QLFS project announcement

J. Scott Edwards sedwards at xmission.com
Wed Aug 25 06:20:51 PDT 2004


On Tue, 24 Aug 2004, T_B wrote:

> "J. Scott Edwards" <sedwards at xmission.com> wrote in message
> news:Pine.LNX.4.61.0408230645590.13668 at xmission.xmission.com...
>>
>> On Sun, 22 Aug 2004, Archaic wrote:
>>
>>
>> Would it be a good idea to try to build HLFS without any of the security
>> patches and then if that works add the security patches one at a time
>> until I find out what is causing the failure?
>>
> I am experiencing the same problems with glibc build in chapter 6 as Jan
> Mattila reported earlier and you seem to be reporting here.  I was following
> the CVS HLFS book that is available here
> http://linuxfromscratch.org/~robert/hlfs/downloads/cvs/

<SNIP>

> Now since the only difference between the build of glibc in chapter 5 and
> chapter 6 is the patches applied to gcc, I assume that this may be the
> underlying cause of the problems.  So I proceeded with the assumption that
> maybe it is not glibc that has a problem in chapter 6 but that the test
> suite has a problem and is reporting errors falsely because they have not
> been modified to accommodate the changes that occur as a result of the gcc
> patches.
>

I think you may be onto something.  I don't remember now which tests I had 
fail, but I do remember that it was some numbers that it was complaining 
about and I examined the results and they appeared to be correct to me.  I 
will try to build it again and do what you did in Ch.6 make -k check and I 
will report back which tests I fail.

BTW - I am on a Pentium 4 as well, could it be this is related to a 
specific type of CPU?  It seems strange to me that different CPU's could 
get different mathematical results.  Could a P4 get a (slightly) different 
answer on a floating point operation than say PIII and it not be a bug in 
the CPU?  Like is there fuzz in the least significant bit or something?

Thanks
    -Scott





More information about the hlfs-dev mailing list