genode/repos/ports/include/vmm/vcpu_dispatcher.h

144 lines
4.0 KiB
C
Raw Normal View History

/*
* \brief Utilities for implementing VMMs on Genode/NOVA
* \author Norman Feske
* \date 2013-08-20
*/
/*
* Copyright (C) 2013 Genode Labs GmbH
*
* This file is part of the Genode OS framework, which is distributed
* under the terms of the GNU General Public License version 2.
*/
#ifndef _INCLUDE__VMM__VCPU_DISPATCHER_H_
#define _INCLUDE__VMM__VCPU_DISPATCHER_H_
namespace Vmm {
using namespace Genode;
template <class T>
class Vcpu_dispatcher;
}
/**
* Thread that handles virtualization events of a 'Vmm::Vcpu_thread'
*/
template <class T>
class Vmm::Vcpu_dispatcher : public T
{
private:
Cap_connection &_cap;
/**
* Portal entry point entered on virtualization events
*
* For each event type used as argument of the 'register_handler'
* function template, the compiler automatically generates a separate
* instance of this function. The sole purpose of this function is to
* call the 'Vcpu' member function corresponding to the event type.
*/
template <unsigned EV, typename DISPATCHER, void (DISPATCHER::*FUNC)()>
static void _portal_entry()
{
/* obtain this pointer of the event handler */
Genode::Thread_base *myself = Genode::Thread_base::myself();
DISPATCHER *vd = static_cast<DISPATCHER *>(myself);
/* call event-specific handler function */
(vd->*FUNC)();
/* continue execution of the guest */
Nova::reply(myself->stack_top());
}
public:
Vcpu_dispatcher(size_t stack_size, Cap_connection &cap,
Cpu_session * cpu_session,
Genode::Affinity::Location location)
:
thread API & CPU session: accounting of CPU quota In the init configuration one can configure the donation of CPU time via 'resource' tags that have the attribute 'name' set to "CPU" and the attribute 'quantum' set to the percentage of CPU quota that init shall donate. The pattern is the same as when donating RAM quota. ! <start name="test"> ! <resource name="CPU" quantum="75"/> ! </start> This would cause init to try donating 75% of its CPU quota to the child "test". Init and core do not preserve CPU quota for their own requirements by default as it is done with RAM quota. The CPU quota that a process owns can be applied through the thread constructor. The constructor has been enhanced by an argument that indicates the percentage of the programs CPU quota that shall be granted to the new thread. So 'Thread(33, "test")' would cause the backing CPU session to try to grant 33% of the programs CPU quota to the thread "test". By now, the CPU quota of a thread can't be altered after construction. Constructing a thread with CPU quota 0 doesn't mean the thread gets never scheduled but that the thread has no guaranty to receive CPU time. Such threads have to live with excess CPU time. Threads that already existed in the official repositories of Genode were adapted in the way that they receive a quota of 0. This commit also provides a run test 'cpu_quota' in base-hw (the only kernel that applies the CPU-quota scheme currently). The test basically runs three threads with different physical CPU quota. The threads simply count for 30 seconds each and the test then checks wether the counter values relate to the CPU-quota distribution. fix #1275
2014-10-16 09:15:46 +00:00
T(0, "vCPU dispatcher", stack_size),
_cap(cap)
{
using namespace Genode;
/* place the thread on CPU described by location object */
cpu_session->affinity(T::cap(), location);
/* request creation of a 'local' EC */
T::_tid.ec_sel = Native_thread::INVALID_INDEX - 1;
T::start();
}
template <typename X>
Vcpu_dispatcher(size_t stack_size, Cap_connection &cap,
Cpu_session * cpu_session,
Genode::Affinity::Location location,
X attr, void *(*start_routine) (void *), void *arg)
: T(attr, start_routine, arg, stack_size, "vCPU dispatcher", nullptr),
_cap(cap)
{
using namespace Genode;
/* place the thread on CPU described by location object */
cpu_session->affinity(T::cap(), location);
/* request creation of a 'local' EC */
T::_tid.ec_sel = Native_thread::INVALID_INDEX - 1;
T::start();
}
/**
* Register virtualization event handler
*/
template <unsigned EV, typename DISPATCHER, void (DISPATCHER::*FUNC)()>
bool register_handler(addr_t exc_base, Nova::Mtd mtd)
{
/*
* Let the compiler generate an instance of a portal entry
*/
void (*entry)() = &_portal_entry<EV, DISPATCHER, FUNC>;
/* Create the portal at the desired selector index */
_cap.rcv_window(exc_base + EV);
Native_capability thread_cap(T::tid().ec_sel);
Native_capability handler =
_cap.alloc(thread_cap, (Nova::mword_t)entry, mtd.value());
return handler.valid() && (exc_base + EV == handler.local_name());
}
/**
* Unused member of the 'Thread_base' interface
*
* Similarly to how 'Rpc_entrypoints' are handled, a 'Vcpu_dispatcher'
* comes with a custom initialization procedure, which does not call
* the thread's normal entry function. Instead, the thread's EC gets
* associated with several portals, each for handling a specific
* virtualization event.
*/
void entry() { }
/**
* Return capability selector of the VCPU's SM and EC
*
* The returned number corresponds to the VCPU's semaphore selector.
* The consecutive number corresponds to the EC. The number returned by
* this function is used by the VMM code as a unique identifier of the
* VCPU. I.e., it gets passed as arguments for 'MessageHostOp'
* operations.
*/
Nova::mword_t sel_sm_ec()
{
return T::tid().exc_pt_sel + Nova::SM_SEL_EC;
}
};
#endif /* _INCLUDE__VMM__VCPU_DISPATCHER_H_ */