From: Shuo Liu <[email protected]>
MMIO device passthrough enables an OS in a virtual machine to directly
access a MMIO device in the host. It promises almost the native
performance, which is required in performance-critical scenarios of
ACRN.
HSM provides the following ioctls:
- Assign - ACRN_IOCTL_ASSIGN_MMIODEV
Pass data struct acrn_mmiodev from userspace to the hypervisor, and
inform the hypervisor to assign a MMIO device to a User VM.
- De-assign - ACRN_IOCTL_DEASSIGN_PCIDEV
Pass data struct acrn_mmiodev from userspace to the hypervisor, and
inform the hypervisor to de-assign a MMIO device from a User VM.
Signed-off-by: Shuo Liu <[email protected]>
Signed-off-by: Fei Li <[email protected]>
---
drivers/virt/acrn/hsm.c | 25 +++++++++++++++++++++++++
drivers/virt/acrn/hypercall.h | 26 ++++++++++++++++++++++++++
include/uapi/linux/acrn.h | 28 ++++++++++++++++++++++++++++
3 files changed, 79 insertions(+)
diff --git a/drivers/virt/acrn/hsm.c b/drivers/virt/acrn/hsm.c
index 130e12b8652a..f567ca59d7c2 100644
--- a/drivers/virt/acrn/hsm.c
+++ b/drivers/virt/acrn/hsm.c
@@ -114,6 +114,7 @@ static long acrn_dev_ioctl(struct file *filp, unsigned int cmd,
struct acrn_ptdev_irq *irq_info;
struct acrn_ioeventfd ioeventfd;
struct acrn_vm_memmap memmap;
+ struct acrn_mmiodev *mmiodev;
struct acrn_msi_entry *msi;
struct acrn_pcidev *pcidev;
struct acrn_irqfd irqfd;
@@ -217,6 +218,30 @@ static long acrn_dev_ioctl(struct file *filp, unsigned int cmd,
ret = acrn_vm_memseg_unmap(vm, &memmap);
break;
+ case ACRN_IOCTL_ASSIGN_MMIODEV:
+ mmiodev = memdup_user((void __user *)ioctl_param,
+ sizeof(struct acrn_mmiodev));
+ if (IS_ERR(mmiodev))
+ return PTR_ERR(mmiodev);
+
+ ret = hcall_assign_mmiodev(vm->vmid, virt_to_phys(mmiodev));
+ if (ret < 0)
+ dev_dbg(acrn_dev.this_device,
+ "Failed to assign MMIO device!\n");
+ kfree(mmiodev);
+ break;
+ case ACRN_IOCTL_DEASSIGN_MMIODEV:
+ mmiodev = memdup_user((void __user *)ioctl_param,
+ sizeof(struct acrn_mmiodev));
+ if (IS_ERR(mmiodev))
+ return PTR_ERR(mmiodev);
+
+ ret = hcall_deassign_mmiodev(vm->vmid, virt_to_phys(mmiodev));
+ if (ret < 0)
+ dev_dbg(acrn_dev.this_device,
+ "Failed to deassign MMIO device!\n");
+ kfree(mmiodev);
+ break;
case ACRN_IOCTL_ASSIGN_PCIDEV:
pcidev = memdup_user((void __user *)ioctl_param,
sizeof(struct acrn_pcidev));
diff --git a/drivers/virt/acrn/hypercall.h b/drivers/virt/acrn/hypercall.h
index 0cfad05bd1a9..f0c78e52cebb 100644
--- a/drivers/virt/acrn/hypercall.h
+++ b/drivers/virt/acrn/hypercall.h
@@ -41,6 +41,8 @@
#define HC_RESET_PTDEV_INTR _HC_ID(HC_ID, HC_ID_PCI_BASE + 0x04)
#define HC_ASSIGN_PCIDEV _HC_ID(HC_ID, HC_ID_PCI_BASE + 0x05)
#define HC_DEASSIGN_PCIDEV _HC_ID(HC_ID, HC_ID_PCI_BASE + 0x06)
+#define HC_ASSIGN_MMIODEV _HC_ID(HC_ID, HC_ID_PCI_BASE + 0x07)
+#define HC_DEASSIGN_MMIODEV _HC_ID(HC_ID, HC_ID_PCI_BASE + 0x08)
#define HC_ID_PM_BASE 0x80UL
#define HC_PM_GET_CPU_STATE _HC_ID(HC_ID, HC_ID_PM_BASE + 0x00)
@@ -194,6 +196,30 @@ static inline long hcall_set_memory_regions(u64 regions_pa)
return acrn_hypercall1(HC_VM_SET_MEMORY_REGIONS, regions_pa);
}
+/**
+ * hcall_assign_mmiodev() - Assign a MMIO device to a User VM
+ * @vmid: User VM ID
+ * @addr: Service VM GPA of the &struct acrn_mmiodev
+ *
+ * Return: 0 on success, <0 on failure
+ */
+static inline long hcall_assign_mmiodev(u64 vmid, u64 addr)
+{
+ return acrn_hypercall2(HC_ASSIGN_MMIODEV, vmid, addr);
+}
+
+/**
+ * hcall_deassign_mmiodev() - De-assign a PCI device from a User VM
+ * @vmid: User VM ID
+ * @addr: Service VM GPA of the &struct acrn_mmiodev
+ *
+ * Return: 0 on success, <0 on failure
+ */
+static inline long hcall_deassign_mmiodev(u64 vmid, u64 addr)
+{
+ return acrn_hypercall2(HC_DEASSIGN_MMIODEV, vmid, addr);
+}
+
/**
* hcall_assign_pcidev() - Assign a PCI device to a User VM
* @vmid: User VM ID
diff --git a/include/uapi/linux/acrn.h b/include/uapi/linux/acrn.h
index 353b2a2e4536..470036d6b1ac 100644
--- a/include/uapi/linux/acrn.h
+++ b/include/uapi/linux/acrn.h
@@ -396,6 +396,7 @@ struct acrn_ptdev_irq {
/* Type of PCI device assignment */
#define ACRN_PTDEV_QUIRK_ASSIGN (1U << 0)
+#define ACRN_MMIODEV_RES_NUM 3
#define ACRN_PCI_NUM_BARS 6
/**
* struct acrn_pcidev - Info for assigning or de-assigning a PCI device
@@ -417,6 +418,29 @@ struct acrn_pcidev {
__u32 bar[ACRN_PCI_NUM_BARS];
};
+/**
+ * struct acrn_mmiodev - Info for assigning or de-assigning a MMIO device
+ * @name: Name of the MMIO device.
+ * @res[].user_vm_pa: Physical address of User VM of the MMIO region
+ * for the MMIO device.
+ * @res[].service_vm_pa: Physical address of Service VM of the MMIO
+ * region for the MMIO device.
+ * @res[].size: Size of the MMIO region for the MMIO device.
+ * @res[].mem_type: Memory type of the MMIO region for the MMIO
+ * device.
+ *
+ * This structure will be passed to hypervisor directly.
+ */
+struct acrn_mmiodev {
+ __u8 name[8];
+ struct {
+ __u64 user_vm_pa;
+ __u64 service_vm_pa;
+ __u64 size;
+ __u64 mem_type;
+ } res[ACRN_MMIODEV_RES_NUM];
+};
+
/**
* struct acrn_msi_entry - Info for injecting a MSI interrupt to a VM
* @msi_addr: MSI addr[19:12] with dest vCPU ID
@@ -568,6 +592,10 @@ struct acrn_irqfd {
_IOW(ACRN_IOCTL_TYPE, 0x55, struct acrn_pcidev)
#define ACRN_IOCTL_DEASSIGN_PCIDEV \
_IOW(ACRN_IOCTL_TYPE, 0x56, struct acrn_pcidev)
+#define ACRN_IOCTL_ASSIGN_MMIODEV \
+ _IOW(ACRN_IOCTL_TYPE, 0x57, struct acrn_mmiodev)
+#define ACRN_IOCTL_DEASSIGN_MMIODEV \
+ _IOW(ACRN_IOCTL_TYPE, 0x58, struct acrn_mmiodev)
#define ACRN_IOCTL_PM_GET_CPU_STATE \
_IOWR(ACRN_IOCTL_TYPE, 0x60, __u64)
--
2.25.1
On Wed, Aug 25, 2021 at 05:01:40PM +0800, Fei Li wrote:
> From: Shuo Liu <[email protected]>
>
> MMIO device passthrough enables an OS in a virtual machine to directly
> access a MMIO device in the host. It promises almost the native
> performance, which is required in performance-critical scenarios of
> ACRN.
>
> HSM provides the following ioctls:
> - Assign - ACRN_IOCTL_ASSIGN_MMIODEV
> Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> inform the hypervisor to assign a MMIO device to a User VM.
>
> - De-assign - ACRN_IOCTL_DEASSIGN_PCIDEV
> Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> inform the hypervisor to de-assign a MMIO device from a User VM.
>
> Signed-off-by: Shuo Liu <[email protected]>
> Signed-off-by: Fei Li <[email protected]>
Where is the userspace code that uses this new api?
thanks,
greg k-h
On Wed, Aug 25, 2021 at 10:43:29PM +0200, Greg KH wrote:
> On Wed, Aug 25, 2021 at 05:01:40PM +0800, Fei Li wrote:
> > From: Shuo Liu <[email protected]>
> >
> > MMIO device passthrough enables an OS in a virtual machine to directly
> > access a MMIO device in the host. It promises almost the native
> > performance, which is required in performance-critical scenarios of
> > ACRN.
> >
> > HSM provides the following ioctls:
> > - Assign - ACRN_IOCTL_ASSIGN_MMIODEV
> > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > inform the hypervisor to assign a MMIO device to a User VM.
> >
> > - De-assign - ACRN_IOCTL_DEASSIGN_PCIDEV
> > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > inform the hypervisor to de-assign a MMIO device from a User VM.
> >
> > Signed-off-by: Shuo Liu <[email protected]>
> > Signed-off-by: Fei Li <[email protected]>
>
> Where is the userspace code that uses this new api?
Hi Greg
An ACRN userspace application (like QEMU) would uses these new apis in
https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L562
and
https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L568
thanks.
>
> thanks,
>
> greg k-h
On Thu, Aug 26, 2021 at 09:38:38AM +0800, Li Fei1 wrote:
> On Wed, Aug 25, 2021 at 10:43:29PM +0200, Greg KH wrote:
> > On Wed, Aug 25, 2021 at 05:01:40PM +0800, Fei Li wrote:
> > > From: Shuo Liu <[email protected]>
> > >
> > > MMIO device passthrough enables an OS in a virtual machine to directly
> > > access a MMIO device in the host. It promises almost the native
> > > performance, which is required in performance-critical scenarios of
> > > ACRN.
> > >
> > > HSM provides the following ioctls:
> > > - Assign - ACRN_IOCTL_ASSIGN_MMIODEV
> > > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > > inform the hypervisor to assign a MMIO device to a User VM.
> > >
> > > - De-assign - ACRN_IOCTL_DEASSIGN_PCIDEV
> > > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > > inform the hypervisor to de-assign a MMIO device from a User VM.
> > >
> > > Signed-off-by: Shuo Liu <[email protected]>
> > > Signed-off-by: Fei Li <[email protected]>
> >
> > Where is the userspace code that uses this new api?
>
> Hi Greg
>
> An ACRN userspace application (like QEMU) would uses these new apis in
> https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L562
> and
> https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L568
Please provide links in all of these patches when you resend them, to
the userspace code as well, so that we know how it all works together.
thanks,
greg k-h
On Fri, Aug 27, 2021 at 10:45:32AM +0200, Greg KH wrote:
> On Thu, Aug 26, 2021 at 09:38:38AM +0800, Li Fei1 wrote:
> > On Wed, Aug 25, 2021 at 10:43:29PM +0200, Greg KH wrote:
> > > On Wed, Aug 25, 2021 at 05:01:40PM +0800, Fei Li wrote:
> > > > From: Shuo Liu <[email protected]>
> > > >
> > > > MMIO device passthrough enables an OS in a virtual machine to directly
> > > > access a MMIO device in the host. It promises almost the native
> > > > performance, which is required in performance-critical scenarios of
> > > > ACRN.
> > > >
> > > > HSM provides the following ioctls:
> > > > - Assign - ACRN_IOCTL_ASSIGN_MMIODEV
> > > > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > > > inform the hypervisor to assign a MMIO device to a User VM.
> > > >
> > > > - De-assign - ACRN_IOCTL_DEASSIGN_PCIDEV
> > > > Pass data struct acrn_mmiodev from userspace to the hypervisor, and
> > > > inform the hypervisor to de-assign a MMIO device from a User VM.
> > > >
> > > > Signed-off-by: Shuo Liu <[email protected]>
> > > > Signed-off-by: Fei Li <[email protected]>
> > >
> > > Where is the userspace code that uses this new api?
> >
> > Hi Greg
> >
> > An ACRN userspace application (like QEMU) would uses these new apis in
> > https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L562
> > and
> > https://github.com/projectacrn/acrn-hypervisor/blob/master/devicemodel/core/vmmapi.c#L568
Hi Greg
Sorry to reply late.
>
> Please provide links in all of these patches when you resend them, to
> the userspace code as well, so that we know how it all works together.
>
Sure.
> thanks,
>
> greg k-h