xnu-3789.1.32.tar.gz

[apple/xnu.git] / bsd / kern / kern_core.c

/*
 * Copyright (c) 2000-2006 Apple Computer, Inc. All rights reserved.
 *
 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
 * 
 * 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.
 * 
 * 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, QUIET ENJOYMENT OR NON-INFRINGEMENT.
 * Please see the License for the specific language governing rights and
 * limitations under the License.
 * 
 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
 */
/* Copyright (c) 1991 NeXT Computer, Inc.  All rights reserved.
 *
 *	File:	bsd/kern/kern_core.c
 *
 *	This file contains machine independent code for performing core dumps.
 *
 */
#if CONFIG_COREDUMP

#include <mach/vm_param.h>
#include <mach/thread_status.h>

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/signalvar.h>
#include <sys/resourcevar.h>
#include <sys/namei.h>
#include <sys/vnode_internal.h>
#include <sys/proc_internal.h>
#include <sys/kauth.h>
#include <sys/timeb.h>
#include <sys/times.h>
#include <sys/acct.h>
#include <sys/file_internal.h>
#include <sys/uio.h>
#include <sys/kernel.h>
#include <sys/stat.h>

#include <mach-o/loader.h>
#include <mach/vm_region.h>
#include <mach/vm_statistics.h>

#include <vm/vm_kern.h>
#include <vm/vm_protos.h> /* last */
#include <vm/vm_map.h>		/* current_map() */
#include <mach/mach_vm.h>	/* mach_vm_region_recurse() */
#include <mach/task.h>		/* task_suspend() */
#include <kern/task.h>		/* get_task_numacts() */

#include <security/audit/audit.h>

#if CONFIG_CSR
#include <sys/codesign.h>
#include <sys/csr.h>
#endif

typedef struct {
	int	flavor;			/* the number for this flavor */
	mach_msg_type_number_t	count;	/* count of ints in this flavor */
} mythread_state_flavor_t;

#if defined (__i386__) || defined (__x86_64__)
mythread_state_flavor_t thread_flavor_array [] = { 
		{x86_THREAD_STATE, x86_THREAD_STATE_COUNT},
		{x86_FLOAT_STATE, x86_FLOAT_STATE_COUNT},
		{x86_EXCEPTION_STATE, x86_EXCEPTION_STATE_COUNT},
		};
int mynum_flavors=3;
#else
#error architecture not supported
#endif


typedef struct {
	vm_offset_t header; 
	int  hoffset;
	mythread_state_flavor_t *flavors;
	int tstate_size;
	int flavor_count;
} tir_t;

extern int freespace_mb(vnode_t vp);

/* XXX not in a Mach header anywhere */
kern_return_t thread_getstatus(thread_t act, int flavor,
	thread_state_t tstate, mach_msg_type_number_t *count);
void task_act_iterate_wth_args(task_t, void(*)(thread_t, void *), void *);

#ifdef SECURE_KERNEL
__XNU_PRIVATE_EXTERN int do_coredump = 0;	/* default: don't dump cores */
#else
__XNU_PRIVATE_EXTERN int do_coredump = 1;	/* default: dump cores */
#endif
__XNU_PRIVATE_EXTERN int sugid_coredump = 0; /* default: but not SGUID binaries */


/* cpu_type returns only the most generic indication of the current CPU. */
/* in a core we want to know the kind of process. */

static cpu_type_t
process_cpu_type(proc_t core_proc)
{
	cpu_type_t what_we_think;
#if defined (__i386__) || defined (__x86_64__)
    if (IS_64BIT_PROCESS(core_proc)) {
		what_we_think = CPU_TYPE_X86_64;
	} else {
		what_we_think = CPU_TYPE_I386;
	}
#endif
	return what_we_think;
}

static cpu_type_t
process_cpu_subtype(proc_t core_proc)
{
	cpu_type_t what_we_think;
#if defined (__i386__) || defined (__x86_64__)
    if (IS_64BIT_PROCESS(core_proc)) {
		what_we_think = CPU_SUBTYPE_X86_64_ALL;
	} else {
		what_we_think = CPU_SUBTYPE_I386_ALL;
	}
#endif
	return what_we_think;
}

static void
collectth_state(thread_t th_act, void *tirp)
{
	vm_offset_t	header;
	int  hoffset, i ;
	mythread_state_flavor_t *flavors;
	struct thread_command	*tc;
	tir_t *t = (tir_t *)tirp;

		/*
		 *	Fill in thread command structure.
		 */
		header = t->header;
		hoffset = t->hoffset;
		flavors = t->flavors;
	
		tc = (struct thread_command *) (header + hoffset);
		tc->cmd = LC_THREAD;
		tc->cmdsize = sizeof(struct thread_command)
				+ t->tstate_size;
		hoffset += sizeof(struct thread_command);
		/*
		 * Follow with a struct thread_state_flavor and
		 * the appropriate thread state struct for each
		 * thread state flavor.
		 */
		for (i = 0; i < t->flavor_count; i++) {
			*(mythread_state_flavor_t *)(header+hoffset) =
			  flavors[i];
			hoffset += sizeof(mythread_state_flavor_t);
			thread_getstatus(th_act, flavors[i].flavor,
					(thread_state_t)(header+hoffset),
					&flavors[i].count);
			hoffset += flavors[i].count*sizeof(int);
		}

		t->hoffset = hoffset;
}

/*
 * coredump
 *
 * Description:	Create a core image on the file "core" for the process
 *		indicated
 *
 * Parameters:	core_proc			Process to dump core [*]
 *				reserve_mb			If non-zero, leave filesystem with
 *									at least this much free space.
 *				coredump_flags	Extra options (ignore rlimit, run fsync)
 *
 * Returns:	0				Success
 *		EFAULT				Failed
 *
 * IMPORTANT:	This function can only be called on the current process, due
 *		to assumptions below; see variable declaration section for
 *		details.
 */
#define	MAX_TSTATE_FLAVORS	10
int
coredump(proc_t core_proc, uint32_t reserve_mb, int coredump_flags)
{
/* Begin assumptions that limit us to only the current process */
	vfs_context_t ctx = vfs_context_current();
	vm_map_t	map = current_map();
	task_t		task = current_task();
/* End assumptions */
	kauth_cred_t cred = vfs_context_ucred(ctx);
	int error = 0;
	struct vnode_attr va;
	int		thread_count, segment_count;
	int		command_size, header_size, tstate_size;
	int		hoffset;
	off_t		foffset;
	mach_vm_offset_t vmoffset;
	vm_offset_t	header;
	mach_vm_size_t	vmsize;
	vm_prot_t	prot;
	vm_prot_t	maxprot;
	vm_inherit_t	inherit;
	int		error1 = 0;
	char		stack_name[MAXCOMLEN+6];
	char		*alloced_name = NULL;
	char		*name;
	mythread_state_flavor_t flavors[MAX_TSTATE_FLAVORS];
	vm_size_t	mapsize;
	int		i;
	uint32_t nesting_depth = 0;
	kern_return_t	kret;
	struct vm_region_submap_info_64 vbr;
	mach_msg_type_number_t vbrcount = 0;
	tir_t tir1;
	struct vnode * vp;
	struct mach_header	*mh = NULL;	/* protected by is_64 */
	struct mach_header_64	*mh64 = NULL;	/* protected by is_64 */
	int		is_64 = 0;
	size_t		mach_header_sz = sizeof(struct mach_header);
	size_t		segment_command_sz = sizeof(struct segment_command);
	
	if (current_proc() != core_proc) {
		panic("coredump() called against proc that is not current_proc: %p", core_proc);
	}

	if (do_coredump == 0 ||		/* Not dumping at all */
	    ( (sugid_coredump == 0) &&	/* Not dumping SUID/SGID binaries */
	      ( (kauth_cred_getsvuid(cred) != kauth_cred_getruid(cred)) ||
	        (kauth_cred_getsvgid(cred) != kauth_cred_getrgid(cred))))) {

#if CONFIG_AUDIT
		audit_proc_coredump(core_proc, NULL, EFAULT);
#endif
		return (EFAULT);
	}

#if CONFIG_CSR
	/* If the process is restricted, CSR isn't configured to allow
	 * restricted processes to be debugged, and CSR isn't configured in
	 * AppleInternal mode, then don't dump core. */
	if (cs_restricted(core_proc) &&
	    csr_check(CSR_ALLOW_TASK_FOR_PID) &&
	    csr_check(CSR_ALLOW_APPLE_INTERNAL)) {
#if CONFIG_AUDIT
		audit_proc_coredump(core_proc, NULL, EFAULT);
#endif
		return (EFAULT);
	}
#endif

	if (IS_64BIT_PROCESS(core_proc)) {
		is_64 = 1;
		mach_header_sz = sizeof(struct mach_header_64);
		segment_command_sz = sizeof(struct segment_command_64);
	}

	mapsize = get_vmmap_size(map);

	if (((coredump_flags & COREDUMP_IGNORE_ULIMIT) == 0) &&
	    (mapsize >=  core_proc->p_rlimit[RLIMIT_CORE].rlim_cur))
		return (EFAULT);

	(void) task_suspend_internal(task);

	MALLOC(alloced_name, char *, MAXPATHLEN, M_TEMP, M_NOWAIT | M_ZERO);

	/* create name according to sysctl'able format string */
	/* if name creation fails, fall back to historical behaviour... */
	if (alloced_name == NULL ||
	    proc_core_name(core_proc->p_comm, kauth_cred_getuid(cred),
			   core_proc->p_pid, alloced_name, MAXPATHLEN)) {
		snprintf(stack_name, sizeof(stack_name),
			 "/cores/core.%d", core_proc->p_pid);
		name = stack_name;
	} else
		name = alloced_name;

	if ((error = vnode_open(name, (O_CREAT | FWRITE | O_NOFOLLOW), S_IRUSR, VNODE_LOOKUP_NOFOLLOW, &vp, ctx)))
		goto out2;

	VATTR_INIT(&va);
	VATTR_WANTED(&va, va_nlink);
	/* Don't dump to non-regular files or files with links. */
	if (vp->v_type != VREG ||
	    vnode_getattr(vp, &va, ctx) || va.va_nlink != 1) {
		error = EFAULT;
		goto out;
	}

	VATTR_INIT(&va);	/* better to do it here than waste more stack in vnode_setsize */
	VATTR_SET(&va, va_data_size, 0);
	vnode_setattr(vp, &va, ctx);
	core_proc->p_acflag |= ACORE;

	if ((reserve_mb > 0) &&
	    ((freespace_mb(vp) - (mapsize >> 20)) < reserve_mb)) {
		error = ENOSPC;
		goto out;
	}

	/*
	 *	If the task is modified while dumping the file
	 *	(e.g., changes in threads or VM, the resulting
	 *	file will not necessarily be correct.
	 */

	thread_count = get_task_numacts(task);
	segment_count = get_vmmap_entries(map);	/* XXX */
	tir1.flavor_count = sizeof(thread_flavor_array)/sizeof(mythread_state_flavor_t);
	bcopy(thread_flavor_array, flavors,sizeof(thread_flavor_array));
	tstate_size = 0;
	for (i = 0; i < tir1.flavor_count; i++)
		tstate_size += sizeof(mythread_state_flavor_t) +
		  (flavors[i].count * sizeof(int));
	command_size = segment_count * segment_command_sz +
	  thread_count*sizeof(struct thread_command) +
	  tstate_size*thread_count;

	header_size = command_size + mach_header_sz;

	if (kmem_alloc(kernel_map, &header, (vm_size_t)header_size, VM_KERN_MEMORY_DIAG) != KERN_SUCCESS) {
		error = ENOMEM;
		goto out;
	}

	/*
	 *	Set up Mach-O header.
	 */
	if (is_64) {
		mh64 = (struct mach_header_64 *)header;
		mh64->magic = MH_MAGIC_64;
		mh64->cputype = process_cpu_type(core_proc);
		mh64->cpusubtype = process_cpu_subtype(core_proc);
		mh64->filetype = MH_CORE;
		mh64->ncmds = segment_count + thread_count;
		mh64->sizeofcmds = command_size;
		mh64->reserved = 0;		/* 8 byte alignment */
	} else {
		mh = (struct mach_header *)header;
		mh->magic = MH_MAGIC;
		mh->cputype = process_cpu_type(core_proc);
		mh->cpusubtype = process_cpu_subtype(core_proc);
		mh->filetype = MH_CORE;
		mh->ncmds = segment_count + thread_count;
		mh->sizeofcmds = command_size;
	}

	hoffset = mach_header_sz;	/* offset into header */
	foffset = round_page(header_size);	/* offset into file */
	vmoffset = MACH_VM_MIN_ADDRESS;		/* offset into VM */

	/*
	 * We use to check for an error, here, now we try and get 
	 * as much as we can
	 */
	while (segment_count > 0) {
		struct segment_command		*sc;
		struct segment_command_64	*sc64;

		/*
		 *	Get region information for next region.
		 */
		
		while (1) {
			vbrcount = VM_REGION_SUBMAP_INFO_COUNT_64;
			if((kret = mach_vm_region_recurse(map, 
					&vmoffset, &vmsize, &nesting_depth, 
					(vm_region_recurse_info_t)&vbr,
					&vbrcount)) != KERN_SUCCESS) {
				break;
			}
			/*
			 * If we get a valid mapping back, but we're dumping
			 * a 32 bit process,  and it's over the allowable
			 * address space of a 32 bit process, it's the same
			 * as if mach_vm_region_recurse() failed.
			 */
			if (!(is_64) &&
			    (vmoffset + vmsize > VM_MAX_ADDRESS)) {
			    	kret = KERN_INVALID_ADDRESS;
				break;
			}
			if(vbr.is_submap) {
				nesting_depth++;
				continue;
			} else {
				break;
			}
		}
		if(kret != KERN_SUCCESS)
			break;

		prot = vbr.protection;
		maxprot = vbr.max_protection;
		inherit = vbr.inheritance;
		/*
		 *	Fill in segment command structure.
		 */
		if (is_64) {
			sc64 = (struct segment_command_64 *)(header + hoffset);
			sc64->cmd = LC_SEGMENT_64;
			sc64->cmdsize = sizeof(struct segment_command_64);
			/* segment name is zeroed by kmem_alloc */
			sc64->segname[0] = 0;
			sc64->vmaddr = vmoffset;
			sc64->vmsize = vmsize;
			sc64->fileoff = foffset;
			sc64->filesize = vmsize;
			sc64->maxprot = maxprot;
			sc64->initprot = prot;
			sc64->nsects = 0;
			sc64->flags = 0;
		} else  {
			sc = (struct segment_command *) (header + hoffset);
			sc->cmd = LC_SEGMENT;
			sc->cmdsize = sizeof(struct segment_command);
			/* segment name is zeroed by kmem_alloc */
			sc->segname[0] = 0;
			sc->vmaddr = CAST_DOWN_EXPLICIT(vm_offset_t,vmoffset);
			sc->vmsize = CAST_DOWN_EXPLICIT(vm_size_t,vmsize);
			sc->fileoff = CAST_DOWN_EXPLICIT(uint32_t,foffset); /* will never truncate */
			sc->filesize = CAST_DOWN_EXPLICIT(uint32_t,vmsize); /* will never truncate */
			sc->maxprot = maxprot;
			sc->initprot = prot;
			sc->nsects = 0;
			sc->flags = 0;
		}

		/*
		 *	Write segment out.  Try as hard as possible to
		 *	get read access to the data.
		 */
		if ((prot & VM_PROT_READ) == 0) {
			mach_vm_protect(map, vmoffset, vmsize, FALSE,
					   prot|VM_PROT_READ);
		}
		/*
		 *	Only actually perform write if we can read.
		 *	Note: if we can't read, then we end up with
		 *	a hole in the file.
		 */
		if ((maxprot & VM_PROT_READ) == VM_PROT_READ
			&& vbr.user_tag != VM_MEMORY_IOKIT
			&& coredumpok(map,vmoffset)) {
			
			error = vn_rdwr_64(UIO_WRITE, vp, vmoffset, vmsize, foffset,
					(IS_64BIT_PROCESS(core_proc) ? UIO_USERSPACE64 : UIO_USERSPACE32), 
					IO_NOCACHE|IO_NODELOCKED|IO_UNIT, cred, (int64_t *) 0, core_proc);

		}

		hoffset += segment_command_sz;
		foffset += vmsize;
		vmoffset += vmsize;
		segment_count--;
	}

	/*
	 * If there are remaining segments which have not been written
	 * out because break in the loop above, then they were not counted
	 * because they exceed the real address space of the executable
	 * type: remove them from the header's count.  This is OK, since
	 * we are allowed to have a sparse area following the segments.
	 */
	if (is_64) {
		mh64->ncmds -= segment_count;
		mh64->sizeofcmds -= segment_count * segment_command_sz;
	} else {
		mh->ncmds -= segment_count;
		mh->sizeofcmds -= segment_count * segment_command_sz;
	}

	tir1.header = header;
	tir1.hoffset = hoffset;
	tir1.flavors = flavors;
	tir1.tstate_size = tstate_size;
	task_act_iterate_wth_args(task, collectth_state,&tir1);

	/*
	 *	Write out the Mach header at the beginning of the
	 *	file.  OK to use a 32 bit write for this.
	 */
	error = vn_rdwr(UIO_WRITE, vp, (caddr_t)header, header_size, (off_t)0,
			UIO_SYSSPACE, IO_NOCACHE|IO_NODELOCKED|IO_UNIT, cred, (int *) 0, core_proc);
	kmem_free(kernel_map, header, header_size);

	if ((coredump_flags & COREDUMP_FULLFSYNC) && error == 0)
		error = VNOP_IOCTL(vp, F_FULLFSYNC, (caddr_t)NULL, 0, ctx);
out:
	error1 = vnode_close(vp, FWRITE, ctx);
out2:
#if CONFIG_AUDIT
	audit_proc_coredump(core_proc, name, error);
#endif
	if (alloced_name != NULL)
		FREE(alloced_name, M_TEMP);
	if (error == 0)
		error = error1;

	return (error);
}

#else /* CONFIG_COREDUMP */

/* When core dumps aren't needed, no need to compile this file at all */

#error assertion failed: this section is not compiled

#endif /* CONFIG_COREDUMP */