[Qemu-devel] Re: [PATCH] Resend: x86: Reboot CPU on triple fault

[Jan Kiszka]

Alexander Graf wrote:
> 
> 
> 
> 
> On 12.01.2009, at 13:14, Jan Kiszka <address@hidden> wrote:
> 
>> address@hidden wrote:
>>> This is a (slightly adjusted for 2009-01-04 SVN) resend of Jan Kiszka's
>>> Reboot CPU on triple fault patch (see patch file for the exact
>>> reference)
>>>
>>> It seems like a consensus was reached on how to deal with tripple
>>> faults,
>>> but noone commited the last version (8) of the patch anyways.
>>>
>>> Just for the record -- 386BSD relies on this behavior to reset the
>>> CPU --
>>> it unmaps the whole address space in order to trigger a tripple fault.
>>>
>>
>> Good that you picked this up! It is still on my to-do list to get this
>> in, but with medium prio. However, let's try to push it a bit.
>>
>>> This is a slightly adjusted (for 2009-01-04 SVN) "reset on tripple
>>> fault patch"
>>>
>>> Originally from:
>>>
>>> Subject: [Qemu-devel] Re: [PATCH] x86: Reboot CPU on triple fault -
>>> Version 8
>>> Message-ID: <address@hidden>
>>> Date: Tue, 27 May 2008 18:17:18 +0200
>>> From: Jan Kiszka <address@hidden>
>>
>> Note that I posted an enhanced version on 2008-09-02, also covering
>> reset logging for non-x86 archs. Please use that one.
>>
>> ...
>>> Index: target-i386/op_helper.c
>>> ===================================================================
>>> --- target-i386/op_helper.c    (revision 6159)
>>> +++ target-i386/op_helper.c    (working copy)
>>> @@ -1244,6 +1244,9 @@
>>>     }
>>> }
>>>
>>> +/* This should come from sysemu.h - if we could include it here... */
>>> +void qemu_system_reset_request(void);
>>> +
>>> /*
>>>  * Check nested exceptions and change to double or triple fault if
>>>  * needed. It should only be called, if this is not an interrupt.
>>> @@ -1261,9 +1264,19 @@
>>>         fprintf(logfile, "check_exception old: 0x%x new 0x%x\n",
>>>                 env->old_exception, intno);
>>>
>>> -    if (env->old_exception == EXCP08_DBLE)
>>> -        cpu_abort(env, "triple fault");
>>> +#if !defined(CONFIG_USER_ONLY)
>>> +    if (env->old_exception == EXCP08_DBLE) {
>>> +        if (env->intercept)
>>> +            helper_vmexit(SVM_EXIT_SHUTDOWN, 0);
>>>
>>> +        if (loglevel & CPU_LOG_RESET)
>>> +            fprintf(logfile, "Triple fault\n");
>>> +
>>> +        qemu_system_reset_request();
>>> +        return EXCP_HLT;
>>> +    }
>>> +#endif
>>> +
>>>     if ((first_contributory && second_contributory)
>>>         || (env->old_exception == EXCP0E_PAGE &&
>>>             (second_contributory || (intno == EXCP0E_PAGE)))) {
>>
>> I meanwhile think that SVM hook should rather look like this
>>
>> helper_svm_check_intercept_param(SVM_EXIT_SHUTDOWN, 0);
>>
>>
>> in order to properly check if shutdown events are actually intercepted.
>> Alexander, am I right?
> 
> Yes, sounds right. Any reason not to put the intercept in reset_request?
> (asking blindly, I don't have access to the qemu source right now)

You mean qemu_system_reset_request? That's generic code while the hook
is x86-specific.

Jan

-- 
Siemens AG, Corporate Technology, CT SE 26
Corporate Competence Center Embedded Linux