Bug #12030
closedCrash calling Hash#dig when hash contains ActiveRecord objects
Description
This code results in a stack consistency error:
begin
require 'bundler/inline'
rescue LoadError => e
$stderr.puts 'Bundler version 1.10 or later is required. Please update your Bundler'
raise e
end
gemfile(true) do
ruby '2.3.0'
source 'https://rubygems.org'
gem 'activerecord', '4.2.5'
end
require 'active_record'
{ foo: ActiveRecord::Base }.dig(:foo, :bar)
Output¶
crash.rb:5: [BUG] Stack consistency error (sp: 3, bp: 4)
ruby 2.3.0p0 (2015-12-25 revision 53290) [x86_64-darwin13]
[snip]
-- Control frame information -----------------------------------------------
c:0002 p:0050 s:0003 E:001a80 EVAL crash.rb:5 [FINISH]
c:0001 p:0000 s:0002 E:000cd0 (none) [FINISH]
-- Ruby level backtrace information ----------------------------------------
crash.rb:5:in `<main>'
-- C level backtrace information -------------------------------------------
0 ruby 0x000000010a3aa1e4 rb_vm_bugreport + 388
1 ruby 0x000000010a248aa5 rb_bug + 485
2 ruby 0x000000010a38abc3 vm_exec_core + 13315
3 ruby 0x000000010a39b1f9 vm_exec + 121
4 ruby 0x000000010a251ef4 ruby_exec_internal + 148
5 ruby 0x000000010a251e1e ruby_run_node + 78
6 ruby 0x000000010a20820f main + 79
Expected output¶
crash.rb:5:in `dig': #<Class:ActiveRecord::Base> does not have #dig method (TypeError)
Notes¶
In addition to crashing when trying to dig into ActiveRecord::Base, it will crash trying to dig into a child class of ActiveRecord::Base, or into an instance of a child class of ActiveRecord::Base. Putting AR objects into hashes is probably a common practice in Rails apps, and users will dig too far into those hashes by accident (as I did!)
More realistic production code could look more like:
class User < ActiveRecord::Base; end
{ foo: User.first }.dig(:foo, :bar)
This bug has also been reproduced on: ruby 2.3.0p0 (2015-12-25 revision 53290) [x86_64-linux-gnu]
Updated by akhramov (Artem Khramov) almost 9 years ago
Hi, I was able to reproduce it on a bit minified example. Here it is:
class Foo
def self.respond_to?(*args)
super
end
end
{ foo: Foo }.dig(:foo, :foo)
Updated by nobu (Nobuyoshi Nakada) almost 9 years ago
- Status changed from Open to Closed
Applied in changeset r53695.
vm_eval.c: fix hook call
- vm_eval.c (rb_check_funcall_with_hook): also should call the
given hook before returning Qundef when overridden respond_to?
method returned false. [ruby-core:73556] [Bug #12030]
Updated by nagachika (Tomoyuki Chikanaga) almost 9 years ago
- Backport changed from 2.0.0: UNKNOWN, 2.1: UNKNOWN, 2.2: UNKNOWN, 2.3: UNKNOWN to 2.0.0: REQUIRED, 2.1: REQUIRED, 2.2: REQUIRED, 2.3: REQUIRED
r53695 should be backported into ruby_2_3 branch.
rb_check_funcall_with_hook() exists in ruby_2_0_0, ruby_2_1, ruby_2_2 branches and they have same potential bug, but rb_check_funcall_with_hook() is never called in these branches.
I fill REQUIRED
for all stable branches even though only 2.3 branch has real issue fixed by the change.
Updated by naruse (Yui NARUSE) over 8 years ago
- Backport changed from 2.0.0: REQUIRED, 2.1: REQUIRED, 2.2: REQUIRED, 2.3: REQUIRED to 2.0.0: REQUIRED, 2.1: REQUIRED, 2.2: REQUIRED, 2.3: DONE
ruby_2_3 r54402 merged revision(s) 53695.