From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754800AbbKWUni (ORCPT ); Mon, 23 Nov 2015 15:43:38 -0500 Received: from mx1.redhat.com ([209.132.183.28]:39774 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754730AbbKWUne (ORCPT ); Mon, 23 Nov 2015 15:43:34 -0500 Subject: [RFC PATCH 3/3] vfio/pci: Include sparse mmap capability for MSI-X table regions From: Alex Williamson To: alex.williamson@redhat.com Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org Date: Mon, 23 Nov 2015 13:43:33 -0700 Message-ID: <20151123204333.18252.97032.stgit@gimli.home> In-Reply-To: <20151123202614.18252.41590.stgit@gimli.home> References: <20151123202614.18252.41590.stgit@gimli.home> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org vfio-pci has never allowed the user to directly mmap the MSI-X vector table, but we've always relied on implicit knowledge of the user that they cannot do this. Now that we have capability chains that we can expose in the region info ioctl and a sparse mmap capability that represents the sub-areas within the region that can be mmap'd, we can make the mmap constraints more explicit. Signed-off-by: Alex Williamson --- drivers/vfio/pci/vfio_pci.c | 101 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 100 insertions(+), 1 deletion(-) diff --git a/drivers/vfio/pci/vfio_pci.c b/drivers/vfio/pci/vfio_pci.c index 32b88bd..46e7aed 100644 --- a/drivers/vfio/pci/vfio_pci.c +++ b/drivers/vfio/pci/vfio_pci.c @@ -421,6 +421,77 @@ static int vfio_pci_for_each_slot_or_bus(struct pci_dev *pdev, return walk.ret; } +struct caps { + struct vfio_info_cap_header *buf; + size_t size; + size_t head; +}; + +static void *add_region_info_cap(struct caps *caps, + size_t size, u16 id, u16 version) +{ + void *tmp; + struct vfio_info_cap_header *header; + + /* This would be ridiculous and exceeds the ioctl's abilities */ + BUG_ON(caps->size + size + sizeof(struct vfio_region_info) > U32_MAX); + + tmp = krealloc(caps->buf, caps->size + size, GFP_KERNEL); + if (!tmp) { + kfree(caps->buf); + caps->size = 0; + return ERR_PTR(-ENOMEM); + } + + caps->buf = tmp; + header = tmp + caps->size; + header->id = id; + header->version = version; + header->next = caps->head; + caps->head = caps->size + sizeof(struct vfio_region_info); + caps->size += size; + + return header; +} + +static int msix_sparse_mmap_cap(struct vfio_pci_device *vdev, struct caps *caps) +{ + struct vfio_region_info_cap_sparse_mmap *sparse; + size_t end, size; + int nr_areas = 2, i = 0; + + end = pci_resource_len(vdev->pdev, vdev->msix_bar); + + /* If MSI-X table is aligned to the start or end, only one area */ + if (((vdev->msix_offset & PAGE_MASK) == 0) || + (PAGE_ALIGN(vdev->msix_offset + vdev->msix_size) >= end)) + nr_areas = 1; + + size = sizeof(*sparse) + (nr_areas * sizeof(*sparse->areas)); + + sparse = add_region_info_cap(caps, size, + VFIO_REGION_INFO_CAP_SPARSE_MMAP, 1); + if (IS_ERR(sparse)) + return PTR_ERR(sparse); + + sparse->nr_areas = nr_areas; + + if (vdev->msix_offset & PAGE_MASK) { + sparse->areas[i].offset = 0; + sparse->areas[i].size = vdev->msix_offset & PAGE_MASK; + i++; + } + + if (PAGE_ALIGN(vdev->msix_offset + vdev->msix_size) < end) { + sparse->areas[i].offset = PAGE_ALIGN(vdev->msix_offset + + vdev->msix_size); + sparse->areas[i].size = end - sparse->areas[i].offset; + i++; + } + + return 0; +} + static long vfio_pci_ioctl(void *device_data, unsigned int cmd, unsigned long arg) { @@ -451,6 +522,8 @@ static long vfio_pci_ioctl(void *device_data, } else if (cmd == VFIO_DEVICE_GET_REGION_INFO) { struct pci_dev *pdev = vdev->pdev; struct vfio_region_info info; + struct caps caps = { .buf = NULL, .size = 0, .head = 0 }; + int ret; minsz = offsetofend(struct vfio_region_info, offset); @@ -479,8 +552,15 @@ static long vfio_pci_ioctl(void *device_data, VFIO_REGION_INFO_FLAG_WRITE; if (IS_ENABLED(CONFIG_VFIO_PCI_MMAP) && pci_resource_flags(pdev, info.index) & - IORESOURCE_MEM && info.size >= PAGE_SIZE) + IORESOURCE_MEM && info.size >= PAGE_SIZE) { info.flags |= VFIO_REGION_INFO_FLAG_MMAP; + if (info.index == vdev->msix_bar) { + ret = msix_sparse_mmap_cap(vdev, &caps); + if (ret) + return ret; + } + } + break; case VFIO_PCI_ROM_REGION_INDEX: { @@ -520,6 +600,25 @@ static long vfio_pci_ioctl(void *device_data, return -EINVAL; } + if (caps.size) { + info.flags |= VFIO_REGION_INFO_FLAG_CAPS; + if (info.argsz < sizeof(info) + caps.size) { + info.argsz = sizeof(info) + caps.size; + info.cap_offset = 0; + } else { + ret = copy_to_user((void __user *)arg + + sizeof(info), caps.buf, + caps.size); + if (ret) { + kfree(caps.buf); + return ret; + } + info.cap_offset = caps.head; + } + + kfree(caps.buf); + } + return copy_to_user((void __user *)arg, &info, minsz); } else if (cmd == VFIO_DEVICE_GET_IRQ_INFO) {