From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sheng Yang Subject: Re: [PATCH 0/1] IRQ Routing Date: Wed, 14 Jan 2009 16:39:29 +0800 Message-ID: <200901141639.29844.sheng@linux.intel.com> References: <1231876950-10092-1-git-send-email-avi@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Cc: Xiantao Zhang , Marcelo Tosatti , kvm@vger.kernel.org To: Avi Kivity Return-path: Received: from mga10.intel.com ([192.55.52.92]:12694 "EHLO fmsmga102.fm.intel.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751600AbZANIjc (ORCPT ); Wed, 14 Jan 2009 03:39:32 -0500 In-Reply-To: <1231876950-10092-1-git-send-email-avi@redhat.com> Content-Disposition: inline Sender: kvm-owner@vger.kernel.org List-ID: On Wednesday 14 January 2009 04:02:29 Avi Kivity wrote: > Following is my alternative to irq routing. The differences compared to > Sheng's version are: > > - A single ioctl to replace the entire routing table, instead of add/remove > ioctls for individual routing entries. Routing changes are rare, and > we need to track the entire table in userspace anyway (for save/restore, > and for user irqchip). As a side effect changes are atomic. > - Interrupt numbers are allocated by userspace, instead of the kernel > - I implemented irqchip routings rather then MSIs, it should be easy to > add MSIs later on. > > Please review and comment. Look nice to me now... Save/restore is a good reason to maintain a table in userspace. And looking forward to userspace patch. Put a table in kvm_context? -- regards Yang, Sheng > > Avi Kivity (1): > KVM: Userspace controlled irq routing > > arch/ia64/kvm/kvm-ia64.c | 5 ++ > arch/x86/kvm/x86.c | 6 ++ > include/linux/kvm.h | 33 +++++++++ > include/linux/kvm_host.h | 31 +++++++++ > virt/kvm/irq_comm.c | 168 > ++++++++++++++++++++++++++++++++++++++++++++- virt/kvm/kvm_main.c | > 25 +++++++ > 6 files changed, 264 insertions(+), 4 deletions(-)