Hello all,

After a hiatus from working with MARSS, I've come back to it and run into a
bit of an issue. One of the things we are looking to do is to run SPEC06. Up
until now I was using the MARSS-supplied parsec disk images, but now I've
created my own image, but I'm running into problems.

I installed Ubuntu 10.04 server AMD64 into a raw disk image (not qcow2, so I
can easily mount it on loopback and modify it without having to fire up the
VM). I installed Ubuntu into the disk image using qemu-0.12.

The problem that I'm running up against is that using the master branch of
MARSS, I'm getting a segmentation fault when trying to run 999.specrand
(both directly and using runspec). GDB is indicating that the fault is at:

#21278 default_ioport_readw (opaque=<value optimized out>, address=114688)
at qemu/vl.c:330
#21279 0x000000000040f1b8 in ioport_read (opaque=<value optimized out>,
address=114688) at qemu/vl.c:295

but the backtrace function seems to be getting confused because it seems
that the backtrace is infinite... I looked at the git logs and it doesn't
appear that this code is modified by you guys.

If I run specrand without using start_sim, it runs fine, but when I use
start_sim, I get about 1.6M cycles in and it segfaults.

I am going to try to copy just the specrand binary into your parsec image
and see if I can run it there, but in the meantime I was just wondering if
there were some kind of special precautions I had to take when creating disk
images ... ?

Thanks,
Paul
_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to