Experimental Physics and
Industrial Control System

Guy Jennings <[email protected]> · Thu, 6 Aug 2009 11:35:21 -0500

As an aside, if you build epics using the rpm build system this kind

of problem is handled automatically

by the package dependency resolution system.

On Aug 6, 2009, at 10:01 AM, Mark Rivers wrote:

Thanks for the online and offline replies.

My original post had an error: the older Linux system was 32-bit,

not 64-bit.

The problem was missing 32-bit libraries on the new 64-bit Fedora

Core 10 system. The error message certainly could have been

clearer than "Command not found", that's for sure!

Once I installed the following packages I was able to run 32-bit

binaries that were already built on the other Linux system, as well

as run "make" to build 32-bit linux-x86 binaries on the new 64-bit

system.

   116  9:45    sudo yum install glibc-devel.i386
   124  9:48    sudo yum install readline-devel.i386
   127  9:49    sudo yum install ncurses-devel.i386
   130  9:50    sudo yum install libstdc++-devel.i386

Thanks,
Mark

________________________________

From: ZY (Zhijian Yin) [mailto:[email protected]]
Sent: Thu 8/6/2009 9:31 AM
To: Mark Rivers
Cc: [email protected]
Subject: Re: Linux question

Mark,

I think you need to install the i386/i686 compatibility libraries

on the

64bit Linux machine. Do you have library files (libreadline.so.5, for
example) in /usr/lib32 or /lib32?

It's probably easier to compile a separate binary for the 64bit Linux.

ZY

Mark Rivers wrote:

Folks,

This is not exactly an EPICS-specific question, but I'm having

trouble running EPICS binaries built on one recent Linux system on

another recent Linux system. It just gives a "command not found"

error, which is not very helpful. Both systems are Fedora 64-bit;

the binary is built with HOST_ARCH=linux-x86, e.g. 32-bit.

Here is an example. I am trying to run a soft IOC in the current

directly. First I run "file" and "ldd" on it, then try to run it:

[epics@localhost linux-x86]$ file ./simDetectorApp

./simDetectorApp: ELF 32-bit LSB executable, Intel 80386, version

1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux

2.6.9, not stripped

[epics@localhost linux-x86]$ ldd ./simDetectorApp
        not a dynamic executable

[epics@localhost linux-x86]$ ./simDetectorApp
./simDetectorApp: Command not found.

Note that ldd says this is not a dynamic executable, which is

incorrect.

Here is the version information on the system where the binary was

built and where the executable resides:

corvette:1-5beta/ADApp/simDetectorSrc>more /proc/version

Linux version 2.6.27.21-78.2.41.fc9.i686 (mockbuild@) (gcc version

4.3.0 20080428 (Red Hat 4.3.0-8) (GCC) ) #1 SMP Mon Mar 23

23:45:58 EDT 2009

Here is the version information on the system that will not run

the binary:

[epics@localhost linux-x86]$ more /proc/version

Linux version 2.6.27.25-170.2.72.fc10.x86_64

([email protected]) (gcc version 4.3.2

20081105 (Red Hat 4.3.2-7) (GCC) ) #1 SMP Sun Jun 21 18:39:34 EDT

2009

If I run "file" and "ldd" on the file on the system that builds it

and runs it OK I get the following:

corvette:1-5beta/bin/linux-x86>file simDetectorApp

simDetectorApp: ELF 32-bit LSB executable, Intel 80386, version 1

(SYSV), dynamically linked (uses shared libs), for GNU/Linux

2.6.9, not stripped

corvette:1-5beta/bin/linux-x86>ldd simDetectorApp
        linux-gate.so.1 =>  (0x0035b000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x003b3000)
        libreadline.so.5 => /lib/libreadline.so.5 (0x05ac2000)
        libncurses.so.5 => /lib/libncurses.so.5 (0x05f3b000)
        librt.so.1 => /lib/librt.so.1 (0x004c8000)
        libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x066c5000)
        libm.so.6 => /lib/libm.so.6 (0x00363000)
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x066b5000)
        libc.so.6 => /lib/libc.so.6 (0x004d2000)
        /lib/ld-linux.so.2 (0x001d8000)
        libtinfo.so.5 => /lib/libtinfo.so.5 (0x05c57000)
        libdl.so.2 => /lib/libdl.so.2 (0x0038e000)
corvette:1-5beta/bin/linux-x86>./simDetectorApp

Any idea what could be wrong?

Thanks,
Mark

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System