On 14.07.2021 22:37, Bobby Eshleman wrote: > --- a/xen/arch/arm/traps.c > +++ b/xen/arch/arm/traps.c > @@ -16,6 +16,7 @@ > * GNU General Public License for more details. > */ > > +#include <xen/debugger.h>
I don't think this is needed here; instead I think ... > @@ -41,7 +42,6 @@ > #include <asm/acpi.h> > #include <asm/cpuerrata.h> > #include <asm/cpufeature.h> > -#include <asm/debugger.h> ... this wants to be done in patch 1 already. > --- a/xen/common/keyhandler.c > +++ b/xen/common/keyhandler.c > @@ -3,6 +3,7 @@ > */ > > #include <asm/regs.h> > +#include <xen/debugger.h> > #include <xen/delay.h> > #include <xen/keyhandler.h> > #include <xen/param.h> > @@ -20,7 +21,6 @@ > #include <xen/mm.h> > #include <xen/watchdog.h> > #include <xen/init.h> > -#include <asm/debugger.h> > #include <asm/div64.h> Not sure about this - as indicated maybe the code needing the include instead wants to move to x86'es new debugger.c. > --- /dev/null > +++ b/xen/include/xen/debugger.h > @@ -0,0 +1,69 @@ > +/****************************************************************************** > + * Generic hooks into arch-dependent Xen. > + * > + * This program is distributed in the hope that it will be useful, > + * but WITHOUT ANY WARRANTY; without even the implied warranty of > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > + * GNU General Public License for more details. > + * > + * You should have received a copy of the GNU General Public License > + * along with this program; If not, see <http://www.gnu.org/licenses/>. > + * > + * Each debugger should define three functions here: > + * > + * 1. debugger_trap_entry(): > + * Called at start of any synchronous fault or trap, before any other work > + * is done. The idea is that if your debugger deliberately caused the trap > + * (e.g. to implement breakpoints or data watchpoints) then you can take > + * appropriate action and return a non-zero value to cause early exit from > + * the trap function. > + * > + * 2. debugger_trap_fatal(): > + * Called when Xen is about to give up and crash. Typically you will use > this > + * hook to drop into a debug session. It can also be used to hook off > + * deliberately caused traps (which you then handle and return non-zero). > + * > + * 3. debugger_trap_immediate(): > + * Called if we want to drop into a debugger now. This is essentially the > + * same as debugger_trap_fatal, except that we use the current register > state > + * rather than the state which was in effect when we took the trap. > + * For example: if we're dying because of an unhandled exception, we call > + * debugger_trap_fatal; if we're dying because of a panic() we call > + * debugger_trap_immediate(). > + */ > + > +#ifndef __XEN_DEBUGGER_H__ > +#define __XEN_DEBUGGER_H__ > + > +#ifdef CONFIG_CRASH_DEBUG > + > +#include <asm/debugger.h> > + > +#else > + > +#include <asm/regs.h> > +#include <asm/processor.h> I don't think you need either of these for the stubs below. > +static inline void domain_pause_for_debugger(void) > +{ > +} > + > +static inline bool debugger_trap_fatal( > + unsigned int vector, const struct cpu_user_regs *regs) > +{ > + return false; > +} > + > +static inline void debugger_trap_immediate(void) > +{ > +} > + > +static inline bool debugger_trap_entry( > + unsigned int vector, const struct cpu_user_regs *regs) > +{ > + return false; > +} Of these stubs, after patch 1 only debugger_trap_immediate() is needed outside of arch/x86/. Perhaps everything else wants to remain in x86'es per-arch header? Jan
