qemu-kvm/kvm-vfio-pci-Make-vfio-cdev-pre-openable-by-passing-a-fi.patch
Miroslav Rezanina 9ef7cdf7ca * Wed Jan 24 2024 Miroslav Rezanina <mrezanin@redhat.com> - 8.2.0-3
- kvm-hw-arm-virt-Add-properties-to-disable-high-memory-re.patch [RHEL-19738]
- kvm-vfio-Introduce-base-object-for-VFIOContainer-and-tar.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Introduce-a-empty-VFIOIOMMUOps.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Switch-to-dma_map-unmap-API.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-common-Introduce-vfio_container_init-destroy-he.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-common-Move-giommu_list-in-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-space-field-to-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Switch-to-IOMMU-BE-set_dirty_page_tra.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-per-container-device-list-in-bas.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Convert-functions-to-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-pgsizes-and-dma_max_mappings-to-.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-vrdl_list-to-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-listener-to-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-dirty_pgsizes-and-max_dirty_bitm.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Move-iova_ranges-to-base-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Implement-attach-detach_device.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Introduce-spapr-backend-and-target-interf.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-switch-to-spapr-IOMMU-BE-add-del_section_.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Move-prereg_listener-into-spapr-container.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Move-hostwin_list-into-spapr-container.patch [RHEL-19302 RHEL-21057]
- kvm-backends-iommufd-Introduce-the-iommufd-object.patch [RHEL-19302 RHEL-21057]
- kvm-util-char_dev-Add-open_cdev.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-common-return-early-if-space-isn-t-empty.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Implement-the-iommufd-backend.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Relax-assert-check-for-iommufd-backend.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Add-support-for-iova_ranges-and-pgsizes.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-pci-Extract-out-a-helper-vfio_pci_get_pci_hot_r.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-pci-Introduce-a-vfio-pci-hot-reset-interface.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Enable-pci-hot-reset-through-iommufd-cd.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-pci-Allow-the-selection-of-a-given-iommu-backen.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-pci-Make-vfio-cdev-pre-openable-by-passing-a-fi.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-platform-Allow-the-selection-of-a-given-iommu-b.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-platform-Make-vfio-cdev-pre-openable-by-passing.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ap-Allow-the-selection-of-a-given-iommu-backend.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ap-Make-vfio-cdev-pre-openable-by-passing-a-fil.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ccw-Allow-the-selection-of-a-given-iommu-backen.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ccw-Make-vfio-cdev-pre-openable-by-passing-a-fi.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-Make-VFIOContainerBase-poiner-parameter-const-i.patch [RHEL-19302 RHEL-21057]
- kvm-hw-arm-Activate-IOMMUFD-for-virt-machines.patch [RHEL-19302 RHEL-21057]
- kvm-kconfig-Activate-IOMMUFD-for-s390x-machines.patch [RHEL-19302 RHEL-21057]
- kvm-hw-i386-Activate-IOMMUFD-for-q35-machines.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-pci-Move-VFIODevice-initializations-in-vfio_ins.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-platform-Move-VFIODevice-initializations-in-vfi.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ap-Move-VFIODevice-initializations-in-vfio_ap_i.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-ccw-Move-VFIODevice-initializations-in-vfio_ccw.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-Introduce-a-helper-function-to-initialize-VFIOD.patch [RHEL-19302 RHEL-21057]
- kvm-docs-devel-Add-VFIO-iommufd-backend-documentation.patch [RHEL-19302 RHEL-21057]
- kvm-hw-ppc-Kconfig-Imply-VFIO_PCI.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Extend-VFIOIOMMUOps-with-a-release-handle.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Introduce-vfio_legacy_setup-for-furth.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Initialize-VFIOIOMMUOps-under-vfio_in.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Introduce-a-VFIOIOMMU-QOM-interface.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Introduce-a-VFIOIOMMU-legacy-QOM-inte.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Intoduce-a-new-VFIOIOMMUClass-setup-h.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Introduce-a-sPAPR-VFIOIOMMU-QOM-interface.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Introduce-a-VFIOIOMMU-iommufd-QOM-inter.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-spapr-Only-compile-sPAPR-IOMMU-support-when-nee.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Remove-CONFIG_IOMMUFD-usage.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Replace-basename-with-g_path_get_base.patch [RHEL-19302 RHEL-21057]
- kvm-hw-vfio-fix-iteration-over-global-VFIODevice-list.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-iommufd-Remove-the-use-of-stat-to-check-file-ex.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-container-Rename-vfio_init_container-to-vfio_se.patch [RHEL-19302 RHEL-21057]
- kvm-vfio-migration-Add-helper-function-to-set-state-or-r.patch [RHEL-19302 RHEL-21057]
- kvm-backends-iommufd-Remove-check-on-number-of-backend-u.patch [RHEL-19302 RHEL-21057]
- kvm-backends-iommufd-Remove-mutex.patch [RHEL-19302 RHEL-21057]
- kvm-Compile-IOMMUFD-object-on-aarch64.patch [RHEL-19302 RHEL-21057]
- kvm-Compile-IOMMUFD-on-s390x.patch [RHEL-19302 RHEL-21057]
- kvm-Compile-IOMMUFD-on-x86_64.patch [RHEL-19302 RHEL-21057]
- kvm-target-s390x-kvm-pv-Provide-some-more-useful-informa.patch [RHEL-18212]
- kvm-nbd-server-avoid-per-NBDRequest-nbd_client_get-put.patch [RHEL-15965]
- kvm-nbd-server-only-traverse-NBDExport-clients-from-main.patch [RHEL-15965]
- kvm-nbd-server-introduce-NBDClient-lock-to-protect-field.patch [RHEL-15965]
- kvm-block-file-posix-set-up-Linux-AIO-and-io_uring-in-th.patch [RHEL-15965]
- kvm-virtio-blk-add-lock-to-protect-s-rq.patch [RHEL-15965]
- kvm-virtio-blk-don-t-lock-AioContext-in-the-completion-c.patch [RHEL-15965]
- kvm-virtio-blk-don-t-lock-AioContext-in-the-submission-c.patch [RHEL-15965]
- kvm-scsi-only-access-SCSIDevice-requests-from-one-thread.patch [RHEL-15965]
- kvm-virtio-scsi-don-t-lock-AioContext-around-virtio_queu.patch [RHEL-15965]
- kvm-scsi-don-t-lock-AioContext-in-I-O-code-path.patch [RHEL-15965]
- kvm-dma-helpers-don-t-lock-AioContext-in-dma_blk_cb.patch [RHEL-15965]
- kvm-virtio-scsi-replace-AioContext-lock-with-tmf_bh_lock.patch [RHEL-15965]
- kvm-scsi-assert-that-callbacks-run-in-the-correct-AioCon.patch [RHEL-15965]
- kvm-tests-remove-aio_context_acquire-tests.patch [RHEL-15965]
- kvm-aio-make-aio_context_acquire-aio_context_release-a-n.patch [RHEL-15965]
- kvm-graph-lock-remove-AioContext-locking.patch [RHEL-15965]
- kvm-block-remove-AioContext-locking.patch [RHEL-15965]
- kvm-block-remove-bdrv_co_lock.patch [RHEL-15965]
- kvm-scsi-remove-AioContext-locking.patch [RHEL-15965]
- kvm-aio-wait-draw-equivalence-between-AIO_WAIT_WHILE-and.patch [RHEL-15965]
- kvm-aio-remove-aio_context_acquire-aio_context_release-A.patch [RHEL-15965]
- kvm-docs-remove-AioContext-lock-from-IOThread-docs.patch [RHEL-15965]
- kvm-scsi-remove-outdated-AioContext-lock-comment.patch [RHEL-15965]
- kvm-job-remove-outdated-AioContext-locking-comments.patch [RHEL-15965]
- kvm-block-remove-outdated-AioContext-locking-comments.patch [RHEL-15965]
- kvm-block-coroutine-wrapper-use-qemu_get_current_aio_con.patch [RHEL-15965]
- kvm-s390x-pci-avoid-double-enable-disable-of-aif.patch [RHEL-21169]
- kvm-s390x-pci-refresh-fh-before-disabling-aif.patch [RHEL-21169]
- kvm-s390x-pci-drive-ISM-reset-from-subsystem-reset.patch [RHEL-21169]
- kvm-include-ui-rect.h-fix-qemu_rect_init-mis-assignment.patch [RHEL-21570]
- kvm-virtio-gpu-block-migration-of-VMs-with-blob-true.patch [RHEL-7565]
- kvm-spec-Enable-zstd.patch [RHEL-7361]
- Resolves: RHEL-19738
  (Enable properties allowing to disable high memory regions)
- Resolves: RHEL-19302
  (NVIDIA:Grace-Hopper Backport QEMU IOMMUFD Backend)
- Resolves: RHEL-21057
  (Request backport of 9353b6da430f90e47f352dbf6dc31120c8914da6)
- Resolves: RHEL-18212
  ([RHEL9][Secure-execution][s390x] The error message is not clear when boot up a SE guest with wrong encryption)
- Resolves: RHEL-15965
  ( [qemu-kvm] Remove AioContext lock (no response with QMP command block_resize))
- Resolves: RHEL-21169
  ([s390x] VM fails to start with ISM passed through QEMU 8.2)
- Resolves: RHEL-21570
  (Critical performance degradation for input devices in virtio vnc session)
- Resolves: RHEL-7565
  (qemu crashed when migrate guest with blob resources enabled)
- Resolves: RHEL-7361
  ([qemu-kvm] Enable zstd support for qcow2 files)
2024-01-24 04:26:42 -05:00

238 lines
8.4 KiB
Diff

From 965a44793806fef2094906947bd3b428638bf89a Mon Sep 17 00:00:00 2001
From: Zhenzhong Duan <zhenzhong.duan@intel.com>
Date: Tue, 21 Nov 2023 16:44:10 +0800
Subject: [PATCH 031/101] vfio/pci: Make vfio cdev pre-openable by passing a
file handle
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Eric Auger <eric.auger@redhat.com>
RH-MergeRequest: 211: IOMMUFD backend backport
RH-Jira: RHEL-19302 RHEL-21057
RH-Acked-by: Cédric Le Goater <clg@redhat.com>
RH-Acked-by: Sebastian Ott <sebott@redhat.com>
RH-Commit: [30/67] a14b824b700e8fb36633cd159bcc422d992a316f (eauger1/centos-qemu-kvm)
Conflicts: contextual conflict in hw/vfio/pci.c due to
RHEL-only f73562144e492 vfio: cap number of devices that can be assigned
This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu. This way qemu never needs
to have privilege to open a VFIO or iommu cdev node.
Together with the earlier support of pre-opening /dev/iommu device,
now we have full support of passing a vfio device to unprivileged
qemu by management tool. This mode is no more considered for the
legacy backend. So let's remove the "TODO" comment.
Add helper functions vfio_device_set_fd() and vfio_device_get_name()
to set fd and get device name, they will also be used by other vfio
devices.
There is no easy way to check if a device is mdev with FD passing,
so fail the x-balloon-allowed check unconditionally in this case.
There is also no easy way to get BDF as name with FD passing, so
we fake a name by VFIO_FD[fd].
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
(cherry picked from commit da3e04b26fd8d15b344944504d5ffa9c5f20b54b)
Signed-off-by: Eric Auger <eric.auger@redhat.com>
---
hw/vfio/helpers.c | 43 +++++++++++++++++++++++++++++++++++
hw/vfio/iommufd.c | 12 ++++++----
hw/vfio/pci.c | 28 +++++++++++++----------
include/hw/vfio/vfio-common.h | 4 ++++
4 files changed, 71 insertions(+), 16 deletions(-)
diff --git a/hw/vfio/helpers.c b/hw/vfio/helpers.c
index 168847e7c5..3592c3d54e 100644
--- a/hw/vfio/helpers.c
+++ b/hw/vfio/helpers.c
@@ -27,6 +27,7 @@
#include "trace.h"
#include "qapi/error.h"
#include "qemu/error-report.h"
+#include "monitor/monitor.h"
/*
* Common VFIO interrupt disable
@@ -609,3 +610,45 @@ bool vfio_has_region_cap(VFIODevice *vbasedev, int region, uint16_t cap_type)
return ret;
}
+
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp)
+{
+ struct stat st;
+
+ if (vbasedev->fd < 0) {
+ if (stat(vbasedev->sysfsdev, &st) < 0) {
+ error_setg_errno(errp, errno, "no such host device");
+ error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ return -errno;
+ }
+ /* User may specify a name, e.g: VFIO platform device */
+ if (!vbasedev->name) {
+ vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+ }
+ } else {
+ if (!vbasedev->iommufd) {
+ error_setg(errp, "Use FD passing only with iommufd backend");
+ return -EINVAL;
+ }
+ /*
+ * Give a name with fd so any function printing out vbasedev->name
+ * will not break.
+ */
+ if (!vbasedev->name) {
+ vbasedev->name = g_strdup_printf("VFIO_FD%d", vbasedev->fd);
+ }
+ }
+
+ return 0;
+}
+
+void vfio_device_set_fd(VFIODevice *vbasedev, const char *str, Error **errp)
+{
+ int fd = monitor_fd_param(monitor_cur(), str, errp);
+
+ if (fd < 0) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ vbasedev->fd = fd;
+}
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index 6e53e013ef..5accd26484 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -320,11 +320,15 @@ static int iommufd_cdev_attach(const char *name, VFIODevice *vbasedev,
uint32_t ioas_id;
Error *err = NULL;
- devfd = iommufd_cdev_getfd(vbasedev->sysfsdev, errp);
- if (devfd < 0) {
- return devfd;
+ if (vbasedev->fd < 0) {
+ devfd = iommufd_cdev_getfd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+ } else {
+ devfd = vbasedev->fd;
}
- vbasedev->fd = devfd;
ret = iommufd_cdev_connect_and_bind(vbasedev, errp);
if (ret) {
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 39e6a6678e..3412a63bb1 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -2949,7 +2949,6 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
VFIOGroup *group;
char *tmp, *subsys;
Error *err = NULL;
- struct stat st;
int ret, i = 0;
bool is_mdev;
char uuid[UUID_STR_LEN];
@@ -2976,11 +2975,14 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
return;
}
- if (!vbasedev->sysfsdev) {
+ if (vbasedev->fd < 0 && !vbasedev->sysfsdev) {
if (!(~vdev->host.domain || ~vdev->host.bus ||
~vdev->host.slot || ~vdev->host.function)) {
error_setg(errp, "No provided host device");
error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F "
+#ifdef CONFIG_IOMMUFD
+ "or -device vfio-pci,fd=DEVICE_FD "
+#endif
"or -device vfio-pci,sysfsdev=PATH_TO_DEVICE\n");
return;
}
@@ -2990,13 +2992,9 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
vdev->host.slot, vdev->host.function);
}
- if (stat(vbasedev->sysfsdev, &st) < 0) {
- error_setg_errno(errp, errno, "no such host device");
- error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ if (vfio_device_get_name(vbasedev, errp) < 0) {
return;
}
-
- vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
vbasedev->ops = &vfio_pci_ops;
vbasedev->type = VFIO_DEVICE_TYPE_PCI;
vbasedev->dev = DEVICE(vdev);
@@ -3356,6 +3354,7 @@ static void vfio_instance_init(Object *obj)
vdev->host.bus = ~0U;
vdev->host.slot = ~0U;
vdev->host.function = ~0U;
+ vdev->vbasedev.fd = -1;
vdev->nv_gpudirect_clique = 0xFF;
@@ -3412,11 +3411,6 @@ static Property vfio_pci_dev_properties[] = {
qdev_prop_nv_gpudirect_clique, uint8_t),
DEFINE_PROP_OFF_AUTO_PCIBAR("x-msix-relocation", VFIOPCIDevice, msix_relo,
OFF_AUTOPCIBAR_OFF),
- /*
- * TODO - support passed fds... is this necessary?
- * DEFINE_PROP_STRING("vfiofd", VFIOPCIDevice, vfiofd_name),
- * DEFINE_PROP_STRING("vfiogroupfd, VFIOPCIDevice, vfiogroupfd_name),
- */
#ifdef CONFIG_IOMMUFD
DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
@@ -3424,6 +3418,13 @@ static Property vfio_pci_dev_properties[] = {
DEFINE_PROP_END_OF_LIST(),
};
+#ifdef CONFIG_IOMMUFD
+static void vfio_pci_set_fd(Object *obj, const char *str, Error **errp)
+{
+ vfio_device_set_fd(&VFIO_PCI(obj)->vbasedev, str, errp);
+}
+#endif
+
static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
@@ -3431,6 +3432,9 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
dc->reset = vfio_pci_reset;
device_class_set_props(dc, vfio_pci_dev_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_pci_set_fd);
+#endif
dc->desc = "VFIO-based PCI device assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
pdc->realize = vfio_realize;
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 3dac5c167e..697bf24a35 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -251,4 +251,8 @@ int vfio_devices_query_dirty_bitmap(VFIOContainerBase *bcontainer,
hwaddr size);
int vfio_get_dirty_bitmap(VFIOContainerBase *bcontainer, uint64_t iova,
uint64_t size, ram_addr_t ram_addr);
+
+/* Returns 0 on success, or a negative errno. */
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp);
+void vfio_device_set_fd(VFIODevice *vbasedev, const char *str, Error **errp);
#endif /* HW_VFIO_VFIO_COMMON_H */
--
2.39.3