From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:44771) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cfvEo-0001tp-Vg for qemu-devel@nongnu.org; Mon, 20 Feb 2017 16:11:30 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cfvCH-0007Gm-5S for qemu-devel@nongnu.org; Mon, 20 Feb 2017 16:11:14 -0500 Received: from mx1.redhat.com ([209.132.183.28]:44942) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1cfvCG-0007Gc-PF for qemu-devel@nongnu.org; Mon, 20 Feb 2017 16:08:37 -0500 Date: Mon, 20 Feb 2017 14:08:33 -0700 From: Alex Williamson Message-ID: <20170220140833.7221fcd5@t450s.home> In-Reply-To: <1487554087-15347-2-git-send-email-tianyu.lan@intel.com> References: <1487554087-15347-1-git-send-email-tianyu.lan@intel.com> <1487554087-15347-2-git-send-email-tianyu.lan@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [Resend RFC PATCH 1/4] VFIO: Set eventfd for IOMMU fault event via new vfio cmd List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Lan Tianyu Cc: qemu-devel@nongnu.org, kevin.tian@intel.com, mst@redhat.com, jan.kiszka@siemens.com, jasowang@redhat.com, peterx@redhat.com, david@gibson.dropbear.id.au, yi.l.liu@intel.com On Mon, 20 Feb 2017 09:28:04 +0800 Lan Tianyu wrote: > This patch is to assign an event fd to VFIO IOMMU type1 driver > in order to get notification when IOMMU driver reports fault event. > > Signed-off-by: Lan Tianyu > --- > hw/vfio/common.c | 37 +++++++++++++++++++++++++++++++++++++ > include/hw/vfio/vfio-common.h | 3 +++ > linux-headers/linux/vfio.h | 13 +++++++++++++ > 3 files changed, 53 insertions(+) > > diff --git a/hw/vfio/common.c b/hw/vfio/common.c > index 6b33b9f..628b424 100644 > --- a/hw/vfio/common.c > +++ b/hw/vfio/common.c > @@ -33,6 +33,7 @@ > #include "qemu/error-report.h" > #include "qemu/range.h" > #include "sysemu/kvm.h" > +#include "sysemu/sysemu.h" > #include "trace.h" > #include "qapi/error.h" > > @@ -294,6 +295,34 @@ static bool vfio_listener_skipped_section(MemoryRegionSection *section) > section->offset_within_address_space & (1ULL << 63); > } > > +static void vfio_iommu_fault(void *opaque) > +{ > +} > + > +static int vfio_set_iommu_fault_notifier(struct VFIOContainer *container) > +{ > + struct vfio_iommu_type1_set_fault_eventfd eventfd; > + int ret; > + > + ret = event_notifier_init(&container->fault_notifier, 0); > + if (ret < 0) { > + error_report("vfio: Failed to init notifier for IOMMU fault event"); > + return ret; > + } > + > + eventfd.fd = event_notifier_get_fd(&container->fault_notifier); > + eventfd.argsz = sizeof(eventfd); > + > + ret = ioctl(container->fd, VFIO_IOMMU_SET_FAULT_EVENTFD, &eventfd); > + if (ret < 0) { > + error_report("vfio: Failed to set notifier for IOMMU fault event"); > + return ret; > + } > + > + qemu_set_fd_handler(eventfd.fd, vfio_iommu_fault, NULL, container); > + return 0; > +} > + > /* Called with rcu_read_lock held. */ > static bool vfio_get_vaddr(IOMMUTLBEntry *iotlb, void **vaddr, > bool *read_only) > @@ -1103,6 +1132,14 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as, > goto listener_release_exit; > } > > + if (memory_region_is_iommu(container->space->as->root)) { > + if (vfio_set_iommu_fault_notifier(container)) { > + error_setg_errno(errp, -ret, > + "Fail to set IOMMU fault notifier"); > + goto listener_release_exit; > + } > + } Backwards compatibility needs to be supported, we can't invent a feature and impose it on all users. We can either opportunistically enable it and be silent about it when it's not there, or we can add a feature flag and complain and fail if the user requests the feature and it's not available or errors. > + > container->initialized = true; > > QLIST_INIT(&container->group_list); > diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h > index c582de1..1b594c6 100644 > --- a/include/hw/vfio/vfio-common.h > +++ b/include/hw/vfio/vfio-common.h > @@ -26,6 +26,7 @@ > #include "exec/memory.h" > #include "qemu/queue.h" > #include "qemu/notify.h" > +#include "qemu/event_notifier.h" > #ifdef CONFIG_LINUX > #include > #endif > @@ -81,6 +82,8 @@ typedef struct VFIOContainer { > unsigned iommu_type; > int error; > bool initialized; > + EventNotifier fault_notifier; > + > /* > * This assumes the host IOMMU can support only a single > * contiguous IOVA window. We may need to generalize that in > diff --git a/linux-headers/linux/vfio.h b/linux-headers/linux/vfio.h > index 759b850..ca890ee 100644 > --- a/linux-headers/linux/vfio.h > +++ b/linux-headers/linux/vfio.h > @@ -537,6 +537,19 @@ struct vfio_iommu_type1_dma_unmap { > #define VFIO_IOMMU_ENABLE _IO(VFIO_TYPE, VFIO_BASE + 15) > #define VFIO_IOMMU_DISABLE _IO(VFIO_TYPE, VFIO_BASE + 16) > > +/* > + * VFIO_IOMMU_SET_FAULT_EVENT_FD _IO(VFIO_TYPE, VFIO_BASE + 17) > + * > + * Receive eventfd from userspace to notify fault event from IOMMU. > + */ > +struct vfio_iommu_type1_set_fault_eventfd { > + __u32 argsz; > + __u32 flags; > + __u32 fd; > +}; > + > +#define VFIO_IOMMU_SET_FAULT_EVENTFD _IO(VFIO_TYPE, VFIO_BASE + 17) > + > /* -------- Additional API for SPAPR TCE (Server POWERPC) IOMMU -------- */ > > /*