -399,6 +410,24 @@ class MemoryAnalyzer:
         # Track uncategorized symbols for analysis
         return "other"
 
+    def _match_pstorage_component(self, symbol_name: str) -> str | None:
+        """Match a __pstorage symbol to its ESPHome component.
+
+        Symbol format: {component}__{id}__pstorage
+        The component namespace is embedded by codegen before the double underscore.
+        """
+        prefix = symbol_name[: -len(_PSTORAGE_SUFFIX)]
+        # Extract component namespace before the first double underscore
+        dunder_pos = prefix.find("__")
+        if dunder_pos == -1:
+            return None
+        component_name = prefix[:dunder_pos]
+        if component_name in get_esphome_components():
+            return f"{_COMPONENT_PREFIX_ESPHOME}{component_name}"
+        if component_name in self.external_components:
+            return f"{_COMPONENT_PREFIX_EXTERNAL}{component_name}"
+        return None
+
     def _batch_demangle_symbols(self, symbols: list[str]) -> None:
         """Batch demangle C++ symbol names for efficiency."""
         if not symbols:

esphome/analyze_memory/cli.py

@@ -15,6 +15,7 @@ from . import (
     _COMPONENT_PREFIX_ESPHOME,
     _COMPONENT_PREFIX_EXTERNAL,
     _COMPONENT_PREFIX_LIB,
+    _PSTORAGE_SUFFIX,
     RAM_SECTIONS,
     MemoryAnalyzer,
 )
@@ -23,6 +24,17 @@ if TYPE_CHECKING:
     from . import ComponentMemory
 
 
+def _format_pstorage_name(name: str) -> str:
+    """Format a __pstorage symbol as 'storage for {id}'."""
+    if not name.endswith(_PSTORAGE_SUFFIX):
+        return name
+    prefix = name[: -len(_PSTORAGE_SUFFIX)]
+    # Strip component namespace prefix: {component}__{id} -> {id}
+    dunder_pos = prefix.find("__")
+    var_id = prefix[dunder_pos + 2 :] if dunder_pos != -1 else prefix
+    return f"storage for {var_id}"
+
+
 class MemoryAnalyzerCLI(MemoryAnalyzer):
     """Memory analyzer with CLI-specific report generation."""
 
@@ -148,11 +160,14 @@ class MemoryAnalyzerCLI(MemoryAnalyzer):
         If section is one of the RAM sections (.data or .bss), a label like
         " [data]" or " [bss]" is appended. For non-RAM sections or when
         section is None, no section label is added.
+
+        Placement new storage symbols are formatted as "storage for {id}".
         """
+        display_name = _format_pstorage_name(demangled)
         section_label = ""
         if section in RAM_SECTIONS:
             section_label = f" [{section[1:]}]"  # .data -> [data], .bss -> [bss]
-        return f"{demangled} ({size:,} B){section_label}"
+        return f"{display_name} ({size:,} B){section_label}"
 
     def _add_top_symbols(self, lines: list[str]) -> None:
         """Add a section showing the top largest symbols in the binary."""
@@ -175,11 +190,13 @@ class MemoryAnalyzerCLI(MemoryAnalyzer):
         for i, (_, demangled, size, section, component) in enumerate(top_symbols):
             # Format section label
             section_label = f"[{section[1:]}]" if section else ""
-            # Truncate demangled name if too long
+            # Format storage symbols readably
+            display_name = _format_pstorage_name(demangled)
+            # Truncate if too long
             demangled_display = (
-                f"{demangled[:truncate_limit]}..."
-                if len(demangled) > self.COL_TOP_SYMBOL_NAME
-                else demangled
+                f"{display_name[:truncate_limit]}..."
+                if len(display_name) > self.COL_TOP_SYMBOL_NAME
+                else display_name
             )
             lines.append(
                 f"{i + 1:>2}. {size:>7,} B {section_label:<8} {demangled_display:<{self.COL_TOP_SYMBOL_NAME}} {component}"
@@ -573,15 +590,16 @@ class MemoryAnalyzerCLI(MemoryAnalyzer):
                 lines.append(f"Total size: {comp_mem.flash_total:,} B")
                 lines.append("")
 
-                # Show all symbols above threshold for better visibility
+                # Show symbols above threshold, always include storage symbols
                 large_symbols = [
                     (sym, dem, size, sec)
                     for sym, dem, size, sec in sorted_symbols
                     if size > self.SYMBOL_SIZE_THRESHOLD
+                    or dem.endswith(_PSTORAGE_SUFFIX)
                 ]
 
                 lines.append(
-                    f"{comp_name} Symbols > {self.SYMBOL_SIZE_THRESHOLD} B ({len(large_symbols)} symbols):"
+                    f"{comp_name} Symbols > {self.SYMBOL_SIZE_THRESHOLD} B & storage ({len(large_symbols)} symbols):"
                 )
                 for i, (symbol, demangled, size, section) in enumerate(large_symbols):
                     lines.append(
@@ -604,7 +622,10 @@ class MemoryAnalyzerCLI(MemoryAnalyzer):
             # Sort by size descending
             sorted_ram_syms = sorted(ram_syms, key=lambda x: x[2], reverse=True)
             large_ram_syms = [
-                s for s in sorted_ram_syms if s[2] > self.RAM_SYMBOL_SIZE_THRESHOLD
+                s
+                for s in sorted_ram_syms
+                if s[2] > self.RAM_SYMBOL_SIZE_THRESHOLD
+                or s[1].endswith(_PSTORAGE_SUFFIX)
             ]
 
             lines.append(f"{name} ({mem.ram_total:,} B total RAM):")
@@ -622,13 +643,14 @@ class MemoryAnalyzerCLI(MemoryAnalyzer):
                 for symbol, demangled, size, section in large_ram_syms[:10]:
                     # Format section label consistently by stripping leading dot
                     section_label = section.lstrip(".") if section else ""
+                    display_name = _format_pstorage_name(demangled)
                     # Add ellipsis if name is truncated
-                    demangled_display = (
-                        f"{demangled[:70]}..." if len(demangled) > 70 else demangled
-                    )
-                    lines.append(
-                        f"    {size:>6,} B [{section_label}] {demangled_display}"
+                    display_name = (
+                        f"{display_name[:70]}..."
+                        if len(display_name) > 70
+                        else display_name
                     )
+                    lines.append(f"    {size:>6,} B [{section_label}] {display_name}")
                 if len(large_ram_syms) > 10:
                     lines.append(f"    ... and {len(large_ram_syms) - 10} more")
             lines.append("")

esphome/cpp_generator.py

@@ -584,7 +584,16 @@ def Pvariable(id_: ID, rhs: SafeExpType, type_: "MockObj" = None) -> "MockObj":
         # For 'new' allocations, use placement new into static storage
         # to avoid heap fragmentation on embedded devices.
         the_type = id_.type
-        storage_name = f"{id_.id}__pstorage"
+        # Extract component namespace from type for memory analysis attribution
+        type_str = str(the_type)
+        # Strip leading esphome:: to get the component namespace
+        # e.g. esphome::dsmr::Dsmr -> dsmr, logger::Logger -> logger
+        bare = type_str.removeprefix("esphome::")
+        if "::" in bare:
+            component_ns = bare.split("::", maxsplit=1)[0].rstrip("_")
+        else:
+            component_ns = "esphome"
+        storage_name = f"{component_ns}__{id_.id}__pstorage"
 
         # Declare aligned byte array for the object storage
         CORE.add_global(