On win32, NS_InvokeByIndex is implemented with inline assembly. This
inline assembly assumes that it is wrapped by the compiler with the
standard x86 prologue and epilogue:
push ebp
mov ebp, esp
[inline assembly that manipulates the stack pointer]
pop ebp
ret
In particular, the last instruction of the inline assembly is:
mov esp, ebp
which cancels out the effects of the stack manipulation performed by all
the inline assembly that proceeds the instruction.
When compiling with clang-cl, however, the above assumption does not
hold, as clang-cl inserts a more complex prologue and epilogue,
something like:
push ebp
mov ebp, esp
sub esp, frame_size
[save registers into stack frame]
[inline assembly that manipulates the stack pointer]
[restore registers from stack frame]
add esp, frame_size
mov esp, ebp
pop ebp
ret
Combining this more extensive prologue and epilogue with the assumptions
of the inline assembly leads to interesting crashes when
NS_InvokeByIndex is called: the inline assembly effectively deallocates
the stack allocated by the inline assembly *and* the stack frame
allocated by the compiler itself. The compiler-generated code then
attemptes to deallocate the stack frame, leading to the crash, as the
code now returns to an unspecified address.
To avoid these sorts of problems in clang-cl and make the code more
robust generally, let's move the NS_InvokeByIndex implementation to a
separate assembly file. We can then write exactly what we need to have
happen, safe from any manipulations of the compiler.
Since we don't compile much (any?) code in Gecko with MASM, we need to
add the /SAFESEH flag to the assembler invocation so that the object
file with be appropriately marked as not containing exception handlers;
the linker (which is invoked with the /SAFESEH flag itself) will then
consent to link it into libxul.
The bulk of this commit was generated with a script, executed at the top
level of a typical source code checkout. The only non-machine-generated
part was modifying MFBT's moz.build to reflect the new naming.
CLOSED TREE makes big refactorings like this a piece of cake.
# The main substitution.
find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
xargs perl -p -i -e '
s/nsRefPtr\.h/RefPtr\.h/g; # handle includes
s/nsRefPtr ?</RefPtr</g; # handle declarations and variables
'
# Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h.
perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h
# Handle nsRefPtr.h itself, a couple places that define constructors
# from nsRefPtr, and code generators specially. We do this here, rather
# than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename
# things like nsRefPtrHashtable.
perl -p -i -e 's/nsRefPtr/RefPtr/g' \
mfbt/nsRefPtr.h \
xpcom/glue/nsCOMPtr.h \
xpcom/base/OwningNonNull.h \
ipc/ipdl/ipdl/lower.py \
ipc/ipdl/ipdl/builtin.py \
dom/bindings/Codegen.py \
python/lldbutils/lldbutils/utils.py
# In our indiscriminate substitution above, we renamed
# nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up.
find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \
xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g'
if [ -d .git ]; then
git mv mfbt/nsRefPtr.h mfbt/RefPtr.h
else
hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h
fi
xptcstubs_arm mostly works on iOS but Apple's assembler is ridiculous so
the inline assembly for the SharedStub and the stub methods needs judicious
preprocessor use.
The patch removes 455 occurrences of FAIL_ON_WARNINGS from moz.build files, and
adds 78 instances of ALLOW_COMPILER_WARNINGS. About half of those 78 are in
code we control and which should be removable with a little effort.
After this change, we have ShallowSizeOf{In,Ex}cludingThis(), which don't do
anything to measure children. (They can be combined with iteration to measure
children.)
The bulk of this commit was generated by running:
run-clang-tidy.py \
-checks='-*,llvm-namespace-comment' \
-header-filter=^/.../mozilla-central/.* \
-fix
Adding isMainProcessScriptable() into the middle of nsIInterfaceInfo
caused problems with some binary addons that relied on the ordering of
the methods in nsIInterfaceInfo. In an attempt to placate those addons,
move isMainProcessScriptable() to the end of the vtable. This change is
a no-op for normal libxul usage.
Constructing kComponentsInterfaceShimMap required a static constructor
on some compilers, due to a non-constexpr constructor and the necessity
of copying non-constexpr things like nsIID. This static constructor is
large (several kilobytes of object code on x86-64) and completely
unnecessary.
To fix this, let's add a constexpr (well, MOZ_CONSTEXPR) constructor to
ComponentsInterfaceShimEntry. This change alone doesn't completely
solve our problem, because the nsIID member still needs to be copied.
But doing that copying is silly: we only use the IID for constructing a
ShimInterfaceInfo in ShimInterfaceInfo::MaybeConstruct, and the
ShimInterfaceInfo constructor takes a const reference. So let's store a
const reference in ComponentsInterfaceShimEntry, too, and make that
structure significantly smaller in the process.
As explained in bug 1111355, having avx enabled appears to change the
alignment behavior of alloca (apparently adding an extra 16 bytes) of
padding/alignment (and using 32-byte alignment instead of 16-byte). The
suggestion of using __bultin_alloca_with_align in bug 1111355 didn't fix
the problem, so this seems to be the best available workaround, given
that this code, which should perhaps better be written in assembly, is
written in C++.
Interestingly, this is NOT fixed by #pragma GCC target ("arch=x86-64").
(I determined the (undocumented) name for the default -march value on
x86_64 from the gcc source code (gcc/config/i386/i386.c, function
ix86_option_override_internal, code that sets opts->x_ix86_arch_string .)
I confirmed that this sets the same macros based on the empty diff
between the output of 'gcc -E -dM -x c++ /dev/null' and 'gcc -E -dM -x
c++ -march=x86-64 /dev/null', which was not an empty diff for other
-march values (e.g., k8).)
I confirmed that the push_options and pop_options actually work by
putting the push/pop pair around a different (earlier) function, and
testing that this did not fix the bug (with the pop_options before
NS_InvokeByIndex).
See the gcc documentation at:
https://gcc.gnu.org/onlinedocs/gcc-4.9.0/gcc/Function-Specific-Option-Pragmas.htmlhttps://gcc.gnu.org/onlinedocs/gcc-4.9.0/gcc/Function-Attributes.htmlhttps://gcc.gnu.org/onlinedocs/gcc-4.9.0/gcc/i386-and-x86-64-Options.html
