[PATCH] Temporary fix for PR bootstrap/44146

[IainS]

On 15 May 2010, at 22:20, Jan Hubicka wrote:

>> Currently the bootstrap on *-apple-darwin* is failing periodically
>> with changes in svn since the introduction of...
>>
>> r159371 | hubicka | 2010-05-13 13:47:48 -0400 (Thu, 13 May 2010) |  
>> 3 lines
>>
>>        * varpool.c (decide_is_variable_needed): Drop code checking
>>        TREE_SYMBOL_REFERENCED.
>>
>> Until the exact cause of these bootstrap failures is found and
>> solved, darwin should revert to the previous behavior to allow
>> for general testing of unpatched gcc trunk. Regression tested at
>> r159434 on x86_64-apple-darwin10. Okay for gcc trunk?
>
> Can I have at least a simple testcase showing the problem?

I'm working on looking at it - it is possible that the size of  
relocation entries is changing for some reason.
  (bootstrapping on x86_64 at the moment).
I'll post whatever comes out of this on the PR thread.
Iain