annotate.py improvements

Annotate constants
- functions
- heap values

Collapse nop regions

src/codegen/irgen/irgenerator.cpp

@@ -834,13 +834,16 @@ private:
         return rtn;
     }
 
-    CompilerVariable* getNone() {
+    ConcreteCompilerVariable* getNone() {
         ConcreteCompilerVariable* v = new ConcreteCompilerVariable(
             typeFromClass(none_cls), embedConstantPtr(None, g.llvm_value_type_ptr), false);
         return v;
     }
 
     ConcreteCompilerVariable* _getGlobal(AST_Name* node, UnwindInfo unw_info) {
+        if (node->id.str() == "None")
+            return getNone();
+
         bool do_patchpoint = ENABLE_ICGETGLOBALS && (irstate->getEffortLevel() != EffortLevel::INTERPRETED);
         if (do_patchpoint) {
             ICSetupInfo* pp = createGetGlobalIC(getOpInfoForNode(node, unw_info).getTypeRecorder());

src/runtime/builtin_modules/builtins.cpp

@@ -861,11 +861,19 @@ Box* getreversed(Box* o) {
     return new (seqreviter_cls) BoxedSeqIter(o, len - 1);
 }
 
-Box* pydump(void* p) {
+Box* pydump(Box* p) {
     dump(p);
     return None;
 }
 
+Box* pydumpAddr(Box* p) {
+    if (p->cls != int_cls)
+        raiseExcHelper(TypeError, "Requires an int");
+
+    dump((void*)static_cast<BoxedInt*>(p)->n);
+    return None;
+}
+
 class BoxedEnvironmentError : public BoxedException {
 public:
     // Box* args, *message, *myerrno, *strerror, *filename;
@@ -1102,6 +1110,8 @@ void setupBuiltins() {
     builtins_module->giveAttr("trap", trap_obj);
     builtins_module->giveAttr("dump",
                               new BoxedBuiltinFunctionOrMethod(boxRTFunction((void*)pydump, UNKNOWN, 1), "dump"));
+    builtins_module->giveAttr(
+        "dumpAddr", new BoxedBuiltinFunctionOrMethod(boxRTFunction((void*)pydumpAddr, UNKNOWN, 1), "dumpAddr"));
 
     builtins_module->giveAttr(
         "getattr", new BoxedBuiltinFunctionOrMethod(boxRTFunction((void*)getattrFunc, UNKNOWN, 3, 1, false, false),

src/runtime/objmodel.cpp

@@ -2017,6 +2017,25 @@ extern "C" void dump(void* p) {
             printf("%ld elements\n", static_cast<BoxedList*>(b)->size);
         }
 
+        if (isSubclass(b->cls, module_cls)) {
+            printf("The '%s' module\n", static_cast<BoxedModule*>(b)->name().c_str());
+        }
+
+        /*
+        if (b->cls->instancesHaveHCAttrs()) {
+            HCAttrs* attrs = b->getHCAttrsPtr();
+            printf("Has %ld attrs\n", attrs->hcls->attr_offsets.size());
+            for (const auto& p : attrs->hcls->attr_offsets) {
+                printf("Index %d: %s: %p\n", p.second, p.first.c_str(), attrs->attr_list->attrs[p.second]);
+            }
+        }
+        */
+
+        return;
+    }
+
+    if (al->kind_id == gc::GCKind::HIDDEN_CLASS) {
+        printf("Hidden class object\n");
         return;
     }
 

tools/annotate.py

+import commands
+import os
+import re
 import subprocess
 import sys
 
@@ -13,6 +16,78 @@ def get_objdump(func):
 
     raise Exception("Couldn't find function %r to objdump" % func)
 
+_symbols = None
+def lookupAsSymbol(n):
+    global _symbols
+    if _symbols is None:
+        _symbols = {}
+        nm_output = commands.getoutput("nm pyston_release")
+        for l in nm_output.split('\n'):
+            addr = l[:16]
+            if addr.isalnum():
+                _symbols[int(addr, 16)] = l[19:]
+    sym = _symbols.get(n, None)
+    if not sym:
+        return sym
+
+    demangled = None
+    if sym.startswith('_') and os.path.exists("tools/demangle"):
+        demangled = commands.getoutput("tools/demangle %s" % sym)
+        if demangled != "Error: unable to demangle":
+            return demangled
+    return sym + "()"
+
+_heap_proc = None
+heapmap_args = None
+# heapmap_args = ["./pyston_release", "-i", "minibenchmarks/fasta.py"]
+def lookupAsHeapAddr(n):
+    global _heap_proc
+    if _heap_proc is None:
+        if heapmap_args is None:
+            return None
+
+        _heap_proc = subprocess.Popen(heapmap_args, stdout=subprocess.PIPE, stdin=subprocess.PIPE)
+
+        while True:
+            l = _heap_proc.stdout.readline()
+            if l.startswith("Pyston v0.2"):
+                break
+
+    _heap_proc.stdin.write("dumpAddr(%d)\nprint '!!!!'\n" % n)
+    lines = []
+    while True:
+        l = _heap_proc.stdout.readline()
+        if l == '!!!!\n':
+            break
+        lines.append(l)
+    s = ''.join(lines[1:-1])
+    if "Class: NoneType" in s:
+        return "None"
+    if "non-gc memory" in s:
+        return "(non-gc memory)"
+    if "Hidden class object" in s:
+        return "(hcls object)"
+    if "Class: type" in s:
+        m = re.search("Type name: ([^ \n]+)", s)
+        return "The '%s' class" % m.group(1)
+
+    if "Python object" in s:
+        m = re.search("Class: ([^ \n]+)", s)
+        return "A '%s' object" % m.group(1)
+
+    print s
+
+def lookupConstant(n):
+    sym = lookupAsSymbol(n)
+    if sym:
+        return "# " + sym
+
+    heap = lookupAsHeapAddr(n)
+    if heap:
+        return "# " + heap
+
+    return ""
+
 if __name__ == "__main__":
     # TODO: if it's not passed, maybe default to annotating the
     # first function in the profile (the one in which the plurality of
@@ -37,9 +112,28 @@ if __name__ == "__main__":
 
         counts[addr] = int(count)
 
+    nops = None # (count, num, start, end)
     for l in objdump.split('\n')[7:]:
         addr = l.split(':')[0]
         count = counts.pop(addr.strip(), 0)
-        print str(count).rjust(8), l
+
+        m = re.search("movabs \\$0x([0-9a-f]+),", l)
+        extra = ""
+        if m:
+            n = int(m.group(1), 16)
+
+            extra = lookupConstant(n)
+
+        if l.endswith("\tnop"):
+            addr = l.split()[0][:-1]
+            if not nops:
+                nops = (count, 1, addr, addr)
+            else:
+                nops = (nops[0] + count, nops[1] + 1, nops[2], addr)
+        else:
+            if nops:
+                print str(nops[0]).rjust(8), ("    %s-%s              nop*%d" % (nops[2], nops[3], nops[1])).ljust(70)
+                nops = None
+            print str(count).rjust(8), l.ljust(70), extra
 
     assert not counts, counts