[Click] AlignmentInfo Question

[Eddie Kohler]

Hi Roberto,

The config.log from the qemu compile might help me debug.  Just send it 
to me (they are often too big for the list).

Eddie


Roberto Riggio wrote:
> Eddie,
> 
> I've done some tests on different machines (all x86). On ubuntu (gcc 4.1.2)
> everything work fine.
> 
> However I'm also compiling click inside a qemu virtual machine using a debian etch 
> distribution updated to unstable (gcc 4.1.1).
> 
> In this latter case the configure script is not able to determine if the machine is 
> indifferent to alignment (as you can see from the attached config.h).
> 
> So I cannot understand if the problem is GCC or QEMU.
> 
> Roberto
> 
> ----- Original Message -----
> From: Eddie Kohler <kohler at cs.ucla.edu>
> To: roberto riggio <roberto.riggio at create-net.org>
> Cc: Click <click at amsterdam.lcs.mit.edu>
> Sent: Monday, April 23, 2007 6:12:28 PM GMT+0100 Europe/Berlin
> Subject: Re: [Click] AlignmentInfo Question
> 
> Roberto,
> 
> It looks like recent GCCs caught a mistake in one of our configuration test 
> programs.  As a result, Click's configuration process was not able to 
> determine whether a machine was indifferent to alignment, at leas ton my 
> machine.  This should be fixed now.
> 
> If the updates don't help you, send your generated config.h, and tell us what 
> type of machine you're running on.
> 
> Eddie
> 
> 
> Roberto Riggio wrote:
>> Hi,
>>
>> I'm puzzled, is htat supposed to be just a warning?
>>
>> I'm using the latest CVS version and I've tried to run the click
>> configuration in the following way:
>>
>> click -w  running.click 
>>
>> running.click:221: While configuring 'Classifier at 16 :: Classifier':
>>   no AlignmentInfo available: you may experience unaligned accesses
>> Router could not be initialized!
>>
>> The same script was running fine without an older version of click.
>>
>> Is there a way to fix the problem? 
>>
>> Roberto
>>
>> ----- Original Message -----
>> From: Eddie Kohler <kohler at cs.ucla.edu>
>> To: Fischofer William-B09792 <Bill.Fischofer at freescale.com>
>> Cc: Click <click at pdos.csail.mit.edu>
>> Sent: Monday, March 26, 2007 10:02:21 PM GMT+0100 Europe/Berlin
>> Subject: Re: [Click] AlignmentInfo Question
>>
>> Ah -- the problem was that a test that should have been generic (#if 
>> !HAVE_INDIFFERENT_ALIGNMENT) was architecture specific (#ifndef 
>> __i386__).  Fixed; it may take some time to make its way to the 
>> anonymous CVS.  In the meantime you can just ignore the warning.
>>
>> Eddie
>>
>>
>> Fischofer William-B09792 wrote:
>>> Attached.  Thanks for your help. 
>>>
>>> Bill
>>>
>>> -----Original Message-----
>>> From: Eddie Kohler [mailto:kohler at cs.ucla.edu] 
>>> Sent: Monday, March 26, 2007 2:13 PM
>>> To: Fischofer William-B09792
>>> Cc: click at amsterdam.lcs.mit.edu
>>> Subject: Re: [Click] AlignmentInfo Question
>>>
>>> Huh.  Is "int" 64 bits wide on AMD64?  I'd appreciate it if you sent me
>>> your "config.log" file (don't cc the list, those files are large) so I
>>> can take a look.
>>>
>>> Eddie
>>>
>>>
>>> Fischofer William-B09792 wrote:
>>>> Actually, this is running on Linux on an AMD64 box.  I'll be moving 
>>>> this to a PPC-based platform later but I wanted to get things running 
>>>> on an x86-compatible box first.  I did look up click-align and will 
>>>> use that if it seems advisable but I was just surprised to get that 
>>>> error message (even moreso now that you say I shouldn't have gotten it
>>>> in the first place).
>>>>
>>>> Bill
>> _______________________________________________
>> click mailing list
>> click at amsterdam.lcs.mit.edu
>> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>>
>>
> 
>