From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sean Christopherson Subject: Re: [RFC PATCH v6 01/92] kvm: introduce KVMI (VM introspection subsystem) Date: Tue, 13 Aug 2019 08:01:28 -0700 Message-ID: <20190813150128.GB13991@linux.intel.com> References: <20190809160047.8319-1-alazar@bitdefender.com> <20190809160047.8319-2-alazar@bitdefender.com> <20190812202030.GB1437@linux.intel.com> <5d52a5ae.1c69fb81.5c260.1573SMTPIN_ADDED_BROKEN@mx.google.com> <5fa6bd89-9d02-22cd-24a8-479abaa4f788@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 Return-path: Content-Disposition: inline In-Reply-To: <5fa6bd89-9d02-22cd-24a8-479abaa4f788@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: Paolo Bonzini Cc: Tamas K Lengyel , Weijiang Yang , Yu C , kvm@vger.kernel.org, Konrad Rzeszutek Wilk , Jan Kiszka , Samuel =?iso-8859-1?Q?Laur=E9n?= , Radim =?utf-8?B?S3LEjW3DocWZ?= , Zhang@vger.kernel.org, virtualization@lists.linux-foundation.org, Adalbert =?utf-8?B?TGF6xINy?= , linux-mm@kvack.org, Patrick Colp , Mathieu Tarral , Stefan Hajnoczi , Mircea =?iso-8859-1?Q?C=EErjaliu?= , Mihai =?utf-8?B?RG9uyJt1?= List-Id: virtualization@lists.linuxfoundation.org T24gVHVlLCBBdWcgMTMsIDIwMTkgYXQgMDI6MDk6NTFQTSArMDIwMCwgUGFvbG8gQm9uemluaSB3 cm90ZToKPiBPbiAxMy8wOC8xOSAxMzo1NywgQWRhbGJlcnQgTGF6xINyIHdyb3RlOgo+ID4+IFRo ZSByZWZjb3VudGluZyBhcHByb2FjaCBzZWVtcyBhIGJpdCBiYWNrd2FyZHMsIGFuZCBBRkFJQ1Qg aXMgZHJpdmVuIGJ5Cj4gPj4gaW1wbGVtZW50aW5nIHVuaG9vayB2aWEgYSBtZXNzYWdlLCB3aGlj aCBhbHNvIHNlZW1zIGJhY2t3YXJkcy4gIEkgYXNzdW1lCj4gPj4gaG9vayBhbmQgdW5ob29rIGFy ZSByZWxhdGl2ZWx5IHJhcmUgZXZlbnRzIGFuZCBub3QgcGVyZm9ybWFuY2UgY3JpdGljYWwsCj4g Pj4gc28gbWFrZSB0aG9zZSB0aGUgcmVzdHJpY3RlZC9zbG93IGZsb3dzLCBlLmcuIGZvcmNlIHVz ZXJzcGFjZSB0byBxdWllc2NlCj4gPj4gdGhlIFZNIGJ5IG1ha2luZyB1bmhvb2soKSBtdXR1YWxs eSBleGNsdXNpdmUgd2l0aCBldmVyeSB2Y3B1IGlvY3RsKCkgYW5kCj4gPj4gbWF5YmUgYW55dGhp bmcgdGhhdCB0YWtlcyBrdm0tPmxvY2suIAo+ID4+Cj4gPj4gVGhlbiBrdm1pX2lvY3RsX3VuaG9v aygpIGNhbiB1c2UgdGhyZWFkX3N0b3AoKSBhbmQga3ZtaV9yZWN2KCkganVzdCBuZWVkcwo+ID4+ IHRvIGNoZWNrIGt0aHJlYWRfc2hvdWxkX3N0b3AoKS4KPiA+Pgo+ID4+IFRoYXQgd2F5IGt2bWkg ZG9lc24ndCBuZWVkIHRvIGJlIHJlZmNvdW50ZWQgc2luY2UgaXQncyBndWFyYW50ZWVkIHRvIGJl Cj4gPj4gYWxpdmUgaWYgdGhlIHBvaW50ZXIgaXMgbm9uLW51bGwuICBFbGltaW5hdGluZyB0aGUg cmVmY291bnRpbmcgd2lsbCBjbGVhbgo+ID4+IHVwIGEgbG90IG9mIHRoZSBjb2RlIGJ5IGVsaW1p bmF0aW5nIGNhbGxzIHRvIGt2bWlfe2dldCxwdXR9KCksIGUuZy4KPiA+PiB3cmFwcGVycyBsaWtl IGt2bWlfYnJlYWtwb2ludF9ldmVudCgpIGp1c3QgY2hlY2sgdmNwdS0+a3ZtaSwgb3IgbWF5YmUK PiA+PiBldmVuIGdldCBkcm9wcGVkIGFsdG9nZXRoZXIuCj4gPiAKPiA+IFRoZSB1bmhvb2sgZXZl bnQgaGFzIGJlZW4gYWRkZWQgdG8gY292ZXIgdGhlIGZvbGxvd2luZyBjYXNlOiB3aGlsZSB0aGUK PiA+IGludHJvc3BlY3Rpb24gdG9vbCBydW5zIGluIGFub3RoZXIgVk0sIGJvdGggVk1zLCB0aGUg dmlydHVhbCBhcHBsaWFuY2UKPiA+IGFuZCB0aGUgaW50cm9zcGVjdGVkIFZNLCBjb3VsZCBiZSBw YXVzZWQgYnkgdGhlIHVzZXIuIFdlIG5lZWRlZCBhIHdheQo+ID4gdG8gc2lnbmFsIHRoaXMgdG8g dGhlIGludHJvc3BlY3Rpb24gdG9vbCBhbmQgZ2l2ZSBpdCB0aW1lIHRvIHVuaG9vawo+ID4gKHRo ZSBpbnRyb3NwZWN0ZWQgVk0gaGFzIHRvIHJ1biBhbmQgZXhlY3V0ZSB0aGUgaW50cm9zcGVjdGlv biBjb21tYW5kcwo+ID4gZHVyaW5nIHRoaXMgcGhhc2UpLiBUaGUgcmVjZWl2aW5nIHRocmVhZHMg cXVpdHMgd2hlbiB0aGUgc29ja2V0IGlzIGNsb3NlZAo+ID4gKGJ5IFFFTVUgb3IgYnkgdGhlIGlu dHJvc3BlY3Rpb24gdG9vbCkuCgpXaHkgZG9lcyBjbG9zaW5nIHRoZSBzb2NrZXQgcmVxdWlyZSBk ZXN0cm95aW5nIHRoZSBrdm1pIG9iamVjdD8gIEUuZy4gY2FuCml0IGJlIG1hcmtlZCBhcyBkZWZ1 bmN0IG9yIHdoYXRldmVyIGFuZCBvbmx5IGZ1bGx5IHJlbW92ZWQgb24gYSBzeW5jaHJvbm91cwp1 bmhvb2sgZnJvbSB1c2Vyc3BhY2U/ICBSZS1ob29raW5nIGNvdWxkIGVpdGhlciByZXF1aXJlIHNh aWQgdW5ob29rLCBvcgptYXliZSByZXVzZSB0aGUgZXhpc3Rpbmcga3ZtaSBvYmplY3Qgd2l0aCBh IG5ldyBzb2NrZXQuCgo+ID4gSXQncyBhIGJpdCB1bmNsZWFyIGhvdywgYnV0IHdlJ2xsIHRyeSB0 byBnZXQgcmlkZSBvZiB0aGUgcmVmY291bnQgb2JqZWN0LAo+ID4gd2hpY2ggd2lsbCByZW1vdmUg YSBsb3Qgb2YgY29kZSwgaW5kZWVkLgo+IAo+IFlvdSBjYW4ga2VlcCBpdCBmb3Igbm93LiAgSXQg bWF5IGJlY29tZSBjbGVhcmVyIGhvdyB0byBmaXggaXQgYWZ0ZXIgdGhlCj4gZXZlbnQgbG9vcCBp cyBjbGVhbmVkIHVwLgoKQnkgZXZlbnQgbG9vcCwgZG8geW91IG1lYW4gdGhlIHBlci12Q1BVIGpv YnMgbGlzdD8KX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18K VmlydHVhbGl6YXRpb24gbWFpbGluZyBsaXN0ClZpcnR1YWxpemF0aW9uQGxpc3RzLmxpbnV4LWZv dW5kYXRpb24ub3JnCmh0dHBzOi8vbGlzdHMubGludXhmb3VuZGF0aW9uLm9yZy9tYWlsbWFuL2xp c3RpbmZvL3ZpcnR1YWxpemF0aW9u From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.3 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 22D00C433FF for ; Tue, 13 Aug 2019 15:01:31 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id F1D75205F4 for ; Tue, 13 Aug 2019 15:01:30 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729840AbfHMPB3 (ORCPT ); Tue, 13 Aug 2019 11:01:29 -0400 Received: from mga06.intel.com ([134.134.136.31]:50964 "EHLO mga06.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729665AbfHMPB3 (ORCPT ); Tue, 13 Aug 2019 11:01:29 -0400 X-Amp-Result: UNKNOWN X-Amp-Original-Verdict: FILE UNKNOWN X-Amp-File-Uploaded: False Received: from orsmga005.jf.intel.com ([10.7.209.41]) by orsmga104.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 13 Aug 2019 08:01:28 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,381,1559545200"; d="scan'208";a="351556260" Received: from sjchrist-coffee.jf.intel.com (HELO linux.intel.com) ([10.54.74.41]) by orsmga005.jf.intel.com with ESMTP; 13 Aug 2019 08:01:28 -0700 Date: Tue, 13 Aug 2019 08:01:28 -0700 From: Sean Christopherson To: Paolo Bonzini Cc: Adalbert =?utf-8?B?TGF6xINy?= , kvm@vger.kernel.org, linux-mm@kvack.org, virtualization@lists.linux-foundation.org, Radim =?utf-8?B?S3LEjW3DocWZ?= , Konrad Rzeszutek Wilk , Tamas K Lengyel , Mathieu Tarral , Samuel =?iso-8859-1?Q?Laur=E9n?= , Patrick Colp , Jan Kiszka , Stefan Hajnoczi , Weijiang Yang , Zhang@vger.kernel.org, Yu C , Mihai =?utf-8?B?RG9uyJt1?= , Mircea =?iso-8859-1?Q?C=EErjaliu?= Subject: Re: [RFC PATCH v6 01/92] kvm: introduce KVMI (VM introspection subsystem) Message-ID: <20190813150128.GB13991@linux.intel.com> References: <20190809160047.8319-1-alazar@bitdefender.com> <20190809160047.8319-2-alazar@bitdefender.com> <20190812202030.GB1437@linux.intel.com> <5d52a5ae.1c69fb81.5c260.1573SMTPIN_ADDED_BROKEN@mx.google.com> <5fa6bd89-9d02-22cd-24a8-479abaa4f788@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <5fa6bd89-9d02-22cd-24a8-479abaa4f788@redhat.com> User-Agent: Mutt/1.5.24 (2015-08-30) Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org On Tue, Aug 13, 2019 at 02:09:51PM +0200, Paolo Bonzini wrote: > On 13/08/19 13:57, Adalbert Lazăr wrote: > >> The refcounting approach seems a bit backwards, and AFAICT is driven by > >> implementing unhook via a message, which also seems backwards. I assume > >> hook and unhook are relatively rare events and not performance critical, > >> so make those the restricted/slow flows, e.g. force userspace to quiesce > >> the VM by making unhook() mutually exclusive with every vcpu ioctl() and > >> maybe anything that takes kvm->lock. > >> > >> Then kvmi_ioctl_unhook() can use thread_stop() and kvmi_recv() just needs > >> to check kthread_should_stop(). > >> > >> That way kvmi doesn't need to be refcounted since it's guaranteed to be > >> alive if the pointer is non-null. Eliminating the refcounting will clean > >> up a lot of the code by eliminating calls to kvmi_{get,put}(), e.g. > >> wrappers like kvmi_breakpoint_event() just check vcpu->kvmi, or maybe > >> even get dropped altogether. > > > > The unhook event has been added to cover the following case: while the > > introspection tool runs in another VM, both VMs, the virtual appliance > > and the introspected VM, could be paused by the user. We needed a way > > to signal this to the introspection tool and give it time to unhook > > (the introspected VM has to run and execute the introspection commands > > during this phase). The receiving threads quits when the socket is closed > > (by QEMU or by the introspection tool). Why does closing the socket require destroying the kvmi object? E.g. can it be marked as defunct or whatever and only fully removed on a synchronous unhook from userspace? Re-hooking could either require said unhook, or maybe reuse the existing kvmi object with a new socket. > > It's a bit unclear how, but we'll try to get ride of the refcount object, > > which will remove a lot of code, indeed. > > You can keep it for now. It may become clearer how to fix it after the > event loop is cleaned up. By event loop, do you mean the per-vCPU jobs list?