Re: [patch] Fix linux-ia64 on SIGILL for deleted breakpoint

[Jan Kratochvil <jan.kratochvil@redhat.com>]

On Mon, 19 Jul 2010 11:04:26 +0200, Tristan Gingold wrote:
> On Jul 19, 2010, at 10:58 AM, Jan Kratochvil wrote:
> > in some cases ia64 can generate SIGILL instead of SIGTRAP.  Guessing it is
> > a CPU bug instead of Linux kernel bug (but I may be wrong).
> > 	https://bugzilla.redhat.com/show_bug.cgi?id=615538
> > 	http://sources.redhat.com/cgi-bin/cvsweb.cgi/~checkout~/tests/ptrace-tests/tests/ia64-sigill.c?cvsroot=systemtap
> 
> This is not a bug but a feature: the immediate of break.b is ignored by the processor.

You are right, Intel IA64 manual says:
	For the b_unit_form, imm21 is ignored and the value zero is placed in
	the Interruption Immediate control register (IIM).

And kernel/arch/ia64/kernel/traps.c ia64_bad_break():
              case 0: /* unknown error (used by GCC for __builtin_abort()) */
		[...]
                sig = SIGILL; code = ILL_ILLOPC;

And it really reports SIGILL even for example for:
...0:       10 00 00 00 01 00       [MIB]       nop.m 0x0
...6:       00 00 00 00 00 00                   break.i 0x0
...c:       00 00 00 20                         nop.b 0x0

Thanks for the explanation (embarassingly I did read the IA64 manual first).

(The GDB patch part meaning has not changed by this mail.)


Jan