Re: [Qemu-devel] PATCH: allow Sparc hosts to run arm/mips/sparc-softmmu

[Ben Taylor <sol10x86@cox.net>]

---- Rob Landley <rob@landley.net> wrote: 
> On Tuesday 13 March 2007 10:25 am, Ben Taylor wrote:
> > However, it's very wax-on, wax-off kind of thing.  Without the patch,
> > arm-test and mips-test crash.  With the patch, I can run both tests.
> 
> Could we get a reproduction sequence for the crashes please?

Running 0.9.0-CVS plus a few patches, on a Solaris 10/FCS 
heavily patched host, compiled with blastwave gcc-3.4.3.

When I run the arm-test as suggested by the readme, qemu
core dumps with a SIGSEGV at the line that was patched.  I agree
with Paul that is probalby not the right fix.  It does just happen to
fix it on both mips-test and arm-test.

Uncommmenting the define for DEBUG_EXEC in cpu-exec.c 
and recompiling, and starting the arm-test with "-vnc :0 -S", 
connecting with VNC, going to the qemu monitor and enabling 
all logging and hitting continue, the qemu again dumps core, 
having processed exactly one translation block.   With the patch,
and started exactly the same way, the first TB log looks exactly
the same, but continues without "crashing".

If I put a fprintf(logfile...) above the line, it works, no crash.

I am not very good at debugging as it is, and I'm pretty sure that there's
some sort of left over from the TB that is causing the problem.  I'm
suspecting maybe a page boundry, just from having stepi'd all the
instructions in the TB before it SEGV's on that line.  The last address
looked suspciously like a page boundry.  However, that's just an
uneducated SWAG at what the problem is.

Does that help?

Ben