1291293 - [wasm] Assertion failure: has(reg), at js/src/jit/RegisterSets.h:742

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Christian Holler (:decoder)]

The attached binary WebAssembly testcase crashes on mozilla-inbound revision c0c16cea7afa+ (build with --enable-posix-nspr-emulation --enable-valgrind --enable-gczeal --disable-tests --enable-address-sanitizer --disable-jemalloc --enable-optimize=-O2 --without-intl-api --enable-debug --target=i686-pc-linux-gnu). To reproduce, you can run the following code in the JS shell (running with --wasm-always-baseline might be necessary):

var data = os.file.readFile(file, 'binary');
Wasm.instantiateModule(new Uint8Array(data.buffer));



Backtrace:

==20968==ERROR: AddressSanitizer: SEGV on unknown address 0x00000000 (pc 0x09a2c18a bp 0xfff26178 sp 0xfff25e40 T0)
    #0 0x9a2c189 in js::jit::SpecializedRegSet<js::jit::AllocatableSetAccessors<js::jit::TypedRegisterSet<js::jit::Register> >, js::jit::TypedRegisterSet<js::jit::Register> >::take(js::jit::Register) js/src/jit/RegisterSets.h:742:9
    #1 0x9a2c189 in js::jit::SpecializedRegSet<js::jit::AllocatableSetAccessors<js::jit::TypedRegisterSet<js::jit::Register> >, js::jit::TypedRegisterSet<js::jit::Register> >::takeAny() js/src/jit/RegisterSets.h:768
    #2 0x9a2c189 in js::jit::CodeGeneratorX86::visitDivOrModI64(js::jit::LDivOrModI64*) js/src/jit/x86/CodeGenerator-x86.cpp:1431
    #3 0x9b51c98 in js::jit::LDivOrModI64::accept(js::jit::LElementVisitor*) js/src/jit/x86/LIR-x86.h:115:5
    #4 0x8aa36b6 in js::jit::CodeGenerator::generateBody() js/src/jit/CodeGenerator.cpp:5139:13
    #5 0x8b40604 in js::jit::CodeGenerator::generateWasm(js::wasm::SigIdDesc, js::wasm::FuncOffsets*) js/src/jit/CodeGenerator.cpp:8971:10
    #6 0xb6bac46 in js::wasm::IonCompileFunction(js::wasm::IonCompileTask*) js/src/asmjs/WasmIonCompile.cpp:3484:14
    #7 0xb6e8aa8 in js::wasm::CompileFunction(js::wasm::IonCompileTask*) js/src/asmjs/WasmIonCompile.cpp:3499:16
    #8 0xb653d98 in js::wasm::ModuleGenerator::finishFuncDef(unsigned int, js::wasm::FunctionGenerator*) js/src/asmjs/WasmGenerator.cpp:864:14
    #9 0xb5d7d7f in DecodeFunctionBody(js::wasm::Decoder&, js::wasm::ModuleGenerator&, unsigned int) js/src/asmjs/WasmCompile.cpp:1290:12
    #10 0xb5d7d7f in DecodeCodeSection(js::wasm::Decoder&, js::wasm::ModuleGenerator&) js/src/asmjs/WasmCompile.cpp:1350
    #11 0xb5d7d7f in js::wasm::Compile(js::wasm::ShareableBytes const&, js::wasm::CompileArgs&&, mozilla::UniquePtr<char [], JS::FreePolicy>*) js/src/asmjs/WasmCompile.cpp:1631
    #12 0x82e5d48 in js::wasm::Eval(JSContext*, JS::Handle<js::TypedArrayObject*>, JS::Handle<JSObject*>, JS::MutableHandle<js::WasmInstanceObject*>) js/src/asmjs/WasmJS.cpp:238:27
    #13 0x822079c in WasmLoop(JSContext*, unsigned int, JS::Value*) js/src/shell/js.cpp:5270:14
    #14 0xa5501bd in js::CallJSNative(JSContext*, bool (*)(JSContext*, unsigned int, JS::Value*), JS::CallArgs const&) js/src/jscntxtinlines.h:232:15
[...]
    #28 0x80ae5d6 in _start (/home/ubuntu/build/build/js+0x80ae5d6)

AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV js/src/jit/RegisterSets.h:742:9 in js::jit::SpecializedRegSet<js::jit::AllocatableSetAccessors<js::jit::TypedRegisterSet<js::jit::Register> >, js::jit::TypedRegisterSet<js::jit::Register> >::take(js::jit::Register)
==20968==ABORTING

Comment 1

[Christian Holler (:decoder)]

Attachment: Testcase

Comment 2

[Lars T Hansen [:lth]]

Note, x86 is currently disabled, see bug 1290453.

Comment 3

[Luke Wagner [:luke]]

From the callstack, this seems to be Ion compilation of i64.

Comment 4

[Lars T Hansen [:lth]]

Ah, indeed it is so.

This had better be reproducible without --wasm-always-baseline then.

Comment 5

[Benjamin Bouvier [:bbouvier] (inactive)]

Reproduced on x86 with --no-threads (on the regular ion backend).

Comment 6

[Benjamin Bouvier [:bbouvier] (inactive)]

Attachment: b.js

JS test case, to be run with --no-threads.

Comment 7

[Benjamin Bouvier [:bbouvier] (inactive)]

Attachment: fix.patch

Actually a simple fix.

Comment 8

[Hannes Verschore [:h4writer]]

Comment on attachment 8777507 [details] [diff] [review]
fix.patch

Review of attachment 8777507 [details] [diff] [review]:
-----------------------------------------------------------------

Good find, thanks for stealing.
Don't we need to do that on ARM too?

Comment 9

[Benjamin Bouvier [:bbouvier] (inactive)]

(In reply to Hannes Verschore [:h4writer] from comment #8)
> Comment on attachment 8777507 [details] [diff] [review]
> fix.patch
> 
> Review of attachment 8777507 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Good find, thanks for stealing.
> Don't we need to do that on ARM too?

Good catch, thanks.

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/498fd4aae629
Don't take the same register twice in i64 mod/umod on x86/ARM; r=h4writer

Comment 11

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/498fd4aae629