[SOLVED] [LFS 11.0 - systemd] GCC test suite errors in Ch. 8.26

peristaltic · 09-04-2021, 05:02 PM

Hello --

Book version: LFS 11.0 systemd
Host system: Linux Mint 20.1
All tests passed in the HSR script
I tried using the search function here on LQ but got a "database error".

Everything has gone without a hitch until now, Chapter 8.26 - GCC-11.2.0

I am in the chroot environment. After running the "make check" command as "tester", I reviewed the summary of failures by running:

Code:

../contrib/test_summary | grep -A7 Summ

The number of unexpected failures in g++ worries me.

Code:

		=== g++ Summary ===

# of expected passes		210136
# of unexpected failures	2666
# of expected failures		1073
# of unresolved testcases	203
# of unsupported tests		9048
/sources/gcc-11.2.0/build/gcc/xg++  version 11.2.0 (GCC) 
--

This seems like a very large number, far more than the dozen or so failures listed as expected in the LFS chapter 8.26 page.

My questions are:

should I be concerned about this number of failures?
if this is indeed a problem, what went wrong and how to resolve it? I'm really hoping to avoid starting over because as you know it's a lot of manual typing.

I can provide any and all command output requested that would help.

Thanks!

Keith Hedger · 09-05-2021, 10:29 AM

If the sanity checks pass ok i wouldn't worry too much about the number of test errors, a lot of them really only apply to devs of gcc itself

P.S
As an aside I've been using LFS for years and i never run the tests just the sanity checks.

peristaltic · 09-06-2021, 07:33 PM

Quote:

Originally Posted by Keith Hedger

If the sanity checks pass ok i wouldn't worry too much about the number of test errors, a lot of them really only apply to devs of gcc itself

P.S
As an aside I've been using LFS for years and i never run the tests just the sanity checks.

Thanks for your response. I realize I am being a little paranoid but I also understand the way that each chapter of LFS builds on the previous ones. GCC is probably a fairly important component for everything else I do going forrward. I didn't want to get to the end and find that ignoring these test failures had tainted the final product.

Despite your assurances, I'm not going to mark this thread as "Solved' because I'm not 100% sure of what lies ahead for this LFS build. But believe me if I ultimately encounter problems, someone's gonna hear about it...

If I never update this thread again, take it as a good sign.

anak_bawang · 09-06-2021, 10:51 PM

Quote:

Originally Posted by Keith Hedger

P.S
As an aside I've been using LFS for years and i never run the tests just the sanity checks.

I did all the test only for the first time I built LFS.

Now, I ignore all the tests except for glibc, gmp, mpfr, and mpc.

peristaltic · 09-23-2021, 11:14 PM

Quote:

Originally Posted by peristaltic

Hello --
...
Thanks!

Posting to resolve this thread in case anyone experienced the same concerns I did.

My LFS system has booted and performed successfully so I wouldn't worry about all those warnings when running tests against the build of GCC in Chapter 8.

Side note: I have started an LFS build 3 times since 2005 and this is the first time I reached the ultimate goal of a bootable system. I'm very grateful for the advice I have received here on these forums.

hazel · 09-24-2021, 05:14 AM

I find this thread extraordinary. The Book emphasises the necessity of the tests for a small number of crucial packages (binutils, gcc, mpc, mpfr, gmp and glibc). They are "on no account to be omitted". The rest of the tests are less important and I don't usually bother with them any more, but I religiously carry out the tests on these packages. Fortunately I have never had the kind of error splurge that the OP had.

But now it turns out that you can have all that and still end up with a bootable LFS. So why are these tests described as so essential?