sys/kern/sys_vmm.c

   1 /*
   2  * Copyright (c) 2013 The DragonFly Project.  All rights reserved.
   3  *
   4  * This code is derived from software contributed to The DragonFly Project
   5  * by Mihai Carabas <mihai.carabas@gmail.com>
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  *
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in
  15  *    the documentation and/or other materials provided with the
  16  *    distribution.
  17  * 3. Neither the name of The DragonFly Project nor the names of its
  18  *    contributors may be used to endorse or promote products derived
  19  *    from this software without specific, prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
  25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
  27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
  29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
  30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
  31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  32  * SUCH DAMAGE.
  33  */
  34
  35 #include <sys/param.h>
  36 #include <sys/systm.h>
  37 #include <sys/sysproto.h>
  38 #include <sys/proc.h>
  39 #include <sys/user.h>
  40 #include <sys/wait.h>
  41 #include <sys/vmm.h>
  42
  43 #include <sys/thread2.h>
  44 #include <sys/spinlock2.h>
  45
  46 #include <machine/cpu.h>
  47 #include <machine/vmm.h>
  48
  49 /*
  50  * vmm guest system call:
  51  * - init the calling thread structure
  52  * - prepare for running in non-root mode
  53  */
  54 int
  55 sys_vmm_guest_ctl(struct vmm_guest_ctl_args *uap)
  56 {
  57         int error = 0;
  58         struct vmm_guest_options options;
  59         struct trapframe *tf = uap->sysmsg_frame;
  60         unsigned long stack_limit = USRSTACK;
  61         unsigned char stack_page[PAGE_SIZE];
  62
  63         clear_quickret();
  64
  65         switch (uap->op) {
  66         case VMM_GUEST_RUN:
  67                 error = copyin(uap->options, &options,
  68                                sizeof(struct vmm_guest_options));
  69                 if (error) {
  70                         kprintf("%s: error copyin vmm_guest_options\n",
  71                                 __func__);
  72                         goto out;
  73                 }
  74
  75                 while(stack_limit > tf->tf_sp) {
  76                         stack_limit -= PAGE_SIZE;
  77                         options.new_stack -= PAGE_SIZE;
  78
  79                         error = copyin((const void *)stack_limit,
  80                                        (void *)stack_page, PAGE_SIZE);
  81                         if (error) {
  82                                 kprintf("%s: error copyin stack\n",
  83                                         __func__);
  84                                 goto out;
  85                         }
  86
  87                         error = copyout((const void *)stack_page,
  88                                         (void *)options.new_stack, PAGE_SIZE);
  89                         if (error) {
  90                                 kprintf("%s: error copyout stack\n",
  91                                     __func__);
  92                                 goto out;
  93                         }
  94                 }
  95
  96                 bcopy(tf, &options.tf, sizeof(struct trapframe));
  97
  98                 error = vmm_vminit(&options);
  99                 if (error) {
 100                         if (error == ENODEV) {
 101                                 kprintf("%s: vmm_vminit failed - "
 102                                         "no VMM available \n", __func__);
 103                                 goto out;
 104                         }
 105                         kprintf("%s: vmm_vminit failed\n", __func__);
 106                         goto out_exit;
 107                 }
 108
 109                 generic_lwp_return(curthread->td_lwp, tf);
 110
 111                 error = vmm_vmrun();
 112
 113                 break;
 114         default:
 115                 kprintf("%s: INVALID op\n", __func__);
 116                 error = EINVAL;
 117                 goto out;
 118         }
 119 out_exit:
 120         exit1(W_EXITCODE(error, 0));
 121 out:
 122         return (error);
 123 }
 124
 125 /*
 126  * The remote IPI will force the cpu out of any VMM mode it is
 127  * in.  When combined with bumping pm_invgen we can ensure that
 128  * INVEPT will be called when it returns.
 129  */
 130 static void
 131 vmm_exit_vmm(void *dummy __unused)
 132 {
 133 }
 134
 135 /*
 136  * Swap the 64 bit value between *dstaddr and *srcaddr in a pmap-safe manner
 137  * and invalidate the tlb on all cpus the vkernel is running on.
 138  *
 139  * If dstaddr is NULL, just invalidate the tlb on the current cpu.
 140  *
 141  * v = *srcaddr
 142  * v = swap(dstaddr, v)
 143  * *dstaddr = v
 144  */
 145 int
 146 sys_vmm_guest_sync_addr(struct vmm_guest_sync_addr_args *uap)
 147 {
 148         int error = 0;
 149         cpulock_t olock;
 150         cpulock_t nlock;
 151         cpumask_t mask;
 152         struct proc *p = curproc;
 153         long v;
 154
 155         if (p->p_vmm == NULL)
 156                 return ENOSYS;
 157         if (uap->dstaddr == NULL)
 158                 return 0;
 159
 160         crit_enter_id("vmm_inval");
 161
 162         /*
 163          * Acquire CPULOCK_EXCL, spin while we wait.  This will prevent
 164          * any other cpu trying to use related VMMs to wait for us.
 165          */
 166         KKASSERT(CPUMASK_TESTMASK(p->p_vmm_cpumask, mycpu->gd_cpumask) == 0);
 167         for (;;) {
 168                 olock = p->p_vmm_cpulock & ~CPULOCK_EXCL;
 169                 cpu_ccfence();
 170                 nlock = olock | CPULOCK_EXCL;
 171                 if (atomic_cmpset_int(&p->p_vmm_cpulock, olock, nlock))
 172                         break;
 173                 lwkt_process_ipiq();
 174                 cpu_pause();
 175         }
 176
 177         /*
 178          * Wait for other cpu's to exit VMM mode (for this vkernel).  No
 179          * new cpus will enter VMM mode while we hold the lock.  New waiters
 180          * may turn-up though so the wakeup() later on has to be
 181          * unconditional.
 182          *
 183          * We must test on p_vmm_cpulock's counter, not the mask, because
 184          * VMM entries will set the mask bit unconditionally first
 185          * (interlocking our IPI below) and then conditionally bump the
 186          * counter.
 187          */
 188         if (olock & CPULOCK_CNTMASK) {
 189                 mask = p->p_vmm_cpumask;
 190                 CPUMASK_ANDMASK(mask, mycpu->gd_other_cpus);
 191                 lwkt_send_ipiq_mask(mask, vmm_exit_vmm, NULL);
 192                 while (p->p_vmm_cpulock & CPULOCK_CNTMASK) {
 193                         lwkt_process_ipiq();
 194                         cpu_pause();
 195                 }
 196         }
 197
 198 #ifndef _KERNEL_VIRTUAL
 199         /*
 200          * Ensure that any new entries into VMM mode using
 201          * vmm's managed under this process will issue a
 202          * INVEPT before resuming.
 203          */
 204         atomic_add_acq_long(&p->p_vmspace->vm_pmap.pm_invgen, 1);
 205 #endif
 206
 207         /*
 208          * Make the requested modification, wakeup any waiters.
 209          */
 210         v = fuword64(uap->srcaddr);
 211         v = swapu64(uap->dstaddr, v);
 212         suword64(uap->srcaddr, v);
 213
 214         /*
 215          * VMMs on remote cpus will not be re-entered until we
 216          * clear the lock.
 217          */
 218         atomic_clear_int(&p->p_vmm_cpulock, CPULOCK_EXCL);
 219 #if 0
 220         wakeup(&p->p_vmm_cpulock);
 221 #endif
 222
 223         crit_exit_id("vmm_inval");
 224
 225         return error;
 226 }