Re: [commit, s390] Fix regression in gdb.base/callfuncs.exp [Re: [commit, s390] Proper handling of PSW address/mask everywhere]

["Ulrich Weigand" <uweigand@de.ibm.com>]

Jan Kratochvil wrote:
> On Mon, 19 Dec 2011 22:49:16 +0100, Ulrich Weigand wrote:
> > Tested with no regression (fixing the callfuncs.exp one) on s390x (-m31,-m64).
> > Committed to mainline and 7.4 branch.
> 
> Confirming it works, thanks.
> 
> There are yet some regressions of sigstep.exp, I will check them more:

Hmm, I'm not seeing this particular symptom you mention below.

However, I did see failures on old kernels in 31-bit compat mode only;
it turned out these were caused by a kernel bug that is exposed by my
GDB change.

When the kernel sets up the registers for calling a signal handler,
it is supposed to place the return address including the addressing
mode bit into register %r14.  This does in fact happen in native
31-bit mode.  However, in 31-bit compat mode under a 64-bit kernel,
the addressing mode bit was not set.  This has been fixed at least
for regular signal handlers in current git (note that RT signals
are still broken).

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com