887645 - Compilation Failed With js/src/jsinterp.cpp:3684:9: error: indirect goto might cross protected scopes

Status: RESOLVED WONTFIX

(Core :: JavaScript Engine, defect)

Description

[David E. Weekly]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/27.0.1453.116 Safari/537.36

Steps to reproduce:

Download SpiderMonkey 17. Unpack.
cd js/src
./configure
make



Actual results:

The following compilation error:

/usr/local/bin/clang++ -o jsinterp.o -c  -fvisibility=hidden -DENABLE_YARR_JIT=1 -DIMPL_MFBT -DEXPORT_JS_API -DNO_NSPR_10_SUPPORT -DUSE_ZLIB -I./../../mfbt/double-conversion -I. -I. -I./dist/include        -I. -I./assembler -I./yarr  -fPIC -Qunused-arguments  -Qunused-arguments -Wall -Wpointer-arith -Woverloaded-virtual -Werror=return-type -Wtype-limits -Wempty-body -Wno-ctor-dtor-privacy -Wno-overlength-strings -Wno-invalid-offsetof -Wno-variadic-macros -Wno-c++0x-extensions -Wno-extended-offsetof -Wno-unknown-warning-option -Wno-return-type-c-linkage -Wno-mismatched-tags -fno-common -fno-rtti -ffunction-sections -fdata-sections -fno-exceptions -pthread -pipe  -DNDEBUG -DTRIMMED -g -O3 -fno-stack-protector -fomit-frame-pointer -DUSE_SYSTEM_MALLOC=1 -DENABLE_ASSEMBLER=1 -DENABLE_JIT=1  -Qunused-arguments  -DMOZILLA_CLIENT -include ./js-confdefs.h -MD -MF .deps/jsinterp.o.pp /Users/dew/mozjs17.0.0/js/src/jsinterp.cpp
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:3684:9: error: indirect goto might
      cross protected scopes
        DO_NEXT_OP(len);
        ^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:1091:33: note: expanded from macro
      'DO_NEXT_OP'
                                DO_OP();                                      \
                                ^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:1085:47: note: expanded from macro
      'DO_OP'
                                JS_EXTENSION_(goto *jumpTable[op]);           \
                                              ^
./jstypes.h:279:44: note: expanded from macro 'JS_EXTENSION_'
# define JS_EXTENSION_(s) __extension__ ({ s; })
                                           ^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:1420:1: note: possible target of
      indirect goto
ADD_EMPTY_CASE(JSOP_NOP)
^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:1097:29: note: expanded from macro
      'ADD_EMPTY_CASE'
# define ADD_EMPTY_CASE(OP) BEGIN_CASE(OP)                                    \
                            ^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:1094:29: note: expanded from macro
      'BEGIN_CASE'
# define BEGIN_CASE(OP)     L_##OP:
                            ^
<scratch space>:118:1: note: expanded from here
L_JSOP_NOP
^
/Users/dew/mozjs17.0.0/js/src/jsinterp.cpp:3667:25: note: jump exits scope of
      variable with non-trivial destructor
    DebugOnly<uint32_t> blockDepth = regs.fp()->blockChain().stackDepth();
                        ^
1 error generated.
make[1]: *** [jsinterp.o] Error 1
make: *** [default] Error 2



Expected results:

Compilation should have succeeded. This is unmodified SpiderMonkey 17, on OS X 10.8.4 with trunk clang 3.4.

/usr/local/bin/clang++ --version
clang version 3.4 (trunk 184528)
Target: x86_64-apple-darwin12.4.0
Thread model: posix

Comment 1

[Jeff Walden [:Waldo]]

I'm pretty sure you can enclose that DebugOnly line in #ifdef DEBUG / ... / #endif to fix this.

Given when we were working on 17's code, it's likely that clang wasn't quite on people's radar yet (it is now, and I remember this issue being fixed at some point), so nobody noticed clang was broken.  At this point, for an issue that's very simply patched, I don't think we'd spin a "17.0.1" point-release to fix it.  So I suggest making the easy fix locally, and wait for the 24 release for things to just work without changes.

Comment 2

[David E. Weekly]

No problem,  good to know this is a known issue and resolved in 24. Perhaps consider putting this in the FAQ / Known Issues section?

Comment 3

[Jeff Walden [:Waldo]]

<https://developer.mozilla.org/en-US/docs/SpiderMonkey/17> is a wiki, so feel free to edit.  :-)  The whole thing is very much skeletal, tho, because we'd gotten somewhat out of the habit of doing release-notes-y things, and it seemed better to do a release with bad docs than no release with "good" docs.  So adding it there or not probably won't make much difference.

We're trying to do a little better with 24 notes at <https://developer.mozilla.org/en-US/docs/SpiderMonkey/24>, but we started focusing on documenting substantial API changes as they happen a bit into the cycle, so even that won't be the greatest.  But you gotta start somewhere.  :-)  31 is hopefully when we'll really start to get things right, as we note changes incrementally as we go.