This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][GOLD] Avoid linker crashing when merge sections have uneven sizes.


On 04/08/2010 05:02 PM, Doug Kwan (Ãö®¶¼w) wrote:
> Yes, it looks fishy but I did not dig into the gcc sources.  The
> sections are .rodata sections for UTF8 strings for various sizes.  It
> looks like the entries are aligned to 32-bit boundaries, not to the
> entsize of a section.
> 

FWIW:  http://gcc.gnu.org/ml/java-patches/2009-q1/msg00044.html

David Daney

> -Doug
> 
> 2010/4/8 Ian Lance Taylor<iant@google.com>:
>> "Doug Kwan (Ãö®¶¼w)"<dougkwan@google.com>  writes:
>>
>>>      This patch fixes a problem exposed by bootstrapping gcc.  libjava
>>> contains objects with unevenly sized merge sections, whose sections
>>> sizes are not multiples of their entsizes.
>>
>> That sounds like a bug somewhere.
>>
>>
>>> 2010-04-08  Doug Kwan<dougkwan@google.com>
>>>
>>>          * merge.cc (Output_merge_data::set_final_data_size): Handle empty
>>>          Output_merge_data.
>>>          * output.cc (Output_section::add_merge_input_section): Simplify
>>>          code and return status of Output_merge_base::add_input_section.
>>>          Update merge section map only if Output_merge_base::add_input_section
>>>          returns true.
>>
>> This is OK.
>>
>> Thanks.
>>
>> Ian
>>
> 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]