Problem with compile test after compiling glibc LFS 7.2
 

LFS Version: 7.2
Base System: Debian wheezy
Chapter: 5.7.2

Version Check:
Command:
Error:

Did you login to new user or "su -lfs" before going to Chapter 4.2.

Look at this thread here
http://www.linuxquestions.org/questi...bc-4175419132/

EDDY1, I did all steps as user lfs so lfs owns all the files created (su'ed to lfs). lfs owns the directorie /mount/LFS and all sub dirs so thats not the problem.

Also I saw that post first and was then told as the last post on that thred says, to start a new topic for the problem as that one is solved and is for LFS 7.1 where as I am stuck with LFS 7.2

I did get another error with my first attempt and then realised I miss read the version-info and /bin/sh was a sym link to /bin/dash not /bin/bash. I sorted that out then started again and now I get this problem.

I am rinning the host system (Debian Wheezy) on a virtual machine (virtual box) under Debian Squeeze if that makes any diference.

It will build in virtualbox
If you are on debian host you also created lfs user you can enter at login window, but I think that you missed something on you second go around, right now you are still a the early stages of your build I would just start over.

I have the same issue

gcc (Debian 4.7.2-4) 4.7.2 > 4.7.1 might be the issue

dunno yet

Lancelot1 Yea I thought maybe the same but I cant find out how to downgrade from 4.7.2 to 4.7.1
The only other version available for wheezy is 6.x.x, I can't figure out how to use that one instead and I can't uninstall 4.7.2 without messing up my system.

@joneschrisan: Can you post the output of the following commands (log in as user lfs):

$ cat ~/.bash_profile
$ cat ~/.bashrc
$ uname -m
$ echo $PATH
$ echo 'main(){}' > dummy.c
$ $LFS_TGT-gcc -v dummy.c

Did you use /mnt/LFS in all the stages of the book (Why deviate from the book)?

Check to see if those 3 files mentioned are present on your system. As root, run the following commands from your host and post the output:

I did use /mnt/LFS in all stages yes. How is that deviating from the book?

locate -r "/crt[1in].o" gives me no output.
But:

# find / -name "crt[1in].o"
However:

# locate -r "/crt[1in]*"

The book uses /mnt/lfs and not /mnt/LFS, which is easily mixed up.

You did update the locate database first (using updatedb)?

That looks ok, although I'm not sure about the bold part (on debian squeeze they are located in /usr/lib, not /usr/lib/i386-linux-gnu).

It seems that it is looking in the wrong place, are you sure you did not make a mistake in chapter 5.5? Especially this part:

Ok then. Time to start again from th begining. I didnt realise I was doing it diferentle with LFS. Sorry. See you in a couple of days and I'll let you know if it works.

Ok. I have the same problem as joneschrisan.

LFS version 7.2
Host: Backtrack5 (ubuntu 10.04)
Chapter 5.7.1

version-check.sh
cat ~/.bash_profile
cat ~/.bashrc
uname -m
echo $PATH
echo 'main(){}' > dummy.c
$LFS_TGT-gcc -v dummy.c
updatedb; locate -r "/crt[1in].o"
So, here is where I stand. I got 2/3 of the way through an LFS build had a segmentation fault and lost grep somehow and so I decided to rebuild from scratch.

My *second* attempt got me stuck in this same place and I thought I had cleared everything up wrong by just reformatting the partition.

My *third* attempt I deleted the partition and recreated it and I am still stuck at the exact same place.

Here is what joneschrisan says Now I'm wondering, is there an issue somehow with having to re-do the build from step 1?

So I have an update to this now.

I, once again, blew away this try and started from scratch only THIS time I created a new user lfsuser and set everything up for that user exactly as the book says to do for lfs.

It worked!!

The only thing that I can think of is that somehow the profile information for lfs user is changed enough that when starting the build from scratch requires either to delete and recreate the lfs account or start with a new user account like I did.

I did not delete lfs and would love to know what caused this issue, but I am thankfully past this part and can move on and rebuild!

One part to note from the build this time is the last information that was displayed from make install of the glibc.

The important thing to note here is that the test condition on my second and third attempts actually echoed 'stubs.h unchanged'. Where as from a new account you can see that it doesn't echo and actually proceeds with the install. I'm sure there are more of these throughout the install messages but this is one that is easy to see since it's at the end of the instal.

Hope this helps someone to get past this part.