genode/repos/base/include/cpu_session/cpu_session.h

398 lines
13 KiB
C
Raw Normal View History

2011-12-22 15:19:25 +00:00
/*
* \brief CPU (processing time) manager session interface
* \author Christian Helmuth
* \date 2006-06-27
*/
/*
2013-01-10 20:44:47 +00:00
* Copyright (C) 2006-2013 Genode Labs GmbH
2011-12-22 15:19:25 +00:00
*
* 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__CPU_SESSION__CPU_SESSION_H_
#define _INCLUDE__CPU_SESSION__CPU_SESSION_H_
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
#include <cpu_session/capability.h>
2011-12-22 15:19:25 +00:00
#include <base/stdint.h>
#include <base/exception.h>
#include <base/thread_state.h>
#include <base/rpc_args.h>
#include <base/signal.h>
#include <base/affinity.h>
2011-12-22 15:19:25 +00:00
#include <thread/capability.h>
#include <session/session.h>
#include <ram_session/ram_session.h>
#include <pd_session/pd_session.h>
2011-12-22 15:19:25 +00:00
namespace Genode { struct Cpu_session; }
struct Genode::Cpu_session : Session
{
/*********************
** Exception types **
*********************/
class Thread_creation_failed : public Exception { };
class State_access_failed : public Exception { };
class Quota_exceeded : public Thread_creation_failed { };
class Out_of_metadata : public Exception { };
static const char *service_name() { return "CPU"; }
enum { THREAD_NAME_LEN = 32 };
enum { PRIORITY_LIMIT = 1 << 16 };
enum { QUOTA_LIMIT_LOG2 = 15 };
enum { QUOTA_LIMIT = 1 << QUOTA_LIMIT_LOG2 };
enum { DEFAULT_PRIORITY = 0 };
/**
* Thread weight argument type for 'create_thread'
*/
struct Weight
{
enum { DEFAULT_WEIGHT = 10 };
size_t value = DEFAULT_WEIGHT;
Weight() { }
explicit Weight(size_t value) : value(value) { }
};
typedef String<THREAD_NAME_LEN> Name;
/**
* Physical quota configuration
*/
struct Quota;
virtual ~Cpu_session() { }
/**
* Create a new thread
*
* \param pd protection domain where the thread will be executed
* \param name name for the thread
* \param affinity CPU affinity, referring to the session-local
* affinity space
* \param weight CPU quota that shall be granted to the thread
* \param utcb base of the UTCB that will be used by the thread
* \return capability representing the new thread
* \throw Thread_creation_failed
* \throw Out_of_metadata
* \throw Quota_exceeded
*/
virtual Thread_capability create_thread(Capability<Pd_session> pd,
Name const &name,
Affinity::Location affinity,
Weight weight,
addr_t utcb = 0) = 0;
/**
* Get dataspace of the UTCB that is used by the specified thread
*/
virtual Ram_dataspace_capability utcb(Thread_capability thread) = 0;
/**
* Kill an existing thread
*
* \param thread capability of the thread to kill
*/
virtual void kill_thread(Thread_capability thread) = 0;
/**
* Modify instruction and stack pointer of thread - start the
* thread
*
* \param thread thread to start
* \param ip initial instruction pointer
* \param sp initial stack pointer
*
* \return 0 on success
*/
virtual int start(Thread_capability thread, addr_t ip, addr_t sp) = 0;
/**
* Pause the specified thread
*
* After calling this method, the execution of the thread can be
* continued by calling 'resume'.
*/
virtual void pause(Thread_capability thread) = 0;
/**
* Resume the specified thread
*/
virtual void resume(Thread_capability thread) = 0;
/**
* Cancel a currently blocking operation
*
* \param thread thread to unblock
*/
virtual void cancel_blocking(Thread_capability thread) = 0;
/**
* Get the current state of a specific thread
*
* \param thread targeted thread
* \return state of the targeted thread
* \throw State_access_failed
*/
virtual Thread_state state(Thread_capability thread) = 0;
/**
* Override the current state of a specific thread
*
* \param thread targeted thread
* \param state state that shall be applied
* \throw State_access_failed
*/
virtual void state(Thread_capability thread,
Thread_state const &state) = 0;
/**
* Register signal handler for exceptions of the specified thread
*
* If 'thread' is an invalid capability, the default exception
* handler for the CPU session is set. This handler is used for
* all threads that have no explicitly installed exception handler.
* The new default signal handler will take effect for threads
* created after the call.
*
* On Linux, this exception is delivered when the process triggers
* a SIGCHLD. On other platforms, this exception is delivered on
* the occurrence of CPU exceptions such as division by zero.
*/
virtual void exception_handler(Thread_capability thread,
Signal_context_capability handler) = 0;
/**
* Enable/disable single stepping for specified thread.
*
* Since this method is currently supported by a small number of
* platforms, we provide a default implementation
*
* \param thread thread to set into single step mode
* \param enable true = enable single-step mode; false = disable
*/
virtual void single_step(Thread_capability, bool) {}
/**
* Return affinity space of CPU nodes available to the CPU session
*
* The dimension of the affinity space as returned by this method
* represent the physical CPUs that are available.
*/
virtual Affinity::Space affinity_space() const = 0;
/**
* Define affinity of thread to one or multiple CPU nodes
*
* In the normal case, a thread is assigned to a single CPU.
* Specifying more than one CPU node is supposed to principally
* allow a CPU service to balance the load of threads among
* multiple CPUs.
*/
virtual void affinity(Thread_capability thread,
Affinity::Location affinity) = 0;
/**
* Translate generic priority value to kernel-specific priority levels
*
* \param pf_prio_limit maximum priority used for the kernel, must
* be power of 2
* \param prio generic priority value as used by the CPU
* session interface
* \param inverse order of platform priorities, if true
* 'pf_prio_limit' corresponds to the highest
* priority, otherwise it refers to the
* lowest priority.
* \return platform-specific priority value
*/
static unsigned scale_priority(unsigned pf_prio_limit, unsigned prio,
bool inverse = true)
{
/*
* Generic priority values are (0 is highest, 'PRIORITY_LIMIT'
* is lowest. On platforms where priority levels are defined
* the other way round, we have to invert the priority value.
*/
prio = inverse ? Cpu_session::PRIORITY_LIMIT - prio : prio;
/* scale value to platform priority range 0..pf_prio_limit */
return (prio*pf_prio_limit)/Cpu_session::PRIORITY_LIMIT;
}
/**
* Request trace control dataspace
*
* The trace-control dataspace is used to propagate tracing
* control information from core to the threads of a CPU session.
*
* The trace-control dataspace is accounted to the CPU session.
*/
virtual Dataspace_capability trace_control() = 0;
/**
* Request index of a trace control block for given thread
*
* The trace control dataspace contains the control blocks for
* all threads of the CPU session. Each thread gets assigned a
* different index by the CPU service.
*/
virtual unsigned trace_control_index(Thread_capability thread) = 0;
/**
* Request trace buffer for the specified thread
*
* The trace buffer is not accounted to the CPU session. It is
* owned by a TRACE session.
*/
virtual Dataspace_capability trace_buffer(Thread_capability thread) = 0;
/**
* Request trace policy
*
* The trace policy buffer is not accounted to the CPU session. It
* is owned by a TRACE session.
*/
virtual Dataspace_capability trace_policy(Thread_capability thread) = 0;
/**
* Define reference account for the CPU session
*
* \param cpu_session reference account
*
* \return 0 on success
*
* Each CPU session requires another CPU session as reference
* account to transfer quota to and from. The reference account can
* be defined only once.
*/
virtual int ref_account(Cpu_session_capability cpu_session) = 0;
/**
* Transfer quota to another CPU session
*
* \param cpu_session receiver of quota donation
* \param amount percentage of the session quota scaled up to
* the 'QUOTA_LIMIT' space
* \return 0 on success
*
* Quota can only be transfered if the specified CPU session is
* either the reference account for this session or vice versa.
*/
virtual int transfer_quota(Cpu_session_capability cpu_session,
size_t amount) = 0;
/**
* Return quota configuration of the session
*/
virtual Quota quota() = 0;
/**
* Scale up 'value' from its space with 'limit' to the 'QUOTA_LIMIT' space
*/
template<typename T = size_t>
static size_t quota_lim_upscale(size_t const value, size_t const limit) {
return ((T)value << Cpu_session::QUOTA_LIMIT_LOG2) / limit; }
/**
* Scale down 'value' from the 'QUOTA_LIMIT' space to a space with 'limit'
*/
template<typename T = size_t>
static size_t quota_lim_downscale(size_t const value, size_t const limit) {
return ((T)value * limit) >> Cpu_session::QUOTA_LIMIT_LOG2; }
/*****************************************
** Access to kernel-specific interface **
*****************************************/
/**
* Common base class of kernel-specific CPU interfaces
*/
struct Native_cpu { };
/**
* Return capability to kernel-specific CPU operations
*/
virtual Capability<Native_cpu> native_cpu() = 0;
/*********************
** RPC declaration **
*********************/
GENODE_RPC_THROW(Rpc_create_thread, Thread_capability, create_thread,
GENODE_TYPE_LIST(Thread_creation_failed, Out_of_metadata),
Capability<Pd_session>, Name const &, Affinity::Location,
Weight, addr_t);
GENODE_RPC(Rpc_utcb, Ram_dataspace_capability, utcb, Thread_capability);
GENODE_RPC(Rpc_kill_thread, void, kill_thread, Thread_capability);
GENODE_RPC(Rpc_start, int, start, Thread_capability, addr_t, addr_t);
GENODE_RPC(Rpc_pause, void, pause, Thread_capability);
GENODE_RPC(Rpc_resume, void, resume, Thread_capability);
GENODE_RPC(Rpc_cancel_blocking, void, cancel_blocking, Thread_capability);
GENODE_RPC_THROW(Rpc_get_state, Thread_state, state,
GENODE_TYPE_LIST(State_access_failed),
Thread_capability);
GENODE_RPC_THROW(Rpc_set_state, void, state,
GENODE_TYPE_LIST(State_access_failed),
Thread_capability, Thread_state const &);
GENODE_RPC(Rpc_exception_handler, void, exception_handler,
Thread_capability, Signal_context_capability);
GENODE_RPC(Rpc_single_step, void, single_step, Thread_capability, bool);
GENODE_RPC(Rpc_affinity_space, Affinity::Space, affinity_space);
GENODE_RPC(Rpc_affinity, void, affinity, Thread_capability, Affinity::Location);
GENODE_RPC(Rpc_trace_control, Dataspace_capability, trace_control);
GENODE_RPC(Rpc_trace_control_index, unsigned, trace_control_index, Thread_capability);
GENODE_RPC(Rpc_trace_buffer, Dataspace_capability, trace_buffer, Thread_capability);
GENODE_RPC(Rpc_trace_policy, Dataspace_capability, trace_policy, Thread_capability);
GENODE_RPC(Rpc_ref_account, int, ref_account, Cpu_session_capability);
GENODE_RPC(Rpc_transfer_quota, int, transfer_quota, Cpu_session_capability, size_t);
GENODE_RPC(Rpc_quota, Quota, quota);
GENODE_RPC(Rpc_native_cpu, Capability<Native_cpu>, native_cpu);
/*
* 'GENODE_RPC_INTERFACE' declaration done manually
*
* The number of RPC functions of this interface exceeds the maximum
* number of elements supported by 'Meta::Type_list'. Therefore, we
* construct the type list by hand using nested type tuples instead
* of employing the convenience macro 'GENODE_RPC_INTERFACE'.
*/
typedef Meta::Type_tuple<Rpc_create_thread,
Meta::Type_tuple<Rpc_utcb,
Meta::Type_tuple<Rpc_kill_thread,
Meta::Type_tuple<Rpc_start,
Meta::Type_tuple<Rpc_pause,
Meta::Type_tuple<Rpc_resume,
Meta::Type_tuple<Rpc_cancel_blocking,
Meta::Type_tuple<Rpc_set_state,
Meta::Type_tuple<Rpc_get_state,
Meta::Type_tuple<Rpc_exception_handler,
Meta::Type_tuple<Rpc_single_step,
Meta::Type_tuple<Rpc_affinity_space,
Meta::Type_tuple<Rpc_affinity,
Meta::Type_tuple<Rpc_trace_control,
Meta::Type_tuple<Rpc_trace_control_index,
Meta::Type_tuple<Rpc_trace_buffer,
Meta::Type_tuple<Rpc_trace_policy,
Meta::Type_tuple<Rpc_ref_account,
Meta::Type_tuple<Rpc_transfer_quota,
Meta::Type_tuple<Rpc_quota,
Meta::Type_tuple<Rpc_native_cpu,
Meta::Empty>
> > > > > > > > > > > > > > > > > > > > Rpc_functions;
};
struct Genode::Cpu_session::Quota
{
size_t super_period_us;
size_t us;
};
2011-12-22 15:19:25 +00:00
#endif /* _INCLUDE__CPU_SESSION__CPU_SESSION_H_ */