wavpack core dumps doing md5

Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

wavpack core dumps doing md5

Graham Stephens-2
I don't know if this is machine-specific as I only have the one sparc
machine on OpenBSD at the moment (blade 100).
When running wavpack with the -m option (for md5 calculation), I get a
core dump with a message of 'Bus Error'.
It works ok for other options, and of course the standard md5 works ok
as well.

OBSD v6.1 with wavpack v5.1.0

Reply | Threaded
Open this post in threaded view
|

Re: wavpack core dumps doing md5

Christian Weisgerber
On 2017-05-12, Graham Stephens <[hidden email]> wrote:

> I don't know if this is machine-specific as I only have the one sparc
> machine on OpenBSD at the moment (blade 100).
> When running wavpack with the -m option (for md5 calculation), I get a
> core dump with a message of 'Bus Error'.

That is most likely an unaligned access error.

The usual approach is to build a package with debugging symbols
(CFLAGS="-g" INSTALL_STRIP="") and examine the core file with gdb.

Alas, I don't have any strict alignment machines at hand any longer.
I tried to reproduce it on aarch64, but that architecture (or at
least the CPU I have) doesn't seem to actually enforce alignment
despite the #define __STRICT_ALIGNMENT in <machine/endian.h>.

--
Christian "naddy" Weisgerber                          [hidden email]