Bug 1483121 - Generate static atom hash in StaticAtoms.py. r=njn,emilio

Summary: Depends On D3286

Reviewers: njn!, emilio!

Tags: #secure-revision

Bug #: 1483121

Differential Revision: https://phabricator.services.mozilla.com/D3295

Author: heycam

servo/components/style/gecko/regen_atoms.py

@@ -24,7 +24,8 @@
 """[1:]   # NOQA: E501
 
 
-PATTERN = re.compile('^GK_ATOM\(([^,]*),[^"]*"([^"]*)",\s*([^,]*),\s*([^)]*)\)',
+# Matches lines like `GK_ATOM(foo, "foo", 0x12345678, nsStaticAtom, PseudoElementAtom)`.
+PATTERN = re.compile('^GK_ATOM\(([^,]*),[^"]*"([^"]*)",\s*(0x[0-9a-f]+),\s*([^,]*),\s*([^)]*)\)',
                      re.MULTILINE)
 FILE = "include/nsGkAtomList.h"
 CLASS = "nsGkAtoms"
@@ -52,10 +53,11 @@ def map_atom(ident):
 
 
 class Atom:
-    def __init__(self, ident, value, ty, atom_type):
+    def __init__(self, ident, value, hash, ty, atom_type):
         self.ident = "{}_{}".format(CLASS, ident)
         self.original_ident = ident
         self.value = value
+        self.hash = hash
         # The Gecko type: "nsStaticAtom", "nsICSSPseudoElement", or "nsIAnonBoxPseudo"
         self.ty = ty
         # The type of atom: "Atom", "PseudoElement", "NonInheritingAnonBox",
@@ -104,7 +106,8 @@ def collect_atoms(objdir):
     with open(path) as f:
         content = f.read()
         for result in PATTERN.finditer(content):
-            atoms.append(Atom(result.group(1), result.group(2), result.group(3), result.group(4)))
+            atoms.append(Atom(result.group(1), result.group(2), result.group(3),
+                              result.group(4), result.group(5)))
     return atoms
 
 

xpcom/ds/Atom.py

@@ -9,6 +9,7 @@ def __init__(self, ident, string, ty="nsStaticAtom"):
         self.string = string
         self.ty = ty
         self.atom_type = self.__class__.__name__
+        self.hash = hash_string(string)
 
 
 class PseudoElementAtom(Atom):
@@ -29,3 +30,25 @@ def __init__(self, ident, string):
 class InheritingAnonBoxAtom(AnonBoxAtom):
     def __init__(self, ident, string):
         AnonBoxAtom.__init__(self, ident, string)
+
+
+GOLDEN_RATIO_U32 = 0x9E3779B9
+
+
+def rotate_left_5(value):
+    return ((value << 5) | (value >> 27)) & 0xFFFFFFFF
+
+
+def wrapping_multiply(x, y):
+    return (x * y) & 0xFFFFFFFF
+
+
+# Calculate the precomputed hash of the static atom. This is a port of
+# mozilla::HashString(const char16_t*), which is what we use for atomizing
+# strings. An assertion in nsAtomTable::RegisterStaticAtoms ensures that
+# the value we compute here matches what HashString() would produce.
+def hash_string(s):
+    h = 0
+    for c in s:
+        h = wrapping_multiply(GOLDEN_RATIO_U32, rotate_left_5(h) ^ ord(c))
+    return h

xpcom/ds/StaticAtoms.py

@@ -2351,9 +2351,9 @@ def verify():
 def generate_nsgkatomlist_h(output, *ignore):
     verify()
     output.write("/* THIS FILE IS AUTOGENERATED BY StaticAtoms.py.  DO NOT EDIT */\n\n"
-                 "// GK_ATOM(identifier, string, gecko_type, atom_type)\n" +
-                 "".join(["GK_ATOM(%s, \"%s\", %s, %s)\n" %
-                            (a.ident, a.string, a.ty, a.atom_type)
+                 "// GK_ATOM(identifier, string, hash, gecko_type, atom_type)\n" +
+                 "".join(["GK_ATOM(%s, \"%s\", 0x%08x, %s, %s)\n" %
+                            (a.ident, a.string, a.hash, a.ty, a.atom_type)
                           for a in STATIC_ATOMS]))
 
 

xpcom/ds/nsAtom.h

@@ -9,7 +9,6 @@
 
 #include "nsISupportsImpl.h"
 #include "nsString.h"
-#include "mozilla/HashFunctions.h"
 #include "mozilla/UniquePtr.h"
 
 namespace mozilla {
@@ -104,10 +103,10 @@ class nsAtom
 
 protected:
   // Used by nsStaticAtom.
-  constexpr nsAtom(const char16_t* aStr, uint32_t aLength)
+  constexpr nsAtom(const char16_t* aStr, uint32_t aLength, uint32_t aHash)
     : mLength(aLength)
     , mKind(static_cast<uint32_t>(nsAtom::AtomKind::Static))
-    , mHash(mozilla::HashString(aStr))
+    , mHash(aHash)
   {}
 
   // Used by nsDynamicAtom.
@@ -138,9 +137,14 @@ class nsStaticAtom : public nsAtom
   MozExternalRefCountType AddRef() = delete;
   MozExternalRefCountType Release() = delete;
 
+  // The static atom's precomputed hash value is an argument here, but it
+  // must be the same as would be computed by mozilla::HashString(aStr),
+  // which is what we use when atomizing strings. We compute this hash in
+  // Atom.py and assert in nsAtomTable::RegisterStaticAtoms that the two
+  // hashes match.
   constexpr nsStaticAtom(const char16_t* aStr, uint32_t aLength,
-                         uint32_t aStringOffset)
-    : nsAtom(aStr, aLength)
+                         uint32_t aHash, uint32_t aStringOffset)
+    : nsAtom(aStr, aLength, aHash)
     , mStringOffset(aStringOffset)
   {}
 

xpcom/ds/nsAtomTable.cpp

@@ -647,6 +647,11 @@ nsAtomTable::RegisterStaticAtoms(const nsStaticAtom* aAtoms, size_t aAtomsLen)
     MOZ_ASSERT(nsCRT::IsAscii(atom->String()));
     MOZ_ASSERT(NS_strlen(atom->String()) == atom->GetLength());
 
+    // This assertion ensures the static atom's precomputed hash value matches
+    // what would be computed by mozilla::HashString(aStr), which is what we use
+    // when atomizing strings. We compute this hash in Atom.py.
+    MOZ_ASSERT(HashString(atom->String()) == atom->hash());
+
     AtomTableKey key(atom);
     nsAtomSubTable& table = SelectSubTable(key);
     MutexAutoLock lock(table.mLock);

xpcom/ds/nsGkAtoms.cpp

@@ -13,10 +13,9 @@ NS_RegisterStaticAtoms(const nsStaticAtom* aAtoms, size_t aAtomsLen);
 namespace mozilla {
 namespace detail {
 
-MOZ_PUSH_DISABLE_INTEGRAL_CONSTANT_OVERFLOW_WARNING
 extern constexpr GkAtoms gGkAtoms = {
   // The initialization of each atom's string.
-  #define GK_ATOM(name_, value_, type_, atom_type_) \
+  #define GK_ATOM(name_, value_, hash_, type_, atom_type_) \
     u"" value_,
   #include "nsGkAtomList.h"
   #undef GK_ATOM
@@ -26,25 +25,25 @@ extern constexpr GkAtoms gGkAtoms = {
     // Note that |value_| is an 8-bit string, and so |sizeof(value_)| is equal
     // to the number of chars (including the terminating '\0'). The |u""| prefix
     // converts |value_| to a 16-bit string.
-    #define GK_ATOM(name_, value_, type_, atom_type_)                     \
+    #define GK_ATOM(name_, value_, hash_, type_, atom_type_)              \
       nsStaticAtom(u"" value_,                                            \
           sizeof(value_) - 1,                                             \
+          hash_,                                                          \
           offsetof(GkAtoms,                                               \
                    mAtoms[static_cast<size_t>(GkAtoms::Atoms::name_)]) -  \
           offsetof(GkAtoms, name_##_string)),
     #include "nsGkAtomList.h"
     #undef GK_ATOM
   }
 };
-MOZ_POP_DISABLE_INTEGRAL_CONSTANT_OVERFLOW_WARNING
 
 } // namespace detail
 } // namespace mozilla
 
 const nsStaticAtom* const nsGkAtoms::sAtoms = mozilla::detail::gGkAtoms.mAtoms;
 
 // Definition of the pointer to the static atom.
-#define GK_ATOM(name_, value_, type_, atom_type_)                          \
+#define GK_ATOM(name_, value_, hash_, type_, atom_type_)                   \
   type_* nsGkAtoms::name_ = const_cast<type_*>(static_cast<const type_*>(  \
     &mozilla::detail::gGkAtoms.mAtoms[                                     \
       static_cast<size_t>(mozilla::detail::GkAtoms::Atoms::name_)]));

xpcom/ds/nsGkAtoms.h

@@ -48,9 +48,9 @@
 // nsGkAtomList.h is generated by StaticAtoms.py and has entries that look
 // like this:
 //
-//   GK_ATOM(one, "one", nsStaticAtom, Atom)
-//   GK_ATOM(two, "two", nsICSSPseudoElement, PseudoElementAtom)
-//   GK_ATOM(three, "three", nsICSSAnonBoxPseudo, InheritingAnonBoxAtom)
+//   GK_ATOM(one, "one", 0x01234567, nsStaticAtom, Atom)
+//   GK_ATOM(two, "two", 0x12345678, nsICSSPseudoElement, PseudoElementAtom)
+//   GK_ATOM(three, "three", 0x23456789, nsICSSAnonBoxPseudo, InheritingAnonBoxAtom)
 //
 // After macro expansion, the atom definitions look like the following:
 //
@@ -120,14 +120,17 @@
 //       // The initialization of the atoms themselves.
 //       nsStaticAtom(
 //         u"one", 3,
+//         0x01234567,
 //         offsetof(GkAtoms, mAtoms[static_cast<size_t>(GkAtoms::Atoms::one)]) -
 //         offsetof(GkAtoms, one_string)),
 //       nsStaticAtom(
 //         u"two", 3,
+//         0x12345678,
 //         offsetof(GkAtoms, mAtoms[static_cast<size_t>(GkAtoms::Atoms::two)]) -
 //         offsetof(GkAtoms, two_string)),
 //       nsStaticAtom(
 //         u"three", 3,
+//         0x23456789,
 //         offsetof(GkAtoms, mAtoms[static_cast<size_t>(GkAtoms::Atoms::three)]) -
 //         offsetof(GkAtoms, three_string)),
 //     }
@@ -156,12 +159,13 @@
 
 // Trivial subclasses of nsStaticAtom so that function signatures can require
 // an atom from a specific atom list.
-#define DEFINE_STATIC_ATOM_SUBCLASS(name_)                                    \
-  class name_ : public nsStaticAtom                                           \
-  {                                                                           \
-  public:                                                                     \
-    constexpr name_(const char16_t* aStr, uint32_t aLength, uint32_t aOffset) \
-      : nsStaticAtom(aStr, aLength, aOffset) {}                               \
+#define DEFINE_STATIC_ATOM_SUBCLASS(name_)                   \
+  class name_ : public nsStaticAtom                          \
+  {                                                          \
+  public:                                                    \
+    constexpr name_(const char16_t* aStr, uint32_t aLength,  \
+                    uint32_t aHash, uint32_t aOffset)        \
+      : nsStaticAtom(aStr, aLength, aHash, aOffset) {}       \
   };
 
 DEFINE_STATIC_ATOM_SUBCLASS(nsICSSAnonBoxPseudo)
@@ -181,14 +185,14 @@ namespace detail {
 struct GkAtoms
 {
   // The declaration of each atom's string.
-  #define GK_ATOM(name_, value_, type_, atom_type_) \
+  #define GK_ATOM(name_, value_, hash_, type_, atom_type_) \
     const char16_t name_##_string[sizeof(value_)];
   #include "nsGkAtomList.h"
   #undef GK_ATOM
 
   // The enum value for each atom.
   enum class Atoms {
-    #define GK_ATOM(name_, value_, type_, atom_type_) \
+    #define GK_ATOM(name_, value_, hash_, type_, atom_type_) \
       name_,
     #include "nsGkAtomList.h"
     #undef GK_ATOM
@@ -221,7 +225,7 @@ class nsGkAtoms
   //
   // XXX: Eventually this should be combined with its definition and the
   // pointer should be made `constexpr`. See bug 1449787.
-  #define GK_ATOM(name_, value_, type_, atom_type_) \
+  #define GK_ATOM(name_, value_, hash_, type_, atom_type_) \
     static type_* name_;
   #include "nsGkAtomList.h"
   #undef GK_ATOM