// Repros also with ignition_turbo:
x = {};
x = [];
function __f_4() {
  x = {};
}
%OptimizeFunctionOnNextCall(__f_4);
__f_4();
__f_4();
Object.defineProperty(this, "x", { writable:false, value:5 });
__f_4();
print(x);

// Output:
# Compared x64,ignition with x64,ignition_turbo
#
# Flags of x64,ignition:
--abort_on_stack_overflow --expose-gc --allow-natives-syntax --invoke-weak-callbacks --omit-quit --es-staging --random-seed 1234 --ignition --turbo-filter=~ --hydrogen-filter=~ --novalidate-asm --nocrankshaft
# Flags of x64,ignition_turbo:
--abort_on_stack_overflow --expose-gc --allow-natives-syntax --invoke-weak-callbacks --omit-quit --es-staging --random-seed 1234 --ignition-staging --turbo --novalidate-asm
#
# Difference:
- [object Object]
+ 5
#
### Start of configuration x64,ignition:
[object Object]

### End of configuration x64,ignition
#
### Start of configuration x64,ignition_turbo:
5

### End of configuration x64,ignition_turbo

Friendly ping...

It looks like TurboFan is correct here and the correct result is 5. Maybe something wrong with the LOAD_GLOBAL_IC in Ignition?

FYI, these two CLs are the initial work for inlining LoadGlobalIC in the bytecode handler. Maybe worth checking if they're the cause:

63096bc89d674d6c97bdecdbdfbf98871d1825e0
f46f341303ac4efad5cbcc771d10642aa2c826de

Reduced repro:

function set_x() { x = {}; }

set_x();
set_x();   // Needs to be called twice to trigger bug.
print(x);  // {}
Object.defineProperty(this, "x", { writable:false, value:5 });
print(x);  // 5
set_x();
print(x);  // {}, should be 5

Looks to me like the problem is that set_x is allowed to set x despite it being non-writable.

The following revision refers to this bug:
  https://chromium.googlesource.com/v8/v8.git/+/b30ea1646a487ed818093afa9cf56aa2f5c222b3

commit b30ea1646a487ed818093afa9cf56aa2f5c222b3
Author: Igor Sheludko <ishell@chromium.org>
Date: Thu Jul 20 13:39:23 2017

[ic] Properly handle reconfiguring of a global property to 'readonly'.

Bug:  chromium:722783 
Change-Id: Id4612f3d45fb26daca8b4ef2efb0f7bc9ac39ed3
Reviewed-on: https://chromium-review.googlesource.com/579268
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46795}
[modify] https://crrev.com/b30ea1646a487ed818093afa9cf56aa2f5c222b3/src/objects.cc
[add] https://crrev.com/b30ea1646a487ed818093afa9cf56aa2f5c222b3/test/mjsunit/regress/regress-crbug-722783.js

ClusterFuzz has detected this issue as fixed in range 46794:46795.

Detailed report: https://clusterfuzz.com/testcase?key=6506088873852928

Fuzzer: foozzie_js_mutation
Job Type: v8_foozzie
Platform Id: linux

Crash Type: V8 correctness failure
Crash Address: 
Crash State:
  configs: x64,ignition:x64,ignition_turbo_opt
  sources: 38b
  
Sanitizer: address (ASAN)

Fixed: V8: 46794:46795

Reproducer Testcase: https://clusterfuzz.com/download?testcase_id=6506088873852928


See https://github.com/google/clusterfuzz-tools for more information.

If you suspect that the result above is incorrect, try re-doing that job on the test case report page.