Re: g++ serious bug in Ubuntu Lucid Lynx

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



This may help as well:

http://clinuxpro.com/Cprogramming/structurepadding.php

On Wed, Jun 2, 2010 at 12:26 PM, Brian Budge <brian.budge@xxxxxxxxx> wrote:
> I'm surprised this wasn't the case before.  AFAIK, structs with
> doubles have been aligned to at least an 8 byte boundary for years.
>
> You may need to write a new reader that knows the layout on disk and
> reconstructs the data.
>
>  Brian
>
> On Wed, Jun 2, 2010 at 9:16 AM, Christian Convey
> <christian.convey@xxxxxxxxx> wrote:
>> Are you sure that's a bug in the legalistic sense?  IIRC, C and C++
>> don't promise any particular memory layout for structs.
>>
>> On Wed, Jun 2, 2010 at 11:53 AM, David R. Doucette
>> <ddoucette@xxxxxxxxxxxxxxxx> wrote:
>>> We just installed Lucid Lynx a few days ago, and have hit a fatal flaw with
>>> g++ that has forced us to stop the deployment and roll back the systems we
>>> could.
>>>
>>> Structures in Lucid are now mapping differently than in the past, so we can
>>> no longer read our own files! Further, if any files are written by code
>>> produced by the new g++, they are corrupted.
>>>
>>> The problem is that doubles are now being mapped onto a boundary divisible
>>> by 8 rather than a boundary divisible by 4, meaning that offsets change and
>>> structs get bigger.
>>>
>>> I've attached a tar file showing the mapping on Jaunty, Karmic, and Lucid to
>>> show the change. This is a greatly simplified program that shows the
>>> problem, but it's happening in all of our C++ programs.
>>>
>>> HELP!
>>>
>>> System is Ubuntu 10.4 Lucid Lynx 64 bit.
>>>
>>> David Doucette
>>>
>>
>



[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux