Problems with 2.3.5b -> Makedev, Compiler Ops and Bash Problem

Simon Perreault nomis80 at
Wed Jun 21 19:40:00 PDT 2000

About the Celerons: I'm really not sure about those. I'm not sure if they
use Pentium Pro architecture or simple Pentium. i686 is for PPro, ie. PPro,
PII and PIII. i586 is for Pentium and Pentium compatibles, ie. Pentium, k6,
k6-2, k6-3. I'm really not sure about Celerons. But if i686 does not work,
you should use i586. Also, what does the kernel says about computer
architecture selection (when you choose '?' instead of 'y' or 'n')? Maybe it
would tell you what architecture your Celeron uses. It told me that my k6
used i586.

What's very important is that you download PGCC (Pentium GCC) at for any real Pentium optimization (Celeron is
considered a pentium, as are the k6's. Not pentium are i486 and less, SPARC,
PPC, etc.) PGCC is a patch to GCC. There is a patch available for the latest
GCC release (2.95.2). Then you should use "-O6 (that,s an O(letter), not a
0(number)) -mcpu=pentium (or pentiumpro if you have determined that your
Celeron is indeed based on the Pentium Pro architecture) -march=pentium
(same thing)". You put this string in your .bash_profile to define CFLAGS
and CXXFLAGS variables, using "export CFLAGS="<string>"". So then the CFLAGS
and CXXFLAGS are defined (CXXFLAGS is for the C++ compiler).

What the -e switch does to make is that all environment variables take
precedence over those defined in the Makefile. So if the Makefile defines
it's own CFLAGS, and you override it with your own, it may very well create
compilation problems. There are special switches that are used in the
Makefile's that can't be overridden.

When you run ./configure, it checks at your environment defined CFLAGS (the
one you put in .bash_profile) and adds those CFLAGS to all the CFLAGS in the
Makefile. So what is does is preserve the integrity of the CFLAGS in the
Makefile while also using the optimization options you selected. So running
make -e is unnecessary. But here I'm talking about the second part of the
building process, the one where you enter chroot and install programs
dynamically. You should never use -e in that part (that's only my advice, I
still have to hear a refuting argument from Gerard). I've always wondered
why Gerard makes use of the -e switch in the static building. I have not
used it in the static building either, and I have not encountered any
problems. So my advice is still to get rid of all those -e you encounter.

As usual, this is only my newbie advice.

""Ian Chilton"" <ian at> wrote in message
> Hello,
> > celeron up from 300A are based on pentium pro, and the celeron below
> > (that includes the 300) are not, so they should be 586.
> Mine is a Celeron 533....I set it to Pentium Pro in the kernel, but what
> compiler flags do I use then?
> And, when you say don't use -e, where is this?
> In the book, Gerard seems to use:
> ./configure
> make
> make -e --prefix=/lfs/usr install
> So, do I take the '-e' out of the make install, or is the above OK?
> Thanks!
> Bye for Now,
> Ian
>                                 \|||/
>                                 (o o)
>  /---------------------------ooO-(_)-Ooo---------------------------\
>  |  Ian Chilton     (IRC Nick - GadgetMan)     ian at  |
>  |-----------------------------------------------------------------|
>  |  Backup E-Mail -->  ian at      ICQ #: 16007717  |
>  |  Mobile Phone (MAX 150 Chars!) -->  ianchilton at  |
>  |-----------------------------------------------------------------|
>  |  --> Main Page                     |
>  |    --> Own Web Server (When Online)  |
>  |-----------------------------------------------------------------|
>  |  "Unix is user friendly - it's just picky about it's friends."  |
>  \-----------------------------------------------------------------/

Mail archive:
IRC access: server: port: 6667 channel: #LFS
Unsubscribe: email lfs-discuss-request at and put
"unsubscribe" (without the quotation marks) in the body of the message
(no subject is required)

More information about the lfs-dev mailing list