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

Re: [Xen-devel] [PATCH 1/1] Xen ARINC 653 Scheduler


  • To: Kathy Hadley <Kathy.Hadley@xxxxxxxxxxxxxxx>
  • From: George Dunlap <George.Dunlap@xxxxxxxxxxxxx>
  • Date: Tue, 4 May 2010 10:54:06 -0500
  • Cc: Keir.Fraser@xxxxxxxxxx, xen-devel@xxxxxxxxxxxxxxxxxxx
  • Delivery-date: Tue, 04 May 2010 08:55:07 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type; b=h3edUNpM40DVfs2D7+fZrBk4XN4mjtTt5edtYktltALdgKTYJS7Tn+AqVMGEoQpsZ1 f3C5CEwYO15yPWMX037P1t21PVHWMWpVF4TRyYq9Yz2cKMS4nnDcekUc8152HOcJMey3 ucfvQEWAyTLmio1RfGWxVuRtPe9cFMzsz+9XA=
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>

Kathy, thanks for your work on this.  Unfortunately, with the new cpupools feature, your actual scheduler will need a little more modification before it can be merged into -unstable.  Cpu pools carves up the cpus into several "pools", each of which has independent schedulers.  This means that schedulers need to make pointers to per-pool "global" structures, rather than having global static structures.  It should be a fairly straightforward transformation; you can see example transformations done on the credit and sedf schedulers.  Let us know if you need any help.

 -George

On Fri, Apr 16, 2010 at 9:18 AM, Kathy Hadley <Kathy.Hadley@xxxxxxxxxxxxxxx> wrote:

This patch adds an ARINC 653 scheduler to Xen.  This is a modification of an earlier patch ([Xen-devel] [PATCH 1/1] Xen ARINC653 scheduler).  In particular, it has been modified to use the new .adjust_global callback function, which was added in “[Xen-devel] [PATCH 1/1] Add .adjust_global callback”.

 

Thanks and regards,

 

 

cid:image001.jpg@01CAD7E0.50E45D70

Kathy Hadley
DornerWorks, Ltd.
Embedded Systems Engineering

3445 Lake Eastbrook Blvd SE
Grand Rapids, MI  49546

Direct: 616.389.6127

Tel:      616.245.8369

Fax:     616.245.8372

 

Kathy.Hadley@xxxxxxxxxxxxxxx

www.DornerWorks.com

cid:image002.jpg@01CAD7E0.50E45D70

Honored as one of the 2010 “Michigan 50 Companies to Watch”

 

diff -rupN a/tools/libxc/Makefile b/tools/libxc/Makefile

--- a/tools/libxc/Makefile            2010-04-13 10:49:37.573793000 -0400

+++ b/tools/libxc/Makefile         2010-04-14 17:49:26.952638000 -0400

@@ -17,6 +17,7 @@ CTRL_SRCS-y       += xc_physdev.c

 CTRL_SRCS-y       += xc_private.c

 CTRL_SRCS-y       += xc_sedf.c

 CTRL_SRCS-y       += xc_csched.c

+CTRL_SRCS-y       += xc_arinc653.c

 CTRL_SRCS-y       += xc_tbuf.c

 CTRL_SRCS-y       += xc_pm.c

 CTRL_SRCS-y       += xc_cpu_hotplug.c

diff -rupN a/tools/libxc/xc_arinc653.c b/tools/libxc/xc_arinc653.c

--- a/tools/libxc/xc_arinc653.c     1969-12-31 19:00:00.000000000 -0500

+++ b/tools/libxc/xc_arinc653.c 2010-04-14 17:49:26.952638000 -0400

@@ -0,0 +1,28 @@

+/****************************************************************************

+ * (C) 2010 - DornerWorks, Ltd <DornerWorks.com>

+ ****************************************************************************

+ *

+ *        File: xc_arinc653.c

+ *      Author: Josh Holtrop <DornerWorks.com>

+ *

+ * Description: XC Interface to the ARINC 653 scheduler

+ *

+ */

+

+#include "xc_private.h"

+

+int

+xc_sched_arinc653_sched_set(

+    int xc_handle,

+    xen_domctl_sched_arinc653_schedule_t * sched)

+{

+    DECLARE_DOMCTL;

+

+    domctl.cmd = XEN_DOMCTL_scheduler_op;

+    domctl.domain = (domid_t) 0;

+    domctl.u.scheduler_op.sched_id = XEN_SCHEDULER_ARINC653;

+    domctl.u.scheduler_op.cmd = XEN_DOMCTL_SCHEDOP_put_global_info;

+    set_xen_guest_handle(domctl.u.scheduler_op.u.arinc653.schedule, sched);

+

+    return do_domctl(xc_handle, &domctl);

+}

diff -rupN a/tools/libxc/xenctrl.h b/tools/libxc/xenctrl.h

--- a/tools/libxc/xenctrl.h            2010-04-13 10:49:37.573793000 -0400

+++ b/tools/libxc/xenctrl.h         2010-04-14 17:49:26.952638000 -0400

@@ -476,6 +476,16 @@ int xc_sched_credit_domain_get(int xc_ha

                                struct xen_domctl_sched_credit *sdom);

 

 /**

+ * This function sets the global ARINC 653 schedule.

+ *

+ * @parm xc_handle a handle to an open hypervisor interface

+ * @parm sched a pointer to the new ARINC 653 schedule

+ * return 0 on success

+ */

+int xc_sched_arinc653_sched_set(int xc_handle,

+                                xen_domctl_sched_arinc653_schedule_t * sched);

+

+/**

  * This function sends a trigger to a domain.

  *

  * @parm xc_handle a handle to an open hypervisor interface

diff -rupN a/xen/common/Makefile b/xen/common/Makefile

--- a/xen/common/Makefile         2010-04-13 10:49:37.573793000 -0400

+++ b/xen/common/Makefile     2010-04-13 13:00:31.651749000 -0400

@@ -14,6 +14,7 @@ obj-y += page_alloc.o

 obj-y += rangeset.o

 obj-y += sched_credit.o

 obj-y += sched_sedf.o

+obj-y += sched_arinc653.o

 obj-y += schedule.o

 obj-y += shutdown.o

 obj-y += softirq.o

diff -rupN a/xen/common/sched_arinc653.c b/xen/common/sched_arinc653.c

--- a/xen/common/sched_arinc653.c       1969-12-31 19:00:00.000000000 -0500

+++ b/xen/common/sched_arinc653.c    2010-04-14 18:13:26.163404000 -0400

@@ -0,0 +1,590 @@

+/*

+ * File: sched_arinc653.c

+ * Copyright (c) 2010, DornerWorks, Ltd. <DornerWorks.com>

+ *

+ * Description:

+ *   This file provides an ARINC653-compatible scheduling algorithm

+ *   for use in Xen.

+ *

+ * This program is free software; you can redistribute it and/or modify it

+ * under the terms of the GNU General Public License as published by the Free

+ * software Foundation; either version 2 of the License, or (at your option)

+ * any later version.

+ *

+ * This program is distributed in the hope that it will be useful,

+ * but WITHOUT ANY WARRANTY; without even the implied warranty of

+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

+ * See the GNU General Public License for more details.

+ */

+

+

+/**************************************************************************

+ * Includes                                                               *

+ *************************************************************************/

+#include <xen/lib.h>

+#include <xen/sched.h>

+#include <xen/sched-if.h>

+#include <xen/timer.h>

+#include <xen/softirq.h>

+#include <xen/time.h>

+#include <xen/errno.h>

+#include <xen/list.h>

+#include <public/domctl.h>          /* ARINC653_MAX_DOMAINS_PER_SCHEDULE */

+#include <xen/guest_access.h>

+

+

+/**************************************************************************

+ * Private Macros                                                         *

+ **************************************************************************/

+

+/**

+ * Retrieve the idle VCPU for a given physical CPU

+ */

+#define IDLETASK(cpu)  ((struct vcpu *) per_cpu(schedule_data, (cpu)).idle)

+

+/**

+ * Return a pointer to the ARINC 653-specific scheduler data information

+ * associated with the given VCPU (vc)

+ */

+#define AVCPU(vc) ((arinc653_vcpu_t *)(vc)->sched_priv)

+

+/**************************************************************************

+ * Private Type Definitions                                               *

+ **************************************************************************/

+

+/**

+ * The sched_entry_t structure holds a single entry of the

+ * ARINC 653 schedule.

+ */

+typedef struct sched_entry_s

+{

+    /* dom_handle holds the handle ("UUID") for the domain that this

+     * schedule entry refers to. */

+    xen_domain_handle_t dom_handle;

+    /* vcpu_id holds the VCPU number for the VCPU that this schedule

+     * entry refers to. */

+    int                 vcpu_id;

+    /* runtime holds the number of nanoseconds that the VCPU for this

+     * schedule entry should be allowed to run per major frame. */

+    s_time_t            runtime;

+    /* vc holds a pointer to the Xen VCPU structure */

+    struct vcpu *       vc;

+} sched_entry_t;

+

+/**

+ * The arinc653_vcpu_t structure holds ARINC 653-scheduler-specific

+ * information for all non-idle VCPUs

+ */

+typedef struct arinc653_vcpu_s

+{

+    /* vc points to Xen's struct vcpu so we can get to it from an

+     * arinc653_vcpu_t pointer. */

+    struct vcpu *       vc;

+    /* awake holds whether the VCPU has been woken with vcpu_wake() */

+    bool_t              awake;

+    /* list holds the linked list information for the list this VCPU

+     * is stored in */

+    struct list_head    list;

+} arinc653_vcpu_t;

+

+

+/**************************************************************************

+ * Global Data                                                            *

+ **************************************************************************/

+

+/**

+ * This array holds the active ARINC 653 schedule.

+ *

+ * When the system tries to start a new VCPU, this schedule is scanned

+ * to look for a matching (handle, VCPU #) pair. If both the handle ("UUID")

+ * and VCPU number match, then the VCPU is allowed to run. Its run time

+ * (per major frame) is given in the third entry of the schedule.

+ */

+static sched_entry_t arinc653_schedule[ARINC653_MAX_DOMAINS_PER_SCHEDULE] = {

+    { "", 0, MILLISECS(10), NULL }

+};

+

+/**

+ * This variable holds the number of entries that are valid in

+ * the arinc653_schedule table.

+ *

+ * This is not necessarily the same as the number of domains in the

+ * schedule. A domain could be listed multiple times within the schedule,

+ * or a domain with multiple VCPUs could have a different

+ * schedule entry for each VCPU.

+ *

+ * A value of 1 means that only 1 domain (Dom0) will initially be started.

+ */

+static int num_schedule_entries = 1;

+

+/**

+ * arinc653_major_frame holds the major frame time for the ARINC 653 schedule.

+ */

+static s_time_t arinc653_major_frame = MILLISECS(10);

+

+/**

+ * next_major_frame holds the time that the next major frame starts

+ */

+static s_time_t next_major_frame = 0;

+

+/**

+ * vcpu_list holds pointers to all Xen VCPU structures for iterating through

+ */

+static LIST_HEAD(vcpu_list);

+

+/**************************************************************************

+ * Scheduler functions                                                    *

+ **************************************************************************/

+

+/**

+ * This function compares two domain handles.

+ *

+ * @param h1        Pointer to handle 1

+ * @param h2        Pointer to handle 2

+ *

+ * @return          <ul>

+ *                  <li> <0:  handle 1 is less than handle 2

+ *                  <li>  0:  handle 1 is equal to handle 2

+ *                  <li> >0:  handle 1 is greater than handle 2

+ *                  </ul>

+ */

+static int dom_handle_cmp(const xen_domain_handle_t h1,

+                          const xen_domain_handle_t h2)

+{

+    return memcmp(h1, h2, sizeof(xen_domain_handle_t));

+} /* end dom_handle_cmp */

+

+/**

+ * This function searches the vcpu list to find a VCPU that matches

+ * the domain handle and VCPU ID specified.

+ *

+ * @param handle    Pointer to handler

+ * @param vcpu_id   VCPU ID

+ *

+ * @return          <ul>

+ *                  <li> Pointer to the matching VCPU if one is found

+ *                  <li> NULL otherwise

+ *                  </ul>

+ */

+static struct vcpu * find_vcpu(xen_domain_handle_t handle, int vcpu_id)

+{

+    arinc653_vcpu_t * avcpu; /* loop index variable */

+    struct vcpu * vc = NULL;

+

+    /* loop through the vcpu_list looking for the specified VCPU */

+    list_for_each_entry(avcpu, &vcpu_list, list)

+    {

+        /* If the handles & VCPU IDs match, we've found a matching VCPU */

+        if ((dom_handle_cmp(avcpu->vc->domain->handle, handle) == 0)

+             && (vcpu_id == avcpu->vc->vcpu_id))

+        {

+            vc = avcpu->vc;

+            /*

+             * "break" statement used instead of loop control variable because

+             * the macro used for this loop does not support using loop control

+             * variables

+             */

+            break;

+        }

+    }

+

+    return vc;

+} /* end find_vcpu */

+

+/**

+ * This function updates the pointer to the Xen VCPU structure for each entry in

+ * the ARINC 653 schedule.

+ *

+ * @param           <None>

+ * @return          <None>

+ */

+static void update_schedule_vcpus(void)

+{

+    /* Loop through the number of entries in the schedule */

+    for (int i = 0; i < num_schedule_entries; i++)

+    {

+        /* Update the pointer to the Xen VCPU structure for the current entry */

+        arinc653_schedule[i].vc =

+            find_vcpu(arinc653_schedule[i].dom_handle,

+                      arinc653_schedule[i].vcpu_id);

+    }

+} /* end update_schedule_vcpus */

+

+/**

+ * This function is called by the arinc653_adjust_global scheduler

+ * callback function in response to a domain control hypercall with

+ * a scheduler operation.

+ *

+ * The parameter schedule is set to be the address of a local variable from

+ * within arinc653_adjust_global(), so it is guaranteed to not be NULL.

+ *

+ * @param schedule  Pointer to the new ARINC 653 schedule.

+ *

+ * @return          <ul>

+ *                  <li> 0 = success

+ *                  <li> !0 = error

+ *                  </ul>

+ */

+static int arinc653_sched_set(xen_domctl_sched_arinc653_schedule_t * schedule)

+{

+    int ret = 0;

+    s_time_t total_runtime = 0;

+    bool_t found_dom0 = 0;

+    const static xen_domain_handle_t dom0_handle = {0};

+

+    /* check for valid major frame and number of schedule entries */

+    if ( (schedule->major_frame <= 0)

+      || (schedule->num_sched_entries < 1)

+      || (schedule->num_sched_entries > ARINC653_MAX_DOMAINS_PER_SCHEDULE) )

+    {

+        ret = -EINVAL;

+    }

+

+    if (ret == 0)

+    {

+        for (int i = 0; i < schedule->num_sched_entries; i++)

+        {

+            /*

+             * look for domain 0 handle - every schedule must contain

+             * some time for domain 0 to run

+             */

+            if (dom_handle_cmp(schedule->sched_entries[i].dom_handle,

+                               dom0_handle) == 0)

+            {

+                found_dom0 = 1;

+            }

+

+            /* check for a valid VCPU ID and run time */

+            if ( (schedule->sched_entries[i].vcpu_id < 0)

+              || (schedule->sched_entries[i].runtime <= 0) )

+            {

+                ret = -EINVAL;

+            }

+            else

+            {

+                /* Add this entry's run time to total run time */

+                total_runtime += schedule->sched_entries[i].runtime;

+            }

+        } /* end loop through schedule entries */

+    }

+

+    if (ret == 0)

+    {

+        /* error if the schedule doesn't contain a slot for domain 0 */

+        if (found_dom0 == 0)

+        {

+            ret = -EINVAL;

+        }

+    }

+

+    if (ret == 0)

+    {

+        /*

+         * error if the major frame is not large enough to run all entries

+         * as indicated by comparing the total run time to the major frame

+         * length

+         */

+        if (total_runtime > schedule->major_frame)

+        {

+            ret = -EINVAL;

+        }

+    }

+

+    if (ret == 0)

+    {

+        /* copy the new schedule into place */

+        num_schedule_entries = schedule->num_sched_entries;

+        arinc653_major_frame = schedule->major_frame;

+        for (int i = 0; i < num_schedule_entries; i++)

+        {

+            memcpy(arinc653_schedule[i].dom_handle,

+                   schedule->sched_entries[i].dom_handle,

+                   sizeof(arinc653_schedule[i].dom_handle));

+            arinc653_schedule[i].vcpu_id = schedule->sched_entries[i].vcpu_id;

+            arinc653_schedule[i].runtime = schedule->sched_entries[i].runtime;

+        }

+        update_schedule_vcpus();

+

+        /*

+         * The newly-installed schedule takes effect immediately.

+         * We do not even wait for the current major frame to expire.

+         *

+         * Signal a new major frame to begin. The next major frame

+         * is set up by the do_schedule callback function when it

+         * is next invoked.

+         */

+        next_major_frame = NOW();

+    }

+

+    return ret;

+} /* end arinc653_sched_set */

+

+/**

+ * Xen scheduler callback function to adjust global scheduling parameters

+ *

+ * @param op    Pointer to the domain control scheduler operation structure

+ *

+ * @return          <ul>

+ *                  <li> 0 for success

+ *                  <li> !0 if there is an error

+ *                  </ul>

+ */

+static int arinc653_adjust_global(struct xen_domctl_scheduler_op * op)

+{

+    int ret = -1;

+    xen_domctl_sched_arinc653_schedule_t new_sched;

+

+    if (op->cmd == XEN_DOMCTL_SCHEDOP_put_global_info)

+    {

+        if (copy_from_guest(&new_sched, op->u.arinc653.schedule, 1) != 0)

+        {

+            ret = -EFAULT;

+        }

+        else

+        {

+            ret = arinc653_sched_set(&new_sched);

+        }

+    }

+

+    return ret;

+} /* end arinc653_adjust_global */

+

+/**

+ * Xen scheduler callback function to initialize a virtual CPU (VCPU).

+ *

+ * @param v         Pointer to the VCPU structure

+ *

+ * @return          <ul>

+ *                  <li> 0 if the VCPU is allowed to run

+ *                  <li> !0 if there is an error

+ *                  </ul>

+ */

+static int arinc653_init_vcpu(struct vcpu * v)

+{

+    int ret = -1;

+

+    if (is_idle_vcpu(v))

+    {

+        /*

+         * The idle VCPU is created by Xen to run when no domains

+         * are runnable or require CPU time.

+         * It is similar to an "idle task" or "halt loop" process

+         * in an operating system.

+         * We do not track any scheduler information for the idle VCPU.

+         */

+        v->sched_priv = NULL;

+        ret = 0;

+    }

+    else

+    {

+        /*

+         * Allocate memory for the ARINC 653-specific scheduler data information

+         * associated with the given VCPU (vc).

+         */

+        v->sched_priv = xmalloc(arinc653_vcpu_t);

+        if (AVCPU(v) != NULL)

+        {

+            /*

+             * Initialize our ARINC 653 scheduler-specific information

+             * for the VCPU.

+             * The VCPU starts "asleep."

+             * When Xen is ready for the VCPU to run, it will call

+             * the vcpu_wake scheduler callback function and our

+             * scheduler will mark the VCPU awake.

+             */

+            AVCPU(v)->vc = v;

+            AVCPU(v)->awake = 0;

+            list_add(&AVCPU(v)->list, &vcpu_list);

+            ret = 0;

+            update_schedule_vcpus();

+        }

+    }

+

+    return ret;

+} /* end arinc653_init_vcpu */

+

+/**

+ * Xen scheduler callback function to remove a VCPU

+ *

+ * @param v         Pointer to the VCPU structure to remove

+ *

+ * @return          <None>

+ */

+static void arinc653_destroy_vcpu(struct vcpu * v)

+{

+    if (AVCPU(v) != NULL)

+    {

+        /* remove the VCPU from whichever list it is on */

+        list_del(&AVCPU(v)->list);

+        /* free the arinc653_vcpu structure */

+        xfree(AVCPU(v));

+        update_schedule_vcpus();

+    }

+} /* end arinc653_destroy_vcpu */

+

+/**

+ * Xen scheduler callback function to select a VCPU to run.

+ * This is the main scheduler routine.

+ *

+ * @param t         Current time

+ *

+ * @return          Time slice and address of the VCPU structure for the chosen

+ *                  domain

+ */

+static struct task_slice arinc653_do_schedule(s_time_t t)

+{

+    struct task_slice ret;                      /* hold the chosen domain */

+    struct vcpu * new_task = NULL;

+    static int sched_index = 0;

+    static s_time_t last_major_frame;

+    static s_time_t last_switch_time;

+    static s_time_t next_switch_time;

+

+    if (t >= next_major_frame)

+    {

+        /* time to enter a new major frame

+         * the first time this function is called, this will be true */

+        sched_index = 0;

+        last_major_frame = last_switch_time = t;

+        next_major_frame = t + arinc653_major_frame;

+    }

+    else if (t >= next_switch_time)

+    {

+        /* time to switch to the next domain in this major frame */

+        sched_index++;

+        last_switch_time = next_switch_time;

+    }

+

+    /*

+     * If there are more domains to run in the current major frame, set

+     * next_switch_time equal to the last switch time + this domain's run time.

+     * Otherwise, set next_switch_time equal to the start of the next major

+     * frame.

+     */

+    next_switch_time = (sched_index < num_schedule_entries)

+                     ? last_switch_time + arinc653_schedule[sched_index].runtime

+                     : next_major_frame;

+

+    /*

+     * If there are more domains to run in the current major frame, set

+     * new_task equal to the address of next domain's VCPU structure.

+     * Otherwise, set new_task equal to the address of the idle task's VCPU

+     * structure.

+     */

+    new_task = (sched_index < num_schedule_entries)

+                   ? arinc653_schedule[sched_index].vc

+                   : IDLETASK(0);

+

+    /* Check to see if the new task can be run (awake & runnable). */

+    if (!((new_task != NULL)

+            && AVCPU(new_task)->awake

+            && vcpu_runnable(new_task)) )

+    {

+        new_task = IDLETASK(0);

+    }

+    BUG_ON(new_task == NULL);

+

+    /*

+     * Check to make sure we did not miss a major frame.

+     * This is a good test for robust partitioning.

+     */

+    BUG_ON(t >= next_major_frame);

+

+    /*

+     * Return the amount of time the next domain has to run and the address

+     * of the selected task's VCPU structure.

+     */

+    ret.time = next_switch_time - t;

+    ret.task = new_task;

+

+    BUG_ON(ret.time <= 0);

+

+    return ret;

+} /* end arinc653_do_schedule */

+

+/**

+ * Xen scheduler callback function to select a CPU for the VCPU to run on

+ *

+ * @param v         Pointer to the VCPU structure for the current domain

+ *

+ * @return          Number of selected physical CPU

+ */

+static int arinc653_pick_cpu(struct vcpu * v)

+{

+    /* this implementation only supports one physical CPU */

+    return 0;

+} /* end arinc653_pick_cpu */

+

+/**

+ * Xen scheduler callback function to wake up a VCPU

+ *

+ * @param vc        Pointer to the VCPU structure for the current domain

+ *

+ * @return          <None>

+ */

+static void arinc653_vcpu_wake(struct vcpu * vc)

+{

+    /* boolean flag to indicate first run */

+    static bool_t dont_raise_softirq = 0;

+

+    if (AVCPU(vc) != NULL)  /* check that this is a VCPU we are tracking */

+    {

+        AVCPU(vc)->awake = 1;

+    }

+

+    /* the first time the vcpu_wake function is called, we should raise

+     * a softirq to invoke the do_scheduler callback */

+    if (!dont_raise_softirq)

+    {

+        cpu_raise_softirq(vc->processor, SCHEDULE_SOFTIRQ);

+        dont_raise_softirq = 1;

+    }

+} /* end arinc653_vcpu_wake */

+

+/**

+ * Xen scheduler callback function to sleep a VCPU

+ *

+ * @param vc        Pointer to the VCPU structure for the current domain

+ *

+ * @return          <None>

+ */

+static void arinc653_vcpu_sleep(struct vcpu * vc)

+{

+    if (AVCPU(vc) != NULL)  /* check that this is a VCPU we are tracking */

+    {

+        AVCPU(vc)->awake = 0;

+    }

+

+    /* if the VCPU being put to sleep is the same one that is currently

+     * running, raise a softirq to invoke the scheduler to switch domains */

+    if (per_cpu(schedule_data, vc->processor).curr == vc)

+    {

+        cpu_raise_softirq(vc->processor, SCHEDULE_SOFTIRQ);

+    }

+} /* end arinc653_vcpu_sleep */

+

+/**

+ * This structure defines our scheduler for Xen.

+ * The entries tell Xen where to find our scheduler-specific

+ * callback functions.

+ * The symbol must be visible to the rest of Xen at link time.

+ */

+struct scheduler sched_arinc653_def = {

+    .name           = "ARINC 653 Scheduler",

+    .opt_name       = "arinc653",

+    .sched_id       = XEN_SCHEDULER_ARINC653,

+

+    .init_domain    = NULL,

+    .destroy_domain = NULL,

+

+    .init_vcpu      = arinc653_init_vcpu,

+    .destroy_vcpu   = arinc653_destroy_vcpu,

+

+    .do_schedule    = arinc653_do_schedule,

+    .pick_cpu       = arinc653_pick_cpu,

+    .dump_cpu_state = NULL,

+    .sleep          = arinc653_vcpu_sleep,

+    .wake           = arinc653_vcpu_wake,

+    .adjust         = NULL,

+    .adjust_global  = arinc653_adjust_global,

+};

diff -rupN a/xen/common/schedule.c b/xen/common/schedule.c

--- a/xen/common/schedule.c     2010-04-14 10:57:11.262796000 -0400

+++ b/xen/common/schedule.c  2010-04-14 16:40:21.543608000 -0400

@@ -7,7 +7,8 @@

  *        File: common/schedule.c

  *      Author: Rolf Neugebauer & Keir Fraser

  *              Updated for generic API by Mark Williamson

- *

+ *              ARINC653 scheduler added by DornerWorks <DornerWorks.com>

+ *

  * Description: Generic CPU scheduling code

  *              implements support functionality for the Xen scheduler API.

  *

@@ -56,9 +57,11 @@ DEFINE_PER_CPU(struct schedule_data, sch

 

 extern const struct scheduler sched_sedf_def;

 extern const struct scheduler sched_credit_def;

+extern const struct scheduler sched_arinc653_def;

 static const struct scheduler *__initdata schedulers[] = {

     &sched_sedf_def,

     &sched_credit_def,

+    &sched_arinc653_def,

     NULL

 };

 

diff -rupN a/xen/include/public/domctl.h b/xen/include/public/domctl.h

--- a/xen/include/public/domctl.h 2010-04-14 10:57:11.262796000 -0400

+++ b/xen/include/public/domctl.h         2010-04-14 16:40:21.543608000 -0400

@@ -23,6 +23,8 @@

  *

  * Copyright (c) 2002-2003, B Dragovic

  * Copyright (c) 2002-2006, K Fraser

+ *

+ * ARINC653 Scheduler type added by DornerWorks <DornerWorks.com>.

  */

 

 #ifndef __XEN_PUBLIC_DOMCTL_H__

@@ -303,11 +305,43 @@ DEFINE_XEN_GUEST_HANDLE(xen_domctl_max_v

 /* Scheduler types. */

 #define XEN_SCHEDULER_SEDF     4

 #define XEN_SCHEDULER_CREDIT   5

+#define XEN_SCHEDULER_ARINC653 6

+

 /* Set or get info? */

 #define XEN_DOMCTL_SCHEDOP_putinfo 0

 #define XEN_DOMCTL_SCHEDOP_getinfo 1

 #define XEN_DOMCTL_SCHEDOP_put_global_info 2

 #define XEN_DOMCTL_SCHEDOP_get_global_info 3

+

+/*

+ * This structure is used to pass a new ARINC653 schedule from a

+ * privileged domain (ie dom0) to Xen.

+ */

+#define ARINC653_MAX_DOMAINS_PER_SCHEDULE   64

+struct xen_domctl_sched_arinc653_schedule {

+    /* major_frame holds the time for the new schedule's major frame

+     * in nanoseconds. */

+    int64_t     major_frame;

+    /* num_sched_entries holds how many of the entries in the

+     * sched_entries[] array are valid. */

+    uint8_t     num_sched_entries;

+    /* The sched_entries array holds the actual schedule entries. */

+    struct {

+        /* dom_handle must match a domain's UUID */

+        xen_domain_handle_t dom_handle;

+        /* If a domain has multiple VCPUs, vcpu_id specifies which one

+         * this schedule entry applies to. It should be set to 0 if

+         * there is only one VCPU for the domain. */

+        int                 vcpu_id;

+        /* runtime specifies the amount of time that should be allocated

+         * to this VCPU per major frame. It is specified in nanoseconds */

+        int64_t             runtime;

+    } sched_entries[ARINC653_MAX_DOMAINS_PER_SCHEDULE];

+};

+typedef struct xen_domctl_sched_arinc653_schedule

+    xen_domctl_sched_arinc653_schedule_t;

+DEFINE_XEN_GUEST_HANDLE(xen_domctl_sched_arinc653_schedule_t);

+

 struct xen_domctl_scheduler_op {

     uint32_t sched_id;  /* XEN_SCHEDULER_* */

     uint32_t cmd;       /* XEN_DOMCTL_SCHEDOP_* */

@@ -323,6 +357,9 @@ struct xen_domctl_scheduler_op {

             uint16_t weight;

             uint16_t cap;

         } credit;

+        struct xen_domctl_sched_arinc653 {

+            XEN_GUEST_HANDLE(xen_domctl_sched_arinc653_schedule_t) schedule;

+        } arinc653;

     } u;

 };

 typedef struct xen_domctl_scheduler_op xen_domctl_scheduler_op_t;

 

 

 

 

 


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.