From mboxrd@z Thu Jan 1 00:00:00 1970 From: Anthony Liguori Subject: Re: [Qemu-devel] [PATCH V6 3/4] qmp, nmi: convert do_inject_nmi() to QObject Date: Thu, 24 Feb 2011 11:12:29 -0600 Message-ID: <4D66917D.2030206@linux.vnet.ibm.com> References: <4D412AD2.1090509@cn.fujitsu.com> <4D655F32.30307@codemonkey.ws> <4D666DE5.9090800@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Lai Jiangshan , Lai Jiangshan , kvm@vger.kernel.org, qemu-devel@nongnu.org, Luiz Capitulino , Avi Kivity To: Markus Armbruster Return-path: Received: from e3.ny.us.ibm.com ([32.97.182.143]:46333 "EHLO e3.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754667Ab1BXRMT (ORCPT ); Thu, 24 Feb 2011 12:12:19 -0500 Received: from d01relay05.pok.ibm.com (d01relay05.pok.ibm.com [9.56.227.237]) by e3.ny.us.ibm.com (8.14.4/8.13.1) with ESMTP id p1OGqPkf009626 for ; Thu, 24 Feb 2011 11:52:25 -0500 Received: from d01av01.pok.ibm.com (d01av01.pok.ibm.com [9.56.224.215]) by d01relay05.pok.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id p1OHCICB147456 for ; Thu, 24 Feb 2011 12:12:18 -0500 Received: from d01av01.pok.ibm.com (loopback [127.0.0.1]) by d01av01.pok.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id p1OHCG7Q011190 for ; Thu, 24 Feb 2011 12:12:18 -0500 In-Reply-To: Sender: kvm-owner@vger.kernel.org List-ID: On 02/24/2011 10:20 AM, Markus Armbruster wrote: > Anthony Liguori writes: > > >> On 02/24/2011 02:33 AM, Markus Armbruster wrote: >> >>> Anthony Liguori writes: >>> >>> >>> >>>> On 01/27/2011 02:20 AM, Lai Jiangshan wrote: >>>> >>>> >>>>> Make we can inject NMI via qemu-monitor-protocol. >>>>> We use "inject-nmi" for the qmp command name, the meaning is clearer. >>>>> >>>>> Signed-off-by: Lai Jiangshan >>>>> --- >>>>> diff --git a/hmp-commands.hx b/hmp-commands.hx >>>>> index ec1a4db..e763bf9 100644 >>>>> --- a/hmp-commands.hx >>>>> +++ b/hmp-commands.hx >>>>> @@ -725,7 +725,8 @@ ETEXI >>>>> .params = "[cpu]", >>>>> .help = "Inject an NMI on all CPUs if no argument is given, " >>>>> "otherwise inject it on the specified CPU", >>>>> - .mhandler.cmd = do_inject_nmi, >>>>> + .user_print = monitor_user_noop, >>>>> + .mhandler.cmd_new = do_inject_nmi, >>>>> }, >>>>> #endif >>>>> STEXI >>>>> diff --git a/monitor.c b/monitor.c >>>>> index 387b020..1b1c0ba 100644 >>>>> --- a/monitor.c >>>>> +++ b/monitor.c >>>>> @@ -2542,7 +2542,7 @@ static void do_wav_capture(Monitor *mon, const QDict *qdict) >>>>> #endif >>>>> >>>>> #if defined(TARGET_I386) >>>>> -static void do_inject_nmi(Monitor *mon, const QDict *qdict) >>>>> +static int do_inject_nmi(Monitor *mon, const QDict *qdict, QObject **ret_data) >>>>> { >>>>> CPUState *env; >>>>> int cpu_index; >>>>> @@ -2550,7 +2550,7 @@ static void do_inject_nmi(Monitor *mon, const QDict *qdict) >>>>> if (!qdict_haskey(qdict, "cpu-index")) { >>>>> for (env = first_cpu; env != NULL; env = env->next_cpu) >>>>> cpu_interrupt(env, CPU_INTERRUPT_NMI); >>>>> - return; >>>>> + return 0; >>>>> } >>>>> >>>>> cpu_index = qdict_get_int(qdict, "cpu-index"); >>>>> @@ -2560,8 +2560,10 @@ static void do_inject_nmi(Monitor *mon, const QDict *qdict) >>>>> kvm_inject_interrupt(env, CPU_INTERRUPT_NMI); >>>>> else >>>>> cpu_interrupt(env, CPU_INTERRUPT_NMI); >>>>> - break; >>>>> + return 0; >>>>> } >>>>> + >>>>> + return -1; >>>>> } >>>>> #endif >>>>> >>>>> diff --git a/qmp-commands.hx b/qmp-commands.hx >>>>> index 56c4d8b..a887dd5 100644 >>>>> --- a/qmp-commands.hx >>>>> +++ b/qmp-commands.hx >>>>> @@ -429,6 +429,34 @@ Example: >>>>> >>>>> EQMP >>>>> >>>>> +#if defined(TARGET_I386) >>>>> + { >>>>> + .name = "inject-nmi", >>>>> + .args_type = "cpu-index:i?", >>>>> + .params = "[cpu]", >>>>> + .help = "Inject an NMI on all CPUs if no argument is given, " >>>>> + "otherwise inject it on the specified CPU", >>>>> + .user_print = monitor_user_noop, >>>>> + .mhandler.cmd_new = do_inject_nmi, >>>>> + }, >>>>> +#endif >>>>> +SQMP >>>>> +inject-nmi >>>>> +---------- >>>>> + >>>>> +Inject an NMI on all CPUs or the given CPU (x86 only). >>>>> + >>>>> +Arguments: >>>>> + >>>>> +- "cpu-index": the index of the CPU to be injected NMI (json-int, optional) >>>>> + >>>>> +Example: >>>>> + >>>>> +-> { "execute": "inject-nmi", "arguments": { "cpu-index": 0 } } >>>>> +<- { "return": {} } >>>>> >>>>> >>>>> >>>> Please describe all expected errors. >>>> >>>> >>> Quoting qmp-commands.hx: >>> >>> 3. Errors, in special, are not documented. Applications should NOT check >>> for specific errors classes or data (it's strongly recommended to only >>> check for the "error" key) >>> >>> Indeed, not a single error is documented there. This is intentional. >>> >>> >> Yeah, but we're not 0.14 anymore and for 0.15, we need to document >> errors. If you are suggesting I send a patch to remove that section, >> I'm more than happy to. >> > Two separate issues here: 1. Are we ready to commit to the current > design of errors, and 2. Is it fair to reject Lai's patch now because he > doesn't document his errors. > > I'm not commenting on 1. here. > > Regarding 2.: rejecting a patch because it doesn't document an aspect > that current master intentionally leaves undocumented is not how you > treat contributors. At least not if you want any other than certified > masochists who enjoy pain, and professionals who get adequately > compensated for it. > > Lead by example, not by fiat. > http://repo.or.cz/w/qemu/aliguori.git/blob/refs/heads/glib:/qmp-schema.json I am in the process of documenting the errors of every command. It's a royal pain but I'm going to document everything we have right now. It's actually the last bit of work I need to finish before sending QAPI out. So for new commands being added, it would be hugely helpful for the authors to document the errors such that I don't have to reverse engineer all of the possible error conditions. >>> Once we have an error design in place that has a reasonable hope to >>> stand the test of time, and have errors documented for at least some of >>> the commands here, we can start to require proper error documentation >>> for new commands. But not now. >>> >>> >> I'm quite happy with the error design we have today. The only problem >> is that we don't propagate errors in a sane way but I've got that all >> but fixed in my qapi tree. >> > I don't think error propagation is the only problem we have with QError. > > QError makes it way too hard to emit error messages fit for human > consumption. The consequence is that we get errors unfit for humans. > Take a look at http://repo.or.cz/w/qemu/aliguori.git/blob/refs/heads/glib:/hmp.c#l64 Automatically generating an error string based on an error and no additional context is problematic. Given the context of the caller and the right fields in the error, very high quality error messages can be constructed. Regards, Anthony Liguori >>>> Don't hide this command for >>>> !defined(TARGET_I386), instead have it throw an error in the >>>> implementation. >>>> >>>> >>> Works for me. >>> >>> >>> >>>> Don't have commands that multiple behavior based on the presence or >>>> absence of arguments. Make it take a list of cpus if you want the >>>> ability to inject the NMI to more than one CPU. >>>> >>>> >>> Having optional arguments is fine. It's good taste to give them >>> "default semantics", i.e. "no argument" is shorthand for one specific >>> argument value. >>> >>> Luiz already pointed to the thread where we discussed this command >>> before. Executive summary: >>> >>> * Real hardware's NMI button injects all CPUs. This is the primary use >>> case. >>> >>> * Lai said injecting a single CPU can be useful for debugging. Was >>> deemed acceptable as secondary use case. >>> >>> Lai also pointed out that the human monitor's nmi command injects a >>> single CPU. That was dismissed as irrelevant for QMP. >>> >>> * No other use cases have been presented. >>> >>> Therefore, the "list of CPUs" idea was shot down as overly general. >>> >>> >> That's fine, then we should do two commands. Think of it from the >> perspective of the client. This appears as: >> >> in C: >> >> qmp_inject_nmi(sess, false, 0,&err); >> >> in Python: >> >> sess.inject_nmi() >> >> The first example doesn't tell you at all what's happening. The >> second API does look really nice until you see the following later: >> >> sess.inject_nmi(0) >> >> What's the difference between these two functions? You might say this >> is bad form and that an explicit named argument should be given but I >> wouldn't count on it. >> >> Having two commands, nmi_inject and nmi_inject_on_cpu, would result in >> a much more readable API in both C and Python: >> >> in C: >> >> qmp_nmi_inject(sess,&err); >> qmp_nmi_inject_on_cpu(sess, 3,&err); >> >> in Python: >> >> sess.nmi_inject() >> sess.nmi_inject_on_cpu(3) >> > I'm fine with Lai's patch. I'd be just as fine with Lai's patch less > the "inject on one CPU" feature. And I'd be fine with your interface as > well. > > Lai has been trying to get this simple feature in for almost three > months. And we're still here, debating which way he should change his > arguments, and whether we should require him to document his errors > before anyone else. For crying out loud, what's wrong with us? >