On 25.09.2015 14:34, Sergey Fedorov wrote:
> On 18.09.2015 17:07, Sergey Fedorov wrote:
>> On 18.09.2015 16:50, Peter Maydell wrote:
>>> On 14 September 2015 at 11:51, Sergey Fedorov <serge.f...@gmail.com> wrote:
>>> @@ -904,7 +913,16 @@ void arm_debug_excp_handler(CPUState *cs)
>>> arm_debug_target_el(env));
>>> }
>>> } else {
>>> - if (check_breakpoints(cpu)) {
>>> + CPUBreakpoint *bp;
>>> + uint64_t pc = is_a64(env) ? env->pc : env->regs[15];
>>> +
>>> + QTAILQ_FOREACH(bp, &cs->breakpoints, entry) {
>>> + if (bp->pc == pc && !(bp->flags & BP_CPU)) {
>>> + return;
>>> + }
>>> + }
>>> This extra code looks right, but isn't it fixing a different bug?
>> You are right, it would better come to separate patch.
> Actually, I can't think of it as a separate patch. This change is really
> required only if we remove check_breakpoints() here. Otherwise
> check_breakpoints() calls bp_wp_matches() which do the necessary check.
>
...but considering the order of breakpoint enumeration it is not so
simple. The difference is when we have GDB and CPU breakpoint to the
same address. In this case check_breakpoints() returns true, but we
should handle GDB breakpoints first. Sorry for my misunderstanding, I
will split this patch as you suggested.
Best regards,
Sergey
