]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/vm/dp_backing_file.c
xnu-2050.48.11.tar.gz
[apple/xnu.git] / bsd / vm / dp_backing_file.c
index e1ff49d7b048c1ae664a27813c59573a25b95d36..9df1b810effdd830ec465bda7c6dcfb67a778a8b 100644 (file)
@@ -1,46 +1,81 @@
 /*
- * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2010 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.
  * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.
+ * 
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
  * 
- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
+ */
+/*
+ * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
+ * support for mandatory and extensible security protections.  This notice
+ * is included in support of clause 2.2 (b) of the Apple Public License,
+ * Version 2.0.
  */
 
-#include <mach/boolean.h>
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/lock.h>
-#include <sys/proc.h>
+#include <sys/proc_internal.h>
+#include <sys/kauth.h>
 #include <sys/buf.h>
 #include <sys/uio.h>
-#include <sys/vnode.h>
+#include <sys/vnode_internal.h>
 #include <sys/namei.h>
-#include <sys/ubc.h>
+#include <sys/ubc_internal.h>
+#include <sys/malloc.h>
+#include <sys/user.h>
+#if CONFIG_PROTECT
+#include <sys/cprotect.h>
+#endif
+
+#include <default_pager/default_pager_types.h>
+#include <default_pager/default_pager_object.h>
+
+#include <security/audit/audit.h>
+#include <bsm/audit_kevents.h>
 
 #include <mach/mach_types.h>
-#include <vm/vm_map.h>
-#include <vm/vm_kern.h>
+#include <mach/host_priv.h>
+#include <mach/mach_traps.h>
+#include <mach/boolean.h>
+
+#include <kern/kern_types.h>
 #include <kern/host.h>
+#include <kern/task.h>
 #include <kern/zalloc.h>
 #include <kern/kalloc.h>
+#include <kern/assert.h>
+
 #include <libkern/libkern.h>
-#include <sys/malloc.h>
 
+#include <vm/vm_pageout.h>
+#include <vm/vm_map.h>
+#include <vm/vm_kern.h>
 #include <vm/vnode_pager.h>
+#include <vm/vm_protos.h>
+#if CONFIG_MACF
+#include <security/mac_framework.h>
+#endif
 
 /*
  * temporary support for delayed instantiation
@@ -63,7 +98,119 @@ struct bs_map               bs_port_table[MAX_BACKING_STORE] = {
 /* ###################################################### */
 
 
-#include <kern/assert.h>
+/*
+ *     Routine:        macx_backing_store_recovery
+ *     Function:
+ *             Syscall interface to set a tasks privilege
+ *             level so that it is not subject to 
+ *             macx_backing_store_suspend
+ */
+int
+macx_backing_store_recovery(
+       struct macx_backing_store_recovery_args *args)
+{
+       int             pid = args->pid;
+       int             error;
+       struct proc     *p =  current_proc();
+       boolean_t       funnel_state;
+
+       funnel_state = thread_funnel_set(kernel_flock, TRUE);
+       if ((error = suser(kauth_cred_get(), 0)))
+               goto backing_store_recovery_return;
+
+       /* for now restrict backing_store_recovery */
+       /* usage to only present task */
+       if(pid != proc_selfpid()) {
+               error = EINVAL;
+               goto backing_store_recovery_return;
+       }
+
+       task_backing_store_privileged(p->task);
+
+backing_store_recovery_return:
+       (void) thread_funnel_set(kernel_flock, FALSE);
+       return(error);
+}
+
+/*
+ *     Routine:        macx_backing_store_suspend
+ *     Function:
+ *             Syscall interface to stop new demand for 
+ *             backing store when backing store is low
+ */
+
+int
+macx_backing_store_suspend(
+       struct macx_backing_store_suspend_args *args)
+{
+       boolean_t       suspend = args->suspend;
+       int             error;
+       boolean_t       funnel_state;
+
+       funnel_state = thread_funnel_set(kernel_flock, TRUE);
+       if ((error = suser(kauth_cred_get(), 0)))
+               goto backing_store_suspend_return;
+
+       vm_backing_store_disable(suspend);
+
+backing_store_suspend_return:
+       (void) thread_funnel_set(kernel_flock, FALSE);
+       return(error);
+}
+
+extern boolean_t backing_store_stop_compaction;
+
+/*
+ *     Routine:        macx_backing_store_compaction
+ *     Function:
+ *             Turn compaction of swap space on or off.  This is
+ *             used during shutdown/restart so that the kernel 
+ *             doesn't waste time compacting swap files that are 
+ *             about to be deleted anyway.  Compaction is always 
+ *             on by default when the system comes up and is turned 
+ *             off when a shutdown/restart is requested.  It is 
+ *             re-enabled if the shutdown/restart is aborted for any reason.
+ */
+
+int
+macx_backing_store_compaction(int flags)
+{
+       int error;
+
+       if ((error = suser(kauth_cred_get(), 0)))
+               return error;
+
+       if (flags & SWAP_COMPACT_DISABLE) {
+               backing_store_stop_compaction = TRUE;
+
+       } else if (flags & SWAP_COMPACT_ENABLE) {
+               backing_store_stop_compaction = FALSE;
+       }
+
+       return 0;
+}
+
+/*
+ *     Routine:        macx_triggers
+ *     Function:
+ *             Syscall interface to set the call backs for low and
+ *             high water marks.
+ */
+int
+macx_triggers(
+       struct macx_triggers_args *args)
+{
+       int     error;
+
+       error = suser(kauth_cred_get(), 0);
+       if (error)
+               return error;
+
+       return mach_macx_triggers(args);
+}
+
+
+extern boolean_t dp_isssd;
 
 /*
  *     Routine:        macx_swapon
@@ -72,65 +219,77 @@ struct bs_map              bs_port_table[MAX_BACKING_STORE] = {
  */
 int
 macx_swapon(
-       char    *filename,
-       int     flags,
-       long    size,
-       long    priority)
+       struct macx_swapon_args *args)
 {
-       struct vnode            *vp = 0; 
+       int                     size = args->size;
+       vnode_t                 vp = (vnode_t)NULL; 
        struct nameidata        nd, *ndp;
-       struct proc             *p =  current_proc();
-       pager_file_t            pf;
        register int            error;
        kern_return_t           kr;
        mach_port_t             backing_store;
        memory_object_default_t default_pager;
        int                     i;
        boolean_t               funnel_state;
+       off_t                   file_size;
+       vfs_context_t           ctx = vfs_context_current();
+       struct proc             *p =  current_proc();
+       int                     dp_cluster_size;
 
-       struct vattr    vattr;
+
+       AUDIT_MACH_SYSCALL_ENTER(AUE_SWAPON);
+       AUDIT_ARG(value32, args->priority);
 
        funnel_state = thread_funnel_set(kernel_flock, TRUE);
        ndp = &nd;
 
-       if ((error = suser(p->p_ucred, &p->p_acflag)))
+       if ((error = suser(kauth_cred_get(), 0)))
                goto swapon_bailout;
 
-       if(default_pager_init_flag == 0) {
-               start_def_pager(NULL);
-               default_pager_init_flag = 1;
-       }
-
        /*
         * Get a vnode for the paging area.
         */
-       NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
-           filename, p);
+       NDINIT(ndp, LOOKUP, OP_LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
+              ((IS_64BIT_PROCESS(p)) ? UIO_USERSPACE64 : UIO_USERSPACE32),
+              (user_addr_t) args->filename, ctx);
 
        if ((error = namei(ndp)))
                goto swapon_bailout;
+       nameidone(ndp);
        vp = ndp->ni_vp;
 
        if (vp->v_type != VREG) {
                error = EINVAL;
-               VOP_UNLOCK(vp, 0, p);
                goto swapon_bailout;
        }
-       UBCINFOCHECK("macx_swapon", vp);
 
-       if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) {
-               VOP_UNLOCK(vp, 0, p);
+       /* get file size */
+       if ((error = vnode_size(vp, &file_size, ctx)) != 0)
+               goto swapon_bailout;
+#if CONFIG_MACF
+       vnode_lock(vp);
+       error = mac_system_check_swapon(vfs_context_ucred(ctx), vp);
+       vnode_unlock(vp);
+       if (error)
+               goto swapon_bailout;
+#endif
+
+       /* resize to desired size if it's too small */
+       if ((file_size < (off_t)size) && ((error = vnode_setsize(vp, (off_t)size, 0, ctx)) != 0))
                goto swapon_bailout;
-       }
 
-       if (vattr.va_size < (u_quad_t)size) {
-               vattr_null(&vattr);
-               vattr.va_size = (u_quad_t)size;
-               error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
-               if (error) {
-                       VOP_UNLOCK(vp, 0, p);
+#if CONFIG_PROTECT
+       {
+               /* initialize content protection keys manually */
+               if ((error = cp_handle_vnop(vp, CP_WRITE_ACCESS, 0)) != 0) {
                        goto swapon_bailout;
-               }
+               }
+       }
+#endif
+
+
+       if (default_pager_init_flag == 0) {
+               start_def_pager(NULL);
+               default_pager_init_flag = 1;
        }
 
        /* add new backing store to list */
@@ -142,7 +301,6 @@ macx_swapon(
        }
        if(i == MAX_BACKING_STORE) {
                error = ENOMEM;
-               VOP_UNLOCK(vp, 0, p);
                goto swapon_bailout;
        }
 
@@ -157,24 +315,44 @@ macx_swapon(
        kr = host_default_memory_manager(host_priv_self(), &default_pager, 0);
        if(kr != KERN_SUCCESS) {
           error = EAGAIN;
-          VOP_UNLOCK(vp, 0, p);
           bs_port_table[i].vp = 0;
           goto swapon_bailout;
        }
 
+#if CONFIG_EMBEDDED
+       dp_cluster_size = 1 * PAGE_SIZE;
+#else
+       if ((dp_isssd = vnode_pager_isSSD(vp)) == TRUE) {
+               /*
+                * keep the cluster size small since the
+                * seek cost is effectively 0 which means
+                * we don't care much about fragmentation
+                */
+               dp_cluster_size = 2 * PAGE_SIZE;
+       } else {
+               /*
+                * use the default cluster size
+                */
+               dp_cluster_size = 0;
+       }
+#endif
        kr = default_pager_backing_store_create(default_pager, 
                                        -1, /* default priority */
-                                       0, /* default cluster size */
+                                       dp_cluster_size,
                                        &backing_store);
        memory_object_default_deallocate(default_pager);
 
        if(kr != KERN_SUCCESS) {
           error = ENOMEM;
-          VOP_UNLOCK(vp, 0, p);
           bs_port_table[i].vp = 0;
           goto swapon_bailout;
        }
 
+       /* Mark this vnode as being used for swapfile */
+       vnode_lock_spin(vp);
+       SET(vp->v_flag, VSWAP);
+       vnode_unlock(vp);
+
        /*
         * NOTE: we are able to supply PAGE_SIZE here instead of
         *      an actual record size or block number because:
@@ -183,39 +361,45 @@ macx_swapon(
         *      b: because allow paging will be done modulo page size
         */
 
-       VOP_UNLOCK(vp, 0, p);
-       kr = default_pager_add_file(backing_store, vp, PAGE_SIZE, 
-                       ((int)vattr.va_size)/PAGE_SIZE);
+       kr = default_pager_add_file(backing_store, (vnode_ptr_t) vp,
+                               PAGE_SIZE, (int)(file_size/PAGE_SIZE));
        if(kr != KERN_SUCCESS) {
           bs_port_table[i].vp = 0;
           if(kr == KERN_INVALID_ARGUMENT)
                error = EINVAL;
           else 
                error = ENOMEM;
+
+          /* This vnode is not to be used for swapfile */
+          vnode_lock_spin(vp);
+          CLR(vp->v_flag, VSWAP);
+          vnode_unlock(vp);
+
           goto swapon_bailout;
        }
        bs_port_table[i].bs = (void *)backing_store;
        error = 0;
-       if (!ubc_hold(vp))
-               panic("macx_swapon: hold");
 
-       /* Mark this vnode as being used for swapfile */
-       SET(vp->v_flag, VSWAP);
+       ubc_setthreadcred(vp, p, current_thread());
 
        /*
-        * take an extra reference on the vnode to keep
+        * take a long term reference on the vnode to keep
         * vnreclaim() away from this vnode.
         */
-       VREF(vp);
-
-       /* Hold on to the namei  reference to the paging file vnode */
-       vp = 0;
+       vnode_ref(vp);
 
 swapon_bailout:
        if (vp) {
-               vrele(vp);
+               vnode_put(vp);
        }
        (void) thread_funnel_set(kernel_flock, FALSE);
+       AUDIT_MACH_SYSCALL_EXIT(error);
+
+       if (error)
+               printf("macx_swapon FAILED - %d\n", error);
+       else
+               printf("macx_swapon SUCCESS\n");
+
        return(error);
 }
 
@@ -226,9 +410,9 @@ swapon_bailout:
  */
 int
 macx_swapoff(
-       char    *filename,
-       int     flags)
+       struct macx_swapoff_args *args)
 {
+       __unused int    flags = args->flags;
        kern_return_t   kr;
        mach_port_t     backing_store;
 
@@ -238,58 +422,75 @@ macx_swapoff(
        int                     i;
        int                     error;
        boolean_t               funnel_state;
+       vfs_context_t ctx = vfs_context_current();
+       struct uthread  *ut;
+       int                     orig_iopol_disk;
+
+       AUDIT_MACH_SYSCALL_ENTER(AUE_SWAPOFF);
 
        funnel_state = thread_funnel_set(kernel_flock, TRUE);
        backing_store = NULL;
        ndp = &nd;
 
-       if ((error = suser(p->p_ucred, &p->p_acflag)))
+       if ((error = suser(kauth_cred_get(), 0)))
                goto swapoff_bailout;
 
        /*
         * Get the vnode for the paging area.
         */
-       NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
-           filename, p);
+       NDINIT(ndp, LOOKUP, OP_LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
+              ((IS_64BIT_PROCESS(p)) ? UIO_USERSPACE64 : UIO_USERSPACE32),
+              (user_addr_t) args->filename, ctx);
 
        if ((error = namei(ndp)))
                goto swapoff_bailout;
+       nameidone(ndp);
        vp = ndp->ni_vp;
 
        if (vp->v_type != VREG) {
                error = EINVAL;
-               VOP_UNLOCK(vp, 0, p);
                goto swapoff_bailout;
        }
+#if CONFIG_MACF
+       vnode_lock(vp);
+       error = mac_system_check_swapoff(vfs_context_ucred(ctx), vp);
+       vnode_unlock(vp);
+       if (error)
+               goto swapoff_bailout;
+#endif
 
        for(i = 0; i < MAX_BACKING_STORE; i++) {
                if(bs_port_table[i].vp == vp) {
-                       backing_store; 
                        break;
                }
        }
        if (i == MAX_BACKING_STORE) {
                error = EINVAL;
-               VOP_UNLOCK(vp, 0, p);
                goto swapoff_bailout;
        }
        backing_store = (mach_port_t)bs_port_table[i].bs;
 
-       VOP_UNLOCK(vp, 0, p);
+       ut = get_bsdthread_info(current_thread());
+
+       orig_iopol_disk = proc_get_thread_selfdiskacc();
+       proc_apply_thread_selfdiskacc(IOPOL_THROTTLE);
+
        kr = default_pager_backing_store_delete(backing_store);
+
+       proc_apply_thread_selfdiskacc(orig_iopol_disk);
+
        switch (kr) {
                case KERN_SUCCESS:
                        error = 0;
                        bs_port_table[i].vp = 0;
-                       ubc_rele(vp);
                        /* This vnode is no longer used for swapfile */
+                       vnode_lock_spin(vp);
                        CLR(vp->v_flag, VSWAP);
+                       vnode_unlock(vp);
 
-                       /* get rid of macx_swapon() namei() reference */
-                       vrele(vp);
+                       /* get rid of macx_swapon() "long term" reference */
+                       vnode_rele(vp);
 
-                       /* get rid of macx_swapon() "extra" reference */
-                       vrele(vp);
                        break;
                case KERN_FAILURE:
                        error = EAGAIN;
@@ -302,8 +503,84 @@ macx_swapoff(
 swapoff_bailout:
        /* get rid of macx_swapoff() namei() reference */
        if (vp)
-               vrele(vp);
+               vnode_put(vp);
 
        (void) thread_funnel_set(kernel_flock, FALSE);
+       AUDIT_MACH_SYSCALL_EXIT(error);
+
+       if (error)
+               printf("macx_swapoff FAILED - %d\n", error);
+       else
+               printf("macx_swapoff SUCCESS\n");
+
        return(error);
 }
+
+/*
+ *     Routine:        macx_swapinfo
+ *     Function:
+ *             Syscall interface to get general swap statistics
+ */
+int
+macx_swapinfo(
+       memory_object_size_t    *total_p,
+       memory_object_size_t    *avail_p,
+       vm_size_t               *pagesize_p,
+       boolean_t               *encrypted_p)
+{
+       int                     error;
+       memory_object_default_t default_pager;
+       default_pager_info_64_t dpi64;
+       kern_return_t           kr;
+
+       error = 0;
+
+       /*
+        * Get a handle on the default pager.
+        */
+       default_pager = MEMORY_OBJECT_DEFAULT_NULL;
+       kr = host_default_memory_manager(host_priv_self(), &default_pager, 0);
+       if (kr != KERN_SUCCESS) {
+               error = EAGAIN; /* XXX why EAGAIN ? */
+               goto done;
+       }
+       if (default_pager == MEMORY_OBJECT_DEFAULT_NULL) {
+               /*
+                * The default pager has not initialized yet,
+                * so it can't be using any swap space at all.
+                */
+               *total_p = 0;
+               *avail_p = 0;
+               *pagesize_p = 0;
+               *encrypted_p = FALSE;
+               goto done;
+       }
+       
+       /*
+        * Get swap usage data from default pager.
+        */
+       kr = default_pager_info_64(default_pager, &dpi64);
+       if (kr != KERN_SUCCESS) {
+               error = ENOTSUP;
+               goto done;
+       }
+
+       /*
+        * Provide default pager info to caller.
+        */
+       *total_p = dpi64.dpi_total_space;
+       *avail_p = dpi64.dpi_free_space;
+       *pagesize_p = dpi64.dpi_page_size;
+       if (dpi64.dpi_flags & DPI_ENCRYPTED) {
+               *encrypted_p = TRUE;
+       } else {
+               *encrypted_p = FALSE;
+       }
+
+done:
+       if (default_pager != MEMORY_OBJECT_DEFAULT_NULL) {
+               /* release our handle on default pager */
+               memory_object_default_deallocate(default_pager);
+       }
+       return error;
+}