# Broken lapack/blas in Rocky Linux 9.3

**URL:** <https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501>\
**Category:** Rocky Linux Help & Support\
**Created:** [January 15, 2024, 6:32pm UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501 "2024-01-15T18:32:21Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![smith](https://avatars.discourse-cdn.com/v4/letter/s/f04885/32.png) [@smith](https://forums.rockylinux.org/u/smith)\
**Post date:** [January 15, 2024, 6:32pm UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/1 "2024-01-15T18:32:21Z")

</div>

I’m lacking a minimal example, but wanted to reach out to the community. Maybe someone has similar issues or does know more about the lapack package for 9.3

Recently we upgraded in our group (working in the field of computational chemistry) from 9.2 to 9.3. Since then our codes using lapack routines are broken. The programs terminate with internal lapack errors indicating that wrong dimensions were passed to the routines. The problems disappears as soon as we link to a different lapack implementation e.g. Intel MKL (which is anyway the preferred choice)

Did someone experience something similar? Or does someone know more details about the lapack rpm?

---

<div class="post-metadata">

**Author:** ![smith](https://avatars.discourse-cdn.com/v4/letter/s/f04885/32.png) [@smith](https://forums.rockylinux.org/u/smith)\
**Post date:** [January 16, 2024, 2:08pm UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/2 "2024-01-16T14:08:22Z")

</div>

Already a very simple example is broken

```fortran
program bench

  implicit none

  real(8), allocatable :: mat(:,:), a(:,:)

  integer :: n

  n = 100

  ! Create random positive definite matrices
  write(6,'(3x,a,i8)') "I will create matrices of size",n

  allocate(a(n,n), mat(n,n))

  call random_number(a)

  call dgemm('n', 't', n, n, n, 1.d0, a, n, a, n, 0.d0, mat, n)

end program

```

Compiling it with gfortran an linking against the lapack installation

```bash
gfortran -fallow-argument-mismatch main.f90 -L /usr/lib64/ -lblas -L/usr/lib64 -llapack

```

produces an executable which seg-faults in the dgemm call. If the program is compiled an linked against the MKL it works

On the machine lapack-devel is installed.

EDIT:  
I think I understand now what is going on. The lapack/blas version is compiled for a 8 byte integer and not a 4 byte integer, which is the fortran standard. There are flags for the compiler to convert all 4 byte integers into 8 byte integers, but maybe there is also a flag which I can use to link to the 4 byte integer version assuming that it is also shipped

---

<div class="post-metadata">

**Author:** ![johnsow1](https://avatars.discourse-cdn.com/v4/letter/j/85e7bf/32.png) [@johnsow1](https://forums.rockylinux.org/u/johnsow1)\
**Post date:** [January 18, 2024, 9:11pm UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/3 "2024-01-18T21:11:04Z")

</div>

Does anyone know why Rocky 9.2 has lapack64\_ but Rocky 9.3 has lapack64\_ and lapack64 (with no underscore)?

---

<div class="post-metadata">

**Author:** ![gsal](https://avatars.discourse-cdn.com/v4/letter/g/48db29/32.png) [@gsal](https://forums.rockylinux.org/u/gsal)\
**Post date:** [January 19, 2024, 1:20am UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/4 "2024-01-19T01:20:59Z")

</div>

Hey, smith, thanks for the hint about the size of integers.

I run into this issue myself in Rocky 9.3 and couldn’t figure out the problem even after testing on Rocky 9.2 and seeing my program work just fine.

Then, I noticed that even though I am requesting “-llapack” at compile time, the run time error message says that the problem is inside lapack64!

The way I see it, if I say “-llapack”, it should pick up lapack; if I say “-llapack64”, it should then pick up lapack64…but that does not seem to be happening OR, lapack is also 64…and it shouldn’t be?

---

<div class="post-metadata">

**Author:** ![smith](https://avatars.discourse-cdn.com/v4/letter/s/f04885/32.png) [@smith](https://forums.rockylinux.org/u/smith)\
**Post date:** [January 19, 2024, 8:06am UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/5 "2024-01-19T08:06:26Z")

</div>

Hi gsal, I think 9.3 is missing a 32 bit integer version of lapack. I’m wondering if this is a Rocky issue or a Red Hat issue. Rocky should be bit by bit compatible to RHEL, but I don’t know if the rpms are maybe rebuild on the Rocky side and something went wrong in the lapack case.

---

<div class="post-metadata">

**Author:** ![label](https://avatars.discourse-cdn.com/v4/letter/l/c5a1d2/32.png) [@label](https://forums.rockylinux.org/u/label)\
**Post date:** [January 19, 2024, 9:20am UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/6 "2024-01-19T09:20:20Z")

</div>

To preface this post, this is all that changed with the RPM: [import lapack-3.9.0-10.el9 (d6d9cc1a) · Commits · staging / rpms / lapack · GitLab](https://git.rockylinux.org/staging/rpms/lapack/-/commit/d6d9cc1affa3a4e2a3fbef2b9610b65f9444ef7f) - Not much changed here, except for the addition of explicit requires for some packages.

> [@johnsow1](#):
>
> Does anyone know why Rocky 9.2 has lapack64\_ but Rocky 9.3 has lapack64\_ and lapack64 (with no underscore)?

This is how pungi (our initial composing tool) determined what was needed to be in the repositories. The packages have always been built (you can actually verify that [in the vault devel 9.2 repo](https://dl.rockylinux.org/vault/rocky/9.2/devel/x86_64/os/Packages/l/), but pungi is what ultimately determines what will be in the repos by looking at dependencies of each package. Pungi is what decided it needed to be there.

If you look at the rpm description, they both say they are for 64bit integars, but `lapack64_` has a symbol name suffix. see `dnf repoquery -q lapack64 -i` and `dnf repoquery -q lapack64_ -i`.

> [@smith](#):
>
> but I don’t know if the rpms are maybe rebuild on the Rocky side and something went wrong in the lapack case.

Below are the dependencies used to build lapack for 9.3.

```auto
================================================================================
 Package Arch Version Repository Size
================================================================================
Installing:
 gcc-gfortran x86_64 11.4.1-2.1.el9 yumrepofs_0 13 M
Installing dependencies:
 annobin x86_64 12.12-1.el9 yumrepofs_0 977 k
 cpp x86_64 11.4.1-2.1.el9 yumrepofs_0 11 M
 gcc x86_64 11.4.1-2.1.el9 yumrepofs_0 32 M
 gcc-plugin-annobin x86_64 11.4.1-2.1.el9 yumrepofs_0 48 k
 glibc-devel x86_64 2.34-83.el9.7 yumrepofs_0 50 k
 glibc-headers x86_64 2.34-83.el9.7 yumrepofs_0 450 k
 kernel-headers x86_64 5.14.0-362.el9 yumrepofs_0 6.4 M
 libedit x86_64 3.1-38.20210216cvs.el9 yumrepofs_0 103 k
 libgfortran x86_64 11.4.1-2.1.el9 yumrepofs_0 807 k
 libmpc x86_64 1.2.1-4.el9 yumrepofs_0 61 k
 libquadmath x86_64 11.4.1-2.1.el9 yumrepofs_0 197 k
 libquadmath-devel x86_64 11.4.1-2.1.el9 yumrepofs_0 38 k
 libxcrypt-devel x86_64 4.4.18-3.el9 yumrepofs_0 28 k
 llvm-libs x86_64 16.0.6-4.el9 yumrepofs_0 24 M

```

The build logs are [here](https://peridot.build.resf.org/6202c09e-6252-4d3a-bcd3-9c7751682970/tasks/b3c1c9eb-adb7-4d87-88b0-616a301bc3de) if you wish to look at them. (Warning, it is difficult to parse through the logs. We are working in the background to make the UI better.).

Whether or not that you look through the build logs, if you believe something is indeed wrong, we ask that you file a bug report with as much detail as possible. This includes debug data (e.g. gdb), related installed packages, example broken scripts/programs. [Bug Tracker](https://bugs.rockylinux.org).

> [@smith](#):
>
> Already a very simple example is broken

It works if you have glibc-devel.i686 (and other i686 packages) and you use `-m32`, which I’m sure goes against yours (and others) expectations.

```auto
[root@awx ~]# dnf install lapack lapack.i686 lapack64 lapack-devel.i686 lapack-devel glibc-devel.i686 -y
Last metadata expiration check: 3:02:57 ago on Thu 18 Jan 2024 11:04:47 PM MST.
Dependencies resolved.
================================================================================
 Package Arch Version Repository Size
================================================================================
Installing:
 glibc-devel i686 2.34-83.el9.7 appstream 50 k
 lapack i686 3.9.0-10.el9 appstream 8.5 M
 lapack x86_64 3.9.0-10.el9 appstream 8.5 M
 lapack-devel i686 3.9.0-10.el9 crb 62 k
 lapack-devel x86_64 3.9.0-10.el9 crb 63 k
 lapack64 x86_64 3.9.0-10.el9 crb 2.2 M
Installing dependencies:
 blas i686 3.9.0-10.el9 appstream 456 k
 blas x86_64 3.9.0-10.el9 appstream 479 k
 blas-devel i686 3.9.0-10.el9 crb 14 k
 blas-devel x86_64 3.9.0-10.el9 crb 15 k
 blas64 x86_64 3.9.0-10.el9 crb 209 k
 blas64_ x86_64 3.9.0-10.el9 crb 209 k
 cpp x86_64 11.4.1-2.1.el9 appstream 11 M
 gcc x86_64 11.4.1-2.1.el9 appstream 32 M
 gcc-gfortran x86_64 11.4.1-2.1.el9 appstream 13 M
 glibc i686 2.34-83.el9.7 baseos 1.9 M
 glibc-devel x86_64 2.34-83.el9.7 appstream 50 k
 glibc-headers x86_64 2.34-83.el9.7 appstream 450 k
 kernel-headers x86_64 5.14.0-362.13.1.el9_3 appstream 6.4 M
 lapack64_ x86_64 3.9.0-10.el9 crb 2.2 M
 libgcc i686 11.4.1-2.1.el9 baseos 104 k
 libgfortran i686 11.4.1-2.1.el9 baseos 774 k
 libgfortran x86_64 11.4.1-2.1.el9 baseos 807 k
 libmpc x86_64 1.2.1-4.el9 appstream 61 k
 libquadmath i686 11.4.1-2.1.el9 baseos 245 k
 libquadmath x86_64 11.4.1-2.1.el9 baseos 197 k
 libquadmath-devel x86_64 11.4.1-2.1.el9 appstream 38 k
 libxcrypt i686 4.4.18-3.el9 baseos 121 k
 libxcrypt-devel i686 4.4.18-3.el9 appstream 28 k
 libxcrypt-devel x86_64 4.4.18-3.el9 appstream 28 k
 make x86_64 1:4.3-7.el9 baseos 530 k
Installing weak dependencies:
 glibc-gconv-extra i686 2.34-83.el9.7 baseos 1.6 M

Transaction Summary
================================================================================
Install 32 Packages

Total download size: 92 M
Installed size: 225 M
Downloading Packages:
(1/32): libxcrypt-4.4.18-3.el9.i686.rpm 369 kB/s | 121 kB 00:00
(2/32): make-4.3-7.el9.x86_64.rpm 1.3 MB/s | 530 kB 00:00
(3/32): libquadmath-11.4.1-2.1.el9.i686.rpm 577 kB/s | 245 kB 00:00
(4/32): libgcc-11.4.1-2.1.el9.i686.rpm 1.8 MB/s | 104 kB 00:00
(5/32): libquadmath-11.4.1-2.1.el9.x86_64.rpm 2.6 MB/s | 197 kB 00:00
(6/32): libgfortran-11.4.1-2.1.el9.i686.rpm 3.7 MB/s | 774 kB 00:00
(7/32): libgfortran-11.4.1-2.1.el9.x86_64.rpm 5.3 MB/s | 807 kB 00:00
(8/32): glibc-gconv-extra-2.34-83.el9.7.i686.rp 7.9 MB/s | 1.6 MB 00:00
(9/32): glibc-2.34-83.el9.7.i686.rpm 8.3 MB/s | 1.9 MB 00:00
(10/32): libxcrypt-devel-4.4.18-3.el9.x86_64.rp 97 kB/s | 28 kB 00:00
(11/32): libmpc-1.2.1-4.el9.x86_64.rpm 146 kB/s | 61 kB 00:00
(12/32): libxcrypt-devel-4.4.18-3.el9.i686.rpm 97 kB/s | 28 kB 00:00
(13/32): blas-3.9.0-10.el9.x86_64.rpm 1.0 MB/s | 479 kB 00:00
(14/32): lapack-3.9.0-10.el9.x86_64.rpm 10 MB/s | 8.5 MB 00:00
(15/32): kernel-headers-5.14.0-362.13.1.el9_3.x 6.9 MB/s | 6.4 MB 00:00
(16/32): blas-3.9.0-10.el9.i686.rpm 3.5 MB/s | 456 kB 00:00
(17/32): libquadmath-devel-11.4.1-2.1.el9.x86_6 389 kB/s | 38 kB 00:00
(18/32): lapack-3.9.0-10.el9.i686.rpm 12 MB/s | 8.5 MB 00:00
(19/32): gcc-gfortran-11.4.1-2.1.el9.x86_64.rpm 19 MB/s | 13 MB 00:00
(20/32): glibc-headers-2.34-83.el9.7.x86_64.rpm 3.7 MB/s | 450 kB 00:00
(21/32): cpp-11.4.1-2.1.el9.x86_64.rpm 16 MB/s | 11 MB 00:00
(22/32): glibc-devel-2.34-83.el9.7.x86_64.rpm 411 kB/s | 50 kB 00:00
(23/32): glibc-devel-2.34-83.el9.7.i686.rpm 514 kB/s | 50 kB 00:00
(24/32): lapack-devel-3.9.0-10.el9.x86_64.rpm 226 kB/s | 63 kB 00:00
(25/32): blas64_-3.9.0-10.el9.x86_64.rpm 1.3 MB/s | 209 kB 00:00
(26/32): gcc-11.4.1-2.1.el9.x86_64.rpm 19 MB/s | 32 MB 00:01
(27/32): blas-devel-3.9.0-10.el9.x86_64.rpm 48 kB/s | 15 kB 00:00
(28/32): lapack-devel-3.9.0-10.el9.i686.rpm 812 kB/s | 62 kB 00:00
(29/32): lapack64_-3.9.0-10.el9.x86_64.rpm 2.2 MB/s | 2.2 MB 00:00
(30/32): blas-devel-3.9.0-10.el9.i686.rpm 64 kB/s | 14 kB 00:00
(31/32): blas64-3.9.0-10.el9.x86_64.rpm 2.3 MB/s | 209 kB 00:00
(32/32): lapack64-3.9.0-10.el9.x86_64.rpm 5.6 MB/s | 2.2 MB 00:00
--------------------------------------------------------------------------------
Total 18 MB/s | 92 MB 00:05
Running transaction check
Transaction check succeeded.
Running transaction test
Transaction test succeeded.
Running transaction
  Preparing : 1/1
  Installing : libgcc-11.4.1-2.1.el9.i686 1/32
  Running scriptlet: libgcc-11.4.1-2.1.el9.i686 1/32
  Installing : kernel-headers-5.14.0-362.13.1.el9_3.x86_64 2/32
  Installing : glibc-gconv-extra-2.34-83.el9.7.i686 3/32
  Running scriptlet: glibc-gconv-extra-2.34-83.el9.7.i686 3/32
  Running scriptlet: glibc-2.34-83.el9.7.i686 4/32
  Installing : glibc-2.34-83.el9.7.i686 4/32
  Running scriptlet: glibc-2.34-83.el9.7.i686 4/32
  Installing : libquadmath-11.4.1-2.1.el9.x86_64 5/32
  Installing : libgfortran-11.4.1-2.1.el9.x86_64 6/32
  Installing : blas-3.9.0-10.el9.x86_64 7/32
  Installing : libmpc-1.2.1-4.el9.x86_64 8/32
  Installing : blas64_-3.9.0-10.el9.x86_64 9/32
  Installing : blas64-3.9.0-10.el9.x86_64 10/32
  Installing : glibc-headers-2.34-83.el9.7.x86_64 11/32
  Installing : libxcrypt-devel-4.4.18-3.el9.x86_64 12/32
  Installing : glibc-devel-2.34-83.el9.7.x86_64 13/32
  Installing : lapack64-3.9.0-10.el9.x86_64 14/32
  Installing : lapack64_-3.9.0-10.el9.x86_64 15/32
  Installing : cpp-11.4.1-2.1.el9.x86_64 16/32
  Installing : lapack-3.9.0-10.el9.x86_64 17/32
  Installing : make-1:4.3-7.el9.x86_64 18/32
  Installing : gcc-11.4.1-2.1.el9.x86_64 19/32
  Installing : libquadmath-devel-11.4.1-2.1.el9.x86_64 20/32
  Installing : gcc-gfortran-11.4.1-2.1.el9.x86_64 21/32
  Installing : blas-devel-3.9.0-10.el9.x86_64 22/32
  Installing : libxcrypt-4.4.18-3.el9.i686 23/32
  Installing : glibc-devel-2.34-83.el9.7.i686 24/32
  Installing : libxcrypt-devel-4.4.18-3.el9.i686 25/32
  Installing : libquadmath-11.4.1-2.1.el9.i686 26/32
  Installing : lapack-devel-3.9.0-10.el9.x86_64 27/32
  Installing : libgfortran-11.4.1-2.1.el9.i686 28/32
  Installing : blas-3.9.0-10.el9.i686 29/32
  Installing : blas-devel-3.9.0-10.el9.i686 30/32
  Installing : lapack-3.9.0-10.el9.i686 31/32
  Installing : lapack-devel-3.9.0-10.el9.i686 32/32
  Running scriptlet: lapack-devel-3.9.0-10.el9.i686 32/32
  Verifying : libxcrypt-4.4.18-3.el9.i686 1/32
  Verifying : make-1:4.3-7.el9.x86_64 2/32
  Verifying : libquadmath-11.4.1-2.1.el9.i686 3/32
  Verifying : libgfortran-11.4.1-2.1.el9.i686 4/32
  Verifying : libgcc-11.4.1-2.1.el9.i686 5/32
  Verifying : libquadmath-11.4.1-2.1.el9.x86_64 6/32
  Verifying : libgfortran-11.4.1-2.1.el9.x86_64 7/32
  Verifying : glibc-gconv-extra-2.34-83.el9.7.i686 8/32
  Verifying : glibc-2.34-83.el9.7.i686 9/32
  Verifying : libmpc-1.2.1-4.el9.x86_64 10/32
  Verifying : libxcrypt-devel-4.4.18-3.el9.x86_64 11/32
  Verifying : libxcrypt-devel-4.4.18-3.el9.i686 12/32
  Verifying : kernel-headers-5.14.0-362.13.1.el9_3.x86_64 13/32
  Verifying : lapack-3.9.0-10.el9.x86_64 14/32
  Verifying : blas-3.9.0-10.el9.x86_64 15/32
  Verifying : lapack-3.9.0-10.el9.i686 16/32
  Verifying : blas-3.9.0-10.el9.i686 17/32
  Verifying : libquadmath-devel-11.4.1-2.1.el9.x86_64 18/32
  Verifying : gcc-gfortran-11.4.1-2.1.el9.x86_64 19/32
  Verifying : gcc-11.4.1-2.1.el9.x86_64 20/32
  Verifying : cpp-11.4.1-2.1.el9.x86_64 21/32
  Verifying : glibc-headers-2.34-83.el9.7.x86_64 22/32
  Verifying : glibc-devel-2.34-83.el9.7.x86_64 23/32
  Verifying : glibc-devel-2.34-83.el9.7.i686 24/32
  Verifying : lapack64_-3.9.0-10.el9.x86_64 25/32
  Verifying : lapack-devel-3.9.0-10.el9.x86_64 26/32
  Verifying : blas64_-3.9.0-10.el9.x86_64 27/32
  Verifying : blas-devel-3.9.0-10.el9.x86_64 28/32
  Verifying : lapack-devel-3.9.0-10.el9.i686 29/32
  Verifying : blas-devel-3.9.0-10.el9.i686 30/32
  Verifying : lapack64-3.9.0-10.el9.x86_64 31/32
  Verifying : blas64-3.9.0-10.el9.x86_64 32/32

Installed:
  blas-3.9.0-10.el9.i686
  blas-3.9.0-10.el9.x86_64
  blas-devel-3.9.0-10.el9.i686
  blas-devel-3.9.0-10.el9.x86_64
  blas64-3.9.0-10.el9.x86_64
  blas64_-3.9.0-10.el9.x86_64
  cpp-11.4.1-2.1.el9.x86_64
  gcc-11.4.1-2.1.el9.x86_64
  gcc-gfortran-11.4.1-2.1.el9.x86_64
  glibc-2.34-83.el9.7.i686
  glibc-devel-2.34-83.el9.7.i686
  glibc-devel-2.34-83.el9.7.x86_64
  glibc-gconv-extra-2.34-83.el9.7.i686
  glibc-headers-2.34-83.el9.7.x86_64
  kernel-headers-5.14.0-362.13.1.el9_3.x86_64
  lapack-3.9.0-10.el9.i686
  lapack-3.9.0-10.el9.x86_64
  lapack-devel-3.9.0-10.el9.i686
  lapack-devel-3.9.0-10.el9.x86_64
  lapack64-3.9.0-10.el9.x86_64
  lapack64_-3.9.0-10.el9.x86_64
  libgcc-11.4.1-2.1.el9.i686
  libgfortran-11.4.1-2.1.el9.i686
  libgfortran-11.4.1-2.1.el9.x86_64
  libmpc-1.2.1-4.el9.x86_64
  libquadmath-11.4.1-2.1.el9.i686
  libquadmath-11.4.1-2.1.el9.x86_64
  libquadmath-devel-11.4.1-2.1.el9.x86_64
  libxcrypt-4.4.18-3.el9.i686
  libxcrypt-devel-4.4.18-3.el9.i686
  libxcrypt-devel-4.4.18-3.el9.x86_64
  make-1:4.3-7.el9.x86_64

Complete!
[root@awx ~]# cat > main.f90 <<EOF
program bench

  implicit none

  real(8), allocatable :: mat(:,:), a(:,:)

  integer :: n

  n = 100

  ! Create random positive definite matrices
  write(6,'(3x,a,i8)') "I will create matrices of size",n

  allocate(a(n,n), mat(n,n))

  call random_number(a)

  call dgemm('n', 't', n, n, n, 1.d0, a, n, a, n, 0.d0, mat, n)

end program
EOF

# Compiling for 32 bit (i686) only
[root@awx ~]# gfortran -m32 -fallow-argument-mismatch main.f90 -lblas -llapack -o a1.out
[root@awx ~]# ./a1.out
   I will create matrices of size 100
[root@awx ~]# ldd a1.out
        linux-gate.so.1 (0xf7efe000)
        libblas.so.3 => /lib/libblas.so.3 (0xf7e7b000)
        liblapack.so.3 => /lib/liblapack.so.3 (0xf7600000)
        libgfortran.so.5 => /lib/libgfortran.so.5 (0xf7200000)
        libm.so.6 => /lib/libm.so.6 (0xf7db0000)
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0xf7d92000)
        libquadmath.so.0 => /lib/libquadmath.so.0 (0xf7d1b000)
        libc.so.6 => /lib/libc.so.6 (0xf7009000)
        /lib/ld-linux.so.2 (0xf7f00000)

# Compiling for 64 bit (x86_64) only
[root@awx ~]# gfortran -fallow-argument-mismatch main.f90 -lblas -llapack -o a2.out
[root@awx ~]# ./a2.out
   I will create matrices of size 100

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7f6db4423860 in ???
#1 0x7f6db4422a05 in ???
#2 0x7f6db4054daf in ???
#3 0x7f6db402b587 in ???
#4 0x7f6db4fbd664 in ???
#5 0x4016fc in ???
#6 0x401740 in ???
#7 0x7f6db403feaf in ???
#8 0x7f6db403ff5f in ???
#9 0x401114 in ???
#10 0xffffffffffffffff in ???
Segmentation fault (core dumped)
[root@awx ~]# ldd a2.out
        linux-vdso.so.1 (0x00007fff8dd80000)
        libblas.so.3 => /lib64/libblas.so.3 (0x00007f66ec87d000)
        liblapack.so.3 => /lib64/liblapack.so.3 (0x00007f66ec000000)
        libgfortran.so.5 => /lib64/libgfortran.so.5 (0x00007f66ebc00000)
        libm.so.6 => /lib64/libm.so.6 (0x00007f66ec7a2000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f66ec787000)
        libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007f66ec73d000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f66eb800000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f66ec90e000)

# Fails also when asking for lapack64 and blas64
[root@awx ~]# gfortran -fallow-argument-mismatch main.f90 -lblas64 -llapack64 -o a3.out
[root@awx ~]# ./a3.out
   I will create matrices of size 100

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7f6183a23860 in ???
#1 0x7f6183a22a05 in ???
#2 0x7f6183654daf in ???
#3 0x7f618362b587 in ???
#4 0x7f61845b1664 in ???
#5 0x4016fc in ???
#6 0x401740 in ???
#7 0x7f618363feaf in ???
#8 0x7f618363ff5f in ???
#9 0x401114 in ???
#10 0xffffffffffffffff in ???
Segmentation fault (core dumped)
[root@awx ~]# ldd a3.out
        linux-vdso.so.1 (0x00007ffc444e3000)
        libblas.so.3 => /lib64/libblas.so.3 (0x00007fac282f6000)
        liblapack.so.3 => /lib64/liblapack.so.3 (0x00007fac27a00000)
        libgfortran.so.5 => /lib64/libgfortran.so.5 (0x00007fac27600000)
        libm.so.6 => /lib64/libm.so.6 (0x00007fac2821b000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007fac28200000)
        libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007fac281b6000)
        libc.so.6 => /lib64/libc.so.6 (0x00007fac27200000)
        /lib64/ld-linux-x86-64.so.2 (0x00007fac28387000)

```

> [@smith](#):
>
> I think 9.3 is missing a 32 bit integer version of lapack

Several lapack packages are multilib and provide both x86\_64 and i686 packages, if this is what you mean. With that said, below is what changed with lapack. Left side is 9.2. Right side is 9.3. No packages were removed. (Also note the i686 packages mentioned here)

 ![image](https://us1.discourse-cdn.com/flex020/uploads/rockylinux/original/2X/a/ab1cddd4b9f913f155b014a43346e18d9512aeb7.png)

---

<div class="post-metadata">

**Author:** ![smith](https://avatars.discourse-cdn.com/v4/letter/s/f04885/32.png) [@smith](https://forums.rockylinux.org/u/smith)\
**Post date:** [January 19, 2024, 9:48am UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/7 "2024-01-19T09:48:44Z")

</div>

Thank you nazunalika for this detailed post. I think I will file a bug report, although there is a way to link to the correct lapack version. It behaves at least not as I expect it and also many codes we are using. Still, it is good to know that there is an additional way to make them run again for the mean time

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/flex020/uploads/rockylinux/original/1X/91b7219eec10e30013422e4df76c1d898711a5d5.svg) [@system](https://forums.rockylinux.org/u/system)\
**Post date:** [March 19, 2024, 9:49am UTC](https://forums.rockylinux.org/t/broken-lapack-blas-in-rocky-linux-9-3/12501/8 "2024-03-19T09:49:29Z")

</div>

Closed on March 19, 2024, 9:49am UTC

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.