On 6/5/2023 12:49 PM, Alex Elder wrote:
On 5/9/23 3:47 PM, Elliot Berman wrote:
Introduce a framework for Gunyah userspace to install VM functions. VM
functions are optional interfaces to the virtual machine. vCPUs,
ioeventfs, and irqfds are examples of such VM functions and are
s/ioventfs/ioventfds/
Also, these aren't just examples of VM functions, they *are* the
VM functions implemented.
implemented in subsequent patches.
A generic framework is implemented instead of individual ioctls to
create vCPUs, irqfds, etc., in order to simplify the VM manager core
implementation and allow dynamic loading of VM function modules.
This also allows the set of VM functions to be extended without
updating the API (like it or not).
Signed-off-by: Elliot Berman <quic_eberman@xxxxxxxxxxx>
I have a few more comments, but this looks pretty good.
Reviewed-by: Alex Elder <elder@xxxxxxxxxx>
---
Documentation/virt/gunyah/vm-manager.rst | 18 ++
drivers/virt/gunyah/vm_mgr.c | 216 ++++++++++++++++++++++-
drivers/virt/gunyah/vm_mgr.h | 4 +
include/linux/gunyah_vm_mgr.h | 87 +++++++++
include/uapi/linux/gunyah.h | 18 ++
5 files changed, 340 insertions(+), 3 deletions(-)
create mode 100644 include/linux/gunyah_vm_mgr.h
diff --git a/Documentation/virt/gunyah/vm-manager.rst
b/Documentation/virt/gunyah/vm-manager.rst
index 50d8ae7fabcd..3b51bab9d793 100644
--- a/Documentation/virt/gunyah/vm-manager.rst
+++ b/Documentation/virt/gunyah/vm-manager.rst
@@ -17,6 +17,24 @@ sharing userspace memory with a VM is done via the
`GH_VM_SET_USER_MEM_REGION`_
ioctl. The VM itself is configured to use the memory region via the
devicetree.
+Gunyah Functions
+================
+
+Components of a Gunyah VM's configuration that need kernel
configuration are
+called "functions" and are built on top of a framework. Functions are
identified
+by a string and have some argument(s) to configure them. They are
typically
+created by the `GH_VM_ADD_FUNCTION`_ ioctl.
Is a function *type* (e.g., VCPU or ioeventfd) identified by a string?
Or a function *instance* (e.g. four VCPUs)? Or both?
Ah, this should be:
Function types are identified by an enum and have some argument(s)...
+
+Functions typically will always do at least one of these operations:
Typically, or always?
Hmm, I didn't want to use a more absolute term like "always" since it
implies to me that the framework forces this somehow. A VM function
wouldn't do much interesting if it weren't interacting with the VM and
resource tickets/IO handlers are the ways for functions to interact with
VMs.
I'll tweak the wording here.
+
+1. Create resource ticket(s). Resource tickets allow a function to
register
+ itself as the client for a Gunyah resource (e.g. doorbell or vCPU)
and
+ the function is given the pointer to the &struct gh_resource when the
+ VM is starting.
+
What I think this means is that tickets are used to allow functions
to be defined *before* the VM is actually started. So once it starts,
the functions get added. (I might have this slightly wrong, but in
any case I'm not sure the above sentence is very clear.)
I'm going to remove the "and the function is given the pointer to..."
since I agree it is a bit confusing. I think it'll be clearer for me to
put it in the resource ticket kerneldoc where there's context of the
populate() callback in the resource ticket. I'll mention there that the
populate() callback may not be made until the VM is started which could
be a while.
+2. Register IO handler(s). IO handlers allow a function to handle
stage-2 faults
+ from the virtual machine.
+
Sample Userspace VMM
====================
diff --git a/drivers/virt/gunyah/vm_mgr.c b/drivers/virt/gunyah/vm_mgr.c
index a800061f56bf..56464451b262 100644
--- a/drivers/virt/gunyah/vm_mgr.c
+++ b/drivers/virt/gunyah/vm_mgr.c
@@ -6,10 +6,13 @@
#define pr_fmt(fmt) "gh_vm_mgr: " fmt
#include <linux/anon_inodes.h>
+#include <linux/compat.h>
#include <linux/file.h>
#include <linux/gunyah_rsc_mgr.h>
+#include <linux/gunyah_vm_mgr.h>
#include <linux/miscdevice.h>
#include <linux/module.h>
+#include <linux/xarray.h>
#include <uapi/linux/gunyah.h>
@@ -17,6 +20,172 @@
static void gh_vm_free(struct work_struct *work);
+static DEFINE_XARRAY(gh_vm_functions);
+
+static void gh_vm_put_function(struct gh_vm_function *fn)
+{
+ module_put(fn->mod);
+}
+
+static struct gh_vm_function *gh_vm_get_function(u32 type)
+{
+ struct gh_vm_function *fn;
+ int r;
+
+ fn = xa_load(&gh_vm_functions, type);
+ if (!fn) {
+ r = request_module("ghfunc:%d", type);
+ if (r)
+ return ERR_PTR(r > 0 ? -r : r);
Almost all callers of request_module() simply ignore the
return value. What positive values are you expecting to
see here (and are you sure they're positive errno values)?
I can ignore the return value here, too, to follow the convention.
I had observed request_module can return modprobe's exit code.
+
+ fn = xa_load(&gh_vm_functions, type);
+ }
+
+ if (!fn || !try_module_get(fn->mod))
+ fn = ERR_PTR(-ENOENT);
+
+ return fn;
+}
. . .
