Re: [PATCH v9 24/27] virt: gunyah: Add proxy-scheduled vCPUs

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On 2/7/2023 6:43 AM, Srinivas Kandagatla wrote:


On 20/01/2023 22:46, Elliot Berman wrote:
Gunyah allows host virtual machines to schedule guest virtual machines
and handle their MMIO accesses. vCPUs are presented to the host as a
Gunyah resource and represented to userspace as a Gunyah VM function.

Creating the vcpu VM function will create a file descriptor that:
  - can run an ioctl: GH_VCPU_RUN to schedule the guest vCPU until the
    next interrupt occurs on the host or when the guest vCPU can no
    longer be run.
  - can be mmap'd to share a gh_vcpu_run structure which can look up the
    reason why GH_VCPU_RUN returned and provide return values for MMIO
    access.

Co-developed-by: Prakruthi Deepak Heragu <quic_pheragu@xxxxxxxxxxx>
Signed-off-by: Prakruthi Deepak Heragu <quic_pheragu@xxxxxxxxxxx>
Signed-off-by: Elliot Berman <quic_eberman@xxxxxxxxxxx>
---
  Documentation/virt/gunyah/vm-manager.rst |  30 +-
  arch/arm64/gunyah/gunyah_hypercall.c     |  28 ++
  drivers/virt/gunyah/Kconfig              |  11 +
  drivers/virt/gunyah/Makefile             |   2 +
  drivers/virt/gunyah/gunyah_vcpu.c        | 358 +++++++++++++++++++++++
  drivers/virt/gunyah/vm_mgr.c             |  25 ++
  drivers/virt/gunyah/vm_mgr.h             |   1 +
  include/linux/gunyah.h                   |   7 +
  include/uapi/linux/gunyah.h              |  30 ++
  9 files changed, 490 insertions(+), 2 deletions(-)
  create mode 100644 drivers/virt/gunyah/gunyah_vcpu.c

...


diff --git a/include/uapi/linux/gunyah.h b/include/uapi/linux/gunyah.h
index ec8da6fde045..b4afb11f538a 100644
--- a/include/uapi/linux/gunyah.h
+++ b/include/uapi/linux/gunyah.h
@@ -53,9 +53,14 @@ struct gh_vm_dtb_config {
  #define GUNYAH_FUNCTION_NAME_SIZE        32
  #define GUNYAH_FUNCTION_MAX_ARG_SIZE        1024
+struct gh_fn_vcpu_arg {
+    __u32 vcpu_id;
+};
+
  struct gh_vm_function {
      char name[GUNYAH_FUNCTION_NAME_SIZE];
      union {
+        struct gh_device_vcpu_arg vcpu;
          char data[GUNYAH_FUNCTION_MAX_ARG_SIZE];
      };
  };
@@ -63,4 +68,29 @@ struct gh_vm_function {
  #define GH_VM_ADD_FUNCTION    _IOW(GH_IOCTL_TYPE, 0x4, struct gh_vm_function)   #define GH_VM_REMOVE_FUNCTION    _IOW(GH_IOCTL_TYPE, 0x7, struct gh_vm_function)
+/* for GH_VCPU_RUN, returned by mmap(vcpu_fd, offset=0) */
+struct gh_vcpu_run {
this looks unused, I dont see any reference to this.


This structure gets mapped into userspace from the vCPU file descriptor, similar principle to KVM.

Thanks,
Elliot

+    /* in */
+    __u8 immediate_exit;
+    __u8 padding1[7];
+
+    /* out */
+#define GH_VM_EXIT_UNKNOWN            0
+#define GH_VM_EXIT_MMIO               1
+    __u32 exit_reason;

If this is ment to go in any of the ioctl arguments then this suffers same issue of padding.

+
+    union {
+        /* GH_VM_EXIT_MMIO */
+        struct {
+            __u64 phys_addr;
+            __u8  data[8];
+            __u32 len;
+            __u8  is_write;
+        } mmio;
+    };
+};
+
+#define GH_VCPU_RUN        _IO(GH_IOCTL_TYPE, 0x5)
+#define GH_VCPU_MMAP_SIZE    _IO(GH_IOCTL_TYPE, 0x6)
+
  #endif



[Index of Archives]     [Kernel Newbies]     [Security]     [Netfilter]     [Bugtraq]     [Linux FS]     [Yosemite Forum]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]     [Linux Resources]

  Powered by Linux