OS-X: Illegal instruction: 4

[rolkar]

Got this mail:

Hello,
Many thanks for providing this tool!
I have limited knowledge of command line...
Trying to use x3f_extract on OS X 10.7.5. Is this why I receive "Illegal instruction: 4" when trying to
execute "x3f_extract -tiff test.x3f"?
Thanks for help with this.
Regards,
Sean Lanyi
California, USA

I answered this:

Searching, I found this
http://stackoverflow.com/questions/14268887/what-is-the-illegal-instruction-4-error-and-why-does-mmacosx-version-min-10
So, it seems we have compiled the code with a too new compiler, or ... you are running a too old OS-X.
Whichever way you want to see it :)

[erikrk]

SIGILL can occur for different reasons. In essence, it comes down to two different categories.

1. The code generated by the compiler is incompatible with the CPU.
2. Jumping to a memory location which contains something else than code.

This seems to be difficult to debug without more information and/or the possibility to test/reproduce.

[rolkar]

The problem described in the link matches quite well the actual situation.

It is also quite easy to check. Just recompile the executable with the
suggested switch
and send the executable to the guy that had problems.

What is more problematic is if we shall fix this or not. Or if we
shall assume people upgrades their OS instead. Is it a good idea
to compile with switches generating old code? How can we know
how old the code have to be?

/Roland

On 2015-08-22 01:28, Erik Karlsson wrote:

SIGILL can occur for different reasons. In essence, it comes down to
two different categories.

1. The code generated by the compiler is incompatible with the CPU.
2. Jumping to a memory location which contains something else than code.

This seems to be difficult to debug without more information and/or
the possibility to test/reproduce.

—
Reply to this email directly or view it on GitHub
#76 (comment).

[erikrk]

The code is already being compiled with -mmacosx-version-min=10.7 and he
is using 10.7.5, so it should in principle work.

On sáb, 2015-08-22 at 01:52 -0700, Roland Karlsson wrote:

The problem described in the link matches quite well the actual
situation.

It is also quite easy to check. Just recompile the executable with
the
suggested switch
and send the executable to the guy that had problems.

What is more problematic is if we shall fix this or not. Or if we
shall assume people upgrades their OS instead. Is it a good idea
to compile with switches generating old code? How can we know
how old the code have to be?

/Roland

On 2015-08-22 01:28, Erik Karlsson wrote:

SIGILL can occur for different reasons. In essence, it comes down
to
two different categories.

1. The code generated by the compiler is incompatible with the CPU.
2. Jumping to a memory location which contains something else than
   code.

This seems to be difficult to debug without more information and/or
the possibility to test/reproduce.

—
Reply to this email directly or view it on GitHub
#76 (comment).

—
Reply to this email directly or view it on GitHub.