Wireshark-dev: [Wireshark-dev] Simplifying (and fixing) tvbuff [Long]

From: Bill Meier <wmeier@xxxxxxxxxxx>
Date: Mon, 12 Dec 2011 16:55:50 -0500
Summary
-------

I've recently been digging into the tvbuff code.

After doing so, I've come to the conclusion that the current tvbuff implementation of managing tvb's with usage_counts and used_in lists:
- doesn't really provide some of what I understand to be
  the intended functionality;
- is not necessary given the ways that Wireshark actually uses tvbuffs; - can be significantly simplified by doing away with usage counts
  and used_in lists.


(Comment: I started digging into the tvbuff code while reviewing the patch provided with Bug #6573:
  "tvbuff_t values leak if they have a child tvbuff_t (memory leak)".
The patch does, in fact, improve matters significantly but the patched tvbuff still has issues).

Below I give a few examples of issues related to the current tvbuff implementation and then describe how tvbuff can be simplified.

I've attached new versions of tvbuff-int.h and tvbuff.c with revised code for managing REAL_DATA and SUBSET tvb's. The changes are fairly simple. Wireshark seems to work AOK with the new versions (although I certainly haven't yet done extensive testing).

I think the revised code:
- provides the functionality required by Wireshark;
- is simpler and easier to understand.

Comments are welcome !

Bill


Some current tvbuff issues
--------------------------

Based upon comments in tvbuff.h[1] and in epan.c[2] it seems a design goal for the tvbuff code was to allow a dissector to do things like the following:

a. Do tvb_subset_new() and then, when done with the subset, tvb_free() ;

b. Increment the usage count of a tvb which the dissector wishes
   to "save for later use".

Neither of the above sequences currently work properly (and thus are not currently actually used in Wireshark). (See [3] below for some details).

c. The following sequence will fail:
   x1 = tvb_new_real_data();
   x2 = tvb_new_child_real_data();
   tvb_free(x2);
   tvb_free_chain(x1);

   issue: For REAL_DATA there's no "back-pointer" in x2
          pointing to x1 and thus the tvb_free() can't
          clean the "used_in" list for x1.

d. composite tvbuff's don't currently work (with respect to
   usage-counts, used-in lists and tvb_free()).
   The patch with bug #6573 certainly improves matters, but my brain
   fries when I try to work thru all the scenarios.

I expect there are more issues which I didn't see or have forgotten as I worked through the code.


Proposal
--------

Essentially:

1. API:
   Any new tvbs created by a dissector which are based upon or
   chained to a tvb handed to the dissector should not (must not)
   be free'd or "saved for later use" by the dissector.
   (AFAIKT Wireshark dissectors currently work this way so
   no dissector changes are required). (See Note below).

   A dissector can create a new tvb using tvb_new_real_data()
   with data storage allocated by the dissector.
   - Since the tvb is owned and managed by the dissector (and not
     chained to any tvb handed to the dissector), the dissector
     can do whatever is desired with that tvb (e.g., save the tvb for
     later use when dissecting another frame).
   - The dissector must eventually free the tvb
     (and possibly the storage) via a call to tvb_free_chain() for
     that tvb.

   Note: - tvb_free_chain() is the only way to free tvb's.
           [Existing dissector calls to tvb_free() can simply
           be replaced by calls to tvb_free_chain()].

         - tvb_free_chain must be called only for the tvb which
           is the first tvb in the chain.
           [XXX: It may be that it will turn out to be OK to allow
           a dissector to do a tvb_free_chain() for any tvb
           created by that dissector (even if based upon or
           chained to a tvb handed to the dissector).

2. Implementation:
   a. Do away with usage_counts and the used_in lists altogether;
      All the existing tvb_new..() functions work pretty much as is
      except they don't try to maintain usage_counts &
      used_in lists & etc.

   b. tvb's (real, subset & composite) are "chained" using a simple
      doubly-linked list (maintained via new tvbuff_t *pointers in
      the tvb struct ('next' and 'previous').

      (A composite tvb should always be part of a chain so that it
      eventually gets free'd).
      [XXX:
        - This means that all members of a composite must
            be part of the same chain as the composite.
      - One way to add a composite tvb to the chain:
        tvb_child_finalize_composite() or some thing similar.
        TBD].

    c. tvb_free_chain() just goes down the list freeing each tvb in
       turn. The only special actions are to exec the call-back function
       for REAL_DATA tvb's and to clear the COMPOSITE GSLISTs. No
       special action is required for SUBSET tvb's.

      [XXX: If it's possible to guarantee that any
            subset/composite tvb in the list only references tvbs
            earlier in the list then it should be OK to allow
            free_chain() to start at any place in the list].

Notes:

[1]  From tvbuff.h:
  * The caller can artificially increment/decrement the usage count
 * with tvbuff_increment_usage_count()/tvbuff_decrement_usage_count().

[2]  From packet.h
 	/* Free all tvb's created from this tvb, unless dissector
	 * wanted to store the pointer (in which case, the dissector
	 * would have incremented the usage count on that tvbuff_t*) */
	tvb_free_chain(edt->tvb);

[3]
a. tvb_free() of a subset doesn't remove the tvb from the parent "used_in" list and thus a subsequent tvb_free_chain() fails.
(The patch for bug #6573 added some code to fix this).

b. In general, the sequence tvb_increment_usage_count(tvb)/tvb_free(tvb) as implemented does work correctly.

However: Note that most likely a tvb handed to dissector is ultimately based upon a "top-level" real tvb which is based, AFAIKT, on a data buffer filled in by wiretap (which will be overwritten when the next frame is read).

So: trying to "keep around" a tvb in such a case won't actually work
    as intended. The tvb needs to be copied if it is to be saved.
/* tvbuff.c
 *
 * Testy, Virtual(-izable) Buffer of guint8*'s
 *
 * "Testy" -- the buffer gets mad when an attempt to access data
 *		beyond the bounds of the buffer. An exception is thrown.
 *
 * "Virtual" -- the buffer can have its own data, can use a subset of
 *		the data of a backing tvbuff, or can be a composite of
 *		other tvbuffs.
 *
 * $Id: tvbuff.c 39528 2011-10-23 22:02:56Z wmeier $
 *
 * Copyright (c) 2000 by Gilbert Ramirez <gram@xxxxxxxxxxxxxxx>
 *
 * Code to convert IEEE floating point formats to native floating point
 * derived from code Copyright (c) Ashok Narayanan, 2000
 *
 * Wireshark - Network traffic analyzer
 * By Gerald Combs <gerald@xxxxxxxxxxxxx>
 * Copyright 1998 Gerald Combs
 *
 * 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.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 */

#ifdef HAVE_CONFIG_H
# include "config.h"
#endif

#include <string.h>

#ifdef HAVE_LIBZ
#include <zlib.h>
#endif

#include "pint.h"
#include "tvbuff.h"
#include "tvbuff-int.h"
#include "strutil.h"
#include "emem.h"
#include "charsets.h"
#include "proto.h"	/* XXX - only used for DISSECTOR_ASSERT, probably a new header file? */


// ToDo: Determine how best to make stats visible when testing
int tvbuff_real_allocated_count	         = 0;
int tvb_new_child_real_data_count        = 0;
int tvb_new_real_data_count              = 0;
int tvb_set_child_real_data_tvbuff_count = 0;
int tvbuff_subset_allocated_count        = 0;
int tvbuff_composite_allocated_count     = 0;
int tvbuff_real_freed_count              = 0;
int tvbuff_subset_freed_count            = 0;
int tvbuff_composite_freed_count         = 0;
//

static const guint8*
ensure_contiguous_no_exception(tvbuff_t *tvb, const gint offset, const gint length,
		int *exception);

static const guint8*
ensure_contiguous(tvbuff_t *tvb, const gint offset, const gint length);

static void
tvb_init(tvbuff_t *tvb, const tvbuff_type type)
{
	tvb_backing_t	*backing;
	tvb_comp_t	*composite;

        tvb->previous		= NULL;
        tvb->next		= NULL;
	tvb->type		= type;
	tvb->initialized	= FALSE;
	tvb->length		= 0;
	tvb->reported_length	= 0;
	tvb->free_cb		= NULL;
	tvb->real_data		= NULL;
	tvb->raw_offset		= -1;
	tvb->ds_tvb		= NULL;

	switch(type) {
		case TVBUFF_REAL_DATA:
			/* Nothing */
			tvbuff_real_allocated_count += 1;
			break;

		case TVBUFF_SUBSET:
			tvbuff_subset_allocated_count += 1;
			backing = &tvb->tvbuffs.subset;
			backing->tvb	= NULL;
			backing->offset	= 0;
			backing->length	= 0;
			break;

		case TVBUFF_COMPOSITE:
			tvbuff_composite_allocated_count += 1;
			composite = &tvb->tvbuffs.composite;
			composite->tvbs			= NULL;
			composite->start_offsets	= NULL;
			composite->end_offsets		= NULL;
			break;

		default:
			DISSECTOR_ASSERT_NOT_REACHED();
			break;
	}
}


tvbuff_t*
tvb_new(const tvbuff_type type)
{
	tvbuff_t	*tvb;

	tvb = g_slice_new(tvbuff_t);

	tvb_init(tvb, type);

	return tvb;
}

static const unsigned char left_aligned_bitmask[] = {
	0xff,
	0x80,
	0xc0,
	0xe0,
	0xf0,
	0xf8,
	0xfc,
	0xfe
};

tvbuff_t *
tvb_new_octet_aligned(tvbuff_t *tvb, guint32 bit_offset, gint32 no_of_bits)
{
	tvbuff_t *sub_tvb = NULL;
	guint32 byte_offset;
	gint32 datalen, i;
	guint8 left, right, remaining_bits, *buf;
	const guint8 *data;

	byte_offset = bit_offset >> 3;
	left = bit_offset % 8; /* for left-shifting */
	right = 8 - left; /* for right-shifting */

	if (no_of_bits == -1) {
		datalen = tvb_length_remaining(tvb, byte_offset);
		remaining_bits = 0;
	} else {
		datalen = no_of_bits >> 3;
		remaining_bits = no_of_bits % 8;
		if (remaining_bits){
			datalen++;
		}
	}

	/* already aligned -> shortcut */
	if ((left == 0) && (remaining_bits == 0)) {
		return tvb_new_subset(tvb, byte_offset, datalen, -1);
	}

	buf = ep_alloc0(datalen);

	/* if at least one trailing byte is available, we must use the content
 	* of that byte for the last shift (i.e. tvb_get_ptr() must use datalen + 1
 	* if non extra byte is available, the last shifted byte requires
 	* special treatment
 	*/
	if (tvb_length_remaining(tvb, byte_offset) > datalen) {
		data = tvb_get_ptr(tvb, byte_offset, datalen + 1);
		/* shift tvb data bit_offset bits to the left */
		for (i = 0; i < datalen; i++)
			buf[i] = (data[i] << left) | (data[i+1] >> right);
	} else {
		data = tvb_get_ptr(tvb, byte_offset, datalen);
		/* shift tvb data bit_offset bits to the left */
		for (i = 0; i < (datalen-1); i++)
			buf[i] = (data[i] << left) | (data[i+1] >> right);
		buf[datalen-1] = data[datalen-1] << left; /* set last octet */
	}
	buf[datalen-1] &= left_aligned_bitmask[remaining_bits];

	sub_tvb = tvb_new_child_real_data(tvb, buf, datalen, datalen);

	return sub_tvb;
}

static tvbuff_t*
tvb_new_with_subset(const guint subset_tvb_offset, const guint subset_tvb_length)
{
	tvbuff_t *tvb = tvb_new(TVBUFF_SUBSET);
	tvb->tvbuffs.subset.offset = subset_tvb_offset;
	tvb->tvbuffs.subset.length = subset_tvb_length;

	return tvb;
}

static void
tvb_free(tvbuff_t* tvb)
{
	tvb_comp_t	*composite;

	DISSECTOR_ASSERT(tvb);

	switch (tvb->type) {
	case TVBUFF_REAL_DATA:
		if (tvb->free_cb) {
			/*
			 * XXX - do this with a union?
			 */
			tvb->free_cb((gpointer)tvb->real_data);
		}
		tvbuff_real_freed_count += 1;
		break;

	case TVBUFF_SUBSET:
		/* Nothing */
		tvbuff_subset_freed_count += 1;
		break;

	case TVBUFF_COMPOSITE:
		composite = &tvb->tvbuffs.composite;

		g_slist_free(composite->tvbs);

		g_free(composite->start_offsets);
		g_free(composite->end_offsets);
		if (tvb->real_data) {
			/*
			 * XXX - do this with a union?
			 */
			g_free((gpointer)tvb->real_data);
		}

		tvbuff_composite_freed_count += 1;
		break;

	default:
		DISSECTOR_ASSERT_NOT_REACHED();
	}

	g_slice_free(tvbuff_t, tvb);
}

void
tvb_free_chain(tvbuff_t* tvb)
{
	tvbuff_t *next;
	DISSECTOR_ASSERT(tvb);
	DISSECTOR_ASSERT(tvb->previous==NULL);
	while (tvb) {
		next=tvb->next;
		tvb_free(tvb);
		tvb  = next;
	}
}

void
tvb_set_free_cb(tvbuff_t* tvb, const tvbuff_free_cb_t func)
{
	DISSECTOR_ASSERT(tvb);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_REAL_DATA);
	tvb->free_cb = func;
}

static void
add_to_chain(tvbuff_t *parent, tvbuff_t *child)
{
	DISSECTOR_ASSERT(parent && child);
	child->next	= parent->next;
	child->previous = parent;
	if (parent->next)
		parent->next->previous = child;
	parent->next    = child;
}

void
tvb_set_child_real_data_tvbuff(tvbuff_t* parent, tvbuff_t* child)
{
	DISSECTOR_ASSERT(parent && child);
	DISSECTOR_ASSERT(parent->initialized);
	DISSECTOR_ASSERT(child->initialized);
	DISSECTOR_ASSERT(child->type == TVBUFF_REAL_DATA);
	add_to_chain(parent, child);
	tvb_set_child_real_data_tvbuff_count += 1;
}

static void
tvb_set_real_data_no_exceptions(tvbuff_t* tvb, const guint8* data, const guint length, const gint reported_length)
{
	tvb->real_data = data;
	tvb->length = length;
	tvb->reported_length = reported_length;
	tvb->initialized = TRUE;
}

void
tvb_set_real_data(tvbuff_t* tvb, const guint8* data, const guint length, const gint reported_length)
{
	DISSECTOR_ASSERT(tvb);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_REAL_DATA);
	DISSECTOR_ASSERT(!tvb->initialized);

	THROW_ON(reported_length < -1, ReportedBoundsError);

	tvb_set_real_data_no_exceptions(tvb, data, length, reported_length);
}

tvbuff_t*
tvb_new_real_data(const guint8* data, const guint length, const gint reported_length)
{
	tvbuff_t	*tvb;

	THROW_ON(reported_length < -1, ReportedBoundsError);

	tvb = tvb_new(TVBUFF_REAL_DATA);

	tvb_set_real_data_no_exceptions(tvb, data, length, reported_length);

	/*
	 * This is the top-level real tvbuff for this data source,
	 * so its data source tvbuff is itself.
	 */
	tvb->ds_tvb = tvb;
	tvb_new_real_data_count += 1;
	return tvb;
}

tvbuff_t*
tvb_new_child_real_data(tvbuff_t *parent, const guint8* data, const guint length, const gint reported_length)
{
	tvbuff_t *tvb = tvb_new_real_data(data, length, reported_length);
	if (tvb) {
		tvb_set_child_real_data_tvbuff (parent, tvb);
	}

	tvb_new_child_real_data_count += 1;
	return tvb;
}

/* Computes the absolute offset and length based on a possibly-negative offset
 * and a length that is possible -1 (which means "to the end of the data").
 * Returns TRUE/FALSE indicating whether the offset is in bounds or
 * not. The integer ptrs are modified with the new offset and length.
 * No exception is thrown.
 *
 * XXX - we return TRUE, not FALSE, if the offset is positive and right
 * after the end of the tvbuff (i.e., equal to the length).  We do this
 * so that a dissector constructing a subset tvbuff for the next protocol
 * will get a zero-length tvbuff, not an exception, if there's no data
 * left for the next protocol - we want the next protocol to be the one
 * that gets an exception, so the error is reported as an error in that
 * protocol rather than the containing protocol.  */
static gboolean
compute_offset_length(const guint tvb_length_val, const guint tvb_reported_length_val, const gint offset, const gint length_val,
		guint *offset_ptr, guint *length_ptr, int *exception)
{
	DISSECTOR_ASSERT(offset_ptr);
	DISSECTOR_ASSERT(length_ptr);

	/* Compute the offset */
	if (offset >= 0) {
		/* Positive offset - relative to the beginning of the packet. */
		if ((guint) offset > tvb_reported_length_val) {
			if (exception) {
				*exception = ReportedBoundsError;
			}
			return FALSE;
		}
		else if ((guint) offset > tvb_length_val) {
			if (exception) {
				*exception = BoundsError;
			}
			return FALSE;
		}
		else {
			*offset_ptr = offset;
		}
	}
	else {
		/* Negative offset - relative to the end of the packet. */
		if ((guint) -offset > tvb_reported_length_val) {
			if (exception) {
				*exception = ReportedBoundsError;
			}
			return FALSE;
		}
		else if ((guint) -offset > tvb_length_val) {
			if (exception) {
				*exception = BoundsError;
			}
			return FALSE;
		}
		else {
			*offset_ptr = tvb_length_val + offset;
		}
	}

	/* Compute the length */
	if (length_val < -1) {
		if (exception) {
			/* XXX - ReportedBoundsError? */
			*exception = BoundsError;
		}
		return FALSE;
	}
	else if (length_val == -1) {
		*length_ptr = tvb_length_val - *offset_ptr;
	}
	else {
		*length_ptr = length_val;
	}

	return TRUE;
}


static gboolean
check_offset_length_no_exception(const guint tvb_length_val, const guint tvb_reported_length_val, const gint offset, gint const length_val,
		guint *offset_ptr, guint *length_ptr, int *exception)
{
	guint	end_offset;

	if (!compute_offset_length(tvb_length_val, tvb_reported_length_val, offset, length_val, offset_ptr, length_ptr, exception)) {
		return FALSE;
	}

	/*
	 * Compute the offset of the first byte past the length.
	 */
	end_offset = *offset_ptr + *length_ptr;

	/*
	 * Check for an overflow, and clamp "end_offset" at the maximum
	 * if we got an overflow - that should force us to indicate that
	 * we're past the end of the tvbuff.
	 */
	if (end_offset < *offset_ptr)
		end_offset = UINT_MAX;

	/*
	 * Check whether that offset goes more than one byte past the
	 * end of the buffer.
	 *
	 * If not, return TRUE; otherwise, return FALSE and, if "exception"
	 * is non-null, return the appropriate exception through it.
	 */
	if (end_offset <= tvb_length_val) {
		return TRUE;
	}
	else if (end_offset <= tvb_reported_length_val) {
		if (exception) {
			*exception = BoundsError;
		}
	}
	else {
		if (exception) {
			*exception = ReportedBoundsError;
		}
	}

	return FALSE;
}

/* Checks (+/-) offset and length and throws an exception if
 * either is out of bounds. Sets integer ptrs to the new offset
 * and length. */
static void
check_offset_length(const guint tvb_length_val, const guint tvb_reported_length_val, const gint offset, gint const length_val,
		guint *offset_ptr, guint *length_ptr)
{
	int exception = 0;

	if (!check_offset_length_no_exception(tvb_length_val, tvb_reported_length_val, offset, length_val, offset_ptr, length_ptr, &exception)) {
		DISSECTOR_ASSERT(exception > 0);
		THROW(exception);
	}
}

static void
tvb_set_subset_no_exceptions(tvbuff_t *tvb, tvbuff_t *backing, const gint reported_length)
{
	tvb->tvbuffs.subset.tvb		= backing;
	tvb->length			= tvb->tvbuffs.subset.length;

	if (reported_length == -1) {
		tvb->reported_length	= backing->reported_length - tvb->tvbuffs.subset.offset;
	}
	else {
		tvb->reported_length	= reported_length;
	}
	tvb->initialized		= TRUE;
	add_to_chain(backing, tvb);

	/* Optimization. If the backing buffer has a pointer to contiguous, real data,
	 * then we can point directly to our starting offset in that buffer */
	if (backing->real_data != NULL) {
		tvb->real_data = backing->real_data + tvb->tvbuffs.subset.offset;
	}
}

void
tvb_set_subset(tvbuff_t *tvb, tvbuff_t *backing,
		const gint backing_offset, const gint backing_length, const gint reported_length)
{
	DISSECTOR_ASSERT(tvb);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_SUBSET);
	DISSECTOR_ASSERT(!tvb->initialized);

	THROW_ON(reported_length < -1, ReportedBoundsError);

	check_offset_length(backing->length, backing->reported_length, backing_offset, backing_length,
			&tvb->tvbuffs.subset.offset,
			&tvb->tvbuffs.subset.length);

	tvb_set_subset_no_exceptions(tvb, backing, reported_length);
}

tvbuff_t*
tvb_new_subset(tvbuff_t *backing, const gint backing_offset, const gint backing_length, const gint reported_length)
{
	tvbuff_t	*tvb;
	guint		subset_tvb_offset;
	guint		subset_tvb_length;

	DISSECTOR_ASSERT(backing && backing->initialized);

	THROW_ON(reported_length < -1, ReportedBoundsError);

	check_offset_length(backing->length, backing->reported_length, backing_offset, backing_length,
			&subset_tvb_offset,
			&subset_tvb_length);

	tvb = tvb_new_with_subset(subset_tvb_offset, subset_tvb_length);

	tvb_set_subset_no_exceptions(tvb, backing, reported_length);

	/*
	 * The top-level data source of this tvbuff is the top-level
	 * data source of its parent.
	 */
	tvb->ds_tvb = backing->ds_tvb;

	return tvb;
}

tvbuff_t*
tvb_new_subset_remaining(tvbuff_t *backing, const gint backing_offset)
{
	tvbuff_t	*tvb;
	guint		subset_tvb_offset;
	guint		subset_tvb_length;

	check_offset_length(backing->length, backing->reported_length, backing_offset, -1 /* backing_length */,
			&subset_tvb_offset,
			&subset_tvb_length);

	tvb = tvb_new_with_subset(subset_tvb_offset, subset_tvb_length);

	tvb_set_subset_no_exceptions(tvb, backing, -1 /* reported_length */);

	/*
	 * The top-level data source of this tvbuff is the top-level
	 * data source of its parent.
	 */
	tvb->ds_tvb = backing->ds_tvb;

	return tvb;
}

void
tvb_composite_append(tvbuff_t* tvb, tvbuff_t* member)
{
	tvb_comp_t	*composite;

	DISSECTOR_ASSERT(tvb && !tvb->initialized);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_COMPOSITE);
	composite = &tvb->tvbuffs.composite;
	composite->tvbs = g_slist_append( composite->tvbs, member );
}

void
tvb_composite_prepend(tvbuff_t* tvb, tvbuff_t* member)
{
	tvb_comp_t	*composite;

	DISSECTOR_ASSERT(tvb && !tvb->initialized);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_COMPOSITE);
	composite = &tvb->tvbuffs.composite;
	composite->tvbs = g_slist_prepend( composite->tvbs, member );
}

// ToDo: Need version which adds composite tvb to chain.
//       To keep things simple, member tvb's all need
//        to be part of one chain and the composite tvb must
//        be attached to that tvb.
//        How might this be enforced ??
//       Chain in at the finalize so that all members
//        preceed the composite tvb in the chain.
//        tvb_child_finalize_composite() ?
//       Or: just chain to the first member tvb at finalization ?
tvbuff_t*
tvb_new_composite(void)
{
	return tvb_new(TVBUFF_COMPOSITE);
}

void
tvb_composite_finalize(tvbuff_t* tvb)
{
	GSList		*slist;
	guint		num_members;
	tvbuff_t	*member_tvb;
	tvb_comp_t	*composite;
	int		i = 0;

	DISSECTOR_ASSERT(tvb && !tvb->initialized);
	DISSECTOR_ASSERT(tvb->type == TVBUFF_COMPOSITE);
	DISSECTOR_ASSERT(tvb->length == 0);
	DISSECTOR_ASSERT(tvb->reported_length == 0);

	composite = &tvb->tvbuffs.composite;
	num_members = g_slist_length(composite->tvbs);

	composite->start_offsets = g_new(guint, num_members);
	composite->end_offsets = g_new(guint, num_members);

	for (slist = composite->tvbs; slist != NULL; slist = slist->next) {
		DISSECTOR_ASSERT((guint) i < num_members);
		member_tvb = slist->data;
		composite->start_offsets[i] = tvb->length;
		tvb->length += member_tvb->length;
		tvb->reported_length += member_tvb->reported_length;
		composite->end_offsets[i] = tvb->length - 1;
		i++;
	}

	tvb->initialized = TRUE;
}



guint
tvb_length(const tvbuff_t* tvb)
{
	DISSECTOR_ASSERT(tvb && tvb->initialized);

	return tvb->length;
}

gint
tvb_length_remaining(const tvbuff_t *tvb, const gint offset)
{
	guint	abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	if (compute_offset_length(tvb->length, tvb->reported_length, offset, -1, &abs_offset, &abs_length, NULL)) {
		return abs_length;
	}
	else {
		return -1;
	}
}

guint
tvb_ensure_length_remaining(const tvbuff_t *tvb, const gint offset)
{
	guint	abs_offset, abs_length;
	int	exception;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	if (!compute_offset_length(tvb->length, tvb->reported_length, offset, -1, &abs_offset, &abs_length, &exception)) {
		THROW(exception);
	}
	if (abs_length == 0) {
		/*
		 * This routine ensures there's at least one byte available.
		 * There aren't any bytes available, so throw the appropriate
		 * exception.
		 */
		if (abs_offset >= tvb->reported_length)
			THROW(ReportedBoundsError);
		else
			THROW(BoundsError);
	}
	return abs_length;
}




/* Validates that 'length' bytes are available starting from
 * offset (pos/neg). Does not throw an exception. */
gboolean
tvb_bytes_exist(const tvbuff_t *tvb, const gint offset, const gint length)
{
	guint		abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	if (!compute_offset_length(tvb->length, tvb->reported_length, offset, length, &abs_offset, &abs_length, NULL))
		return FALSE;

	if (abs_offset + abs_length <= tvb->length) {
		return TRUE;
	}
	else {
		return FALSE;
	}
}

/* Validates that 'length' bytes are available starting from
 * offset (pos/neg). Throws an exception if they aren't. */
void
tvb_ensure_bytes_exist(const tvbuff_t *tvb, const gint offset, const gint length)
{
	guint		abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	/*
	 * -1 doesn't mean "until end of buffer", as that's pointless
	 * for this routine.  We must treat it as a Really Large Positive
	 * Number, so that we throw an exception; we throw
	 * ReportedBoundsError, as if it were past even the end of a
	 * reassembled packet, and past the end of even the data we
	 * didn't capture.
	 *
	 * We do the same with other negative lengths.
	 */
	if (length < 0) {
		THROW(ReportedBoundsError);
	}
	check_offset_length(tvb->length, tvb->reported_length, offset, length, &abs_offset, &abs_length);
}

gboolean
tvb_offset_exists(const tvbuff_t *tvb, const gint offset)
{
	guint		abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);
	if (!compute_offset_length(tvb->length, tvb->reported_length, offset, -1, &abs_offset, &abs_length, NULL))
		return FALSE;

	if (abs_offset < tvb->length) {
		return TRUE;
	}
	else {
		return FALSE;
	}
}

guint
tvb_reported_length(const tvbuff_t* tvb)
{
	DISSECTOR_ASSERT(tvb && tvb->initialized);

	return tvb->reported_length;
}

gint
tvb_reported_length_remaining(const tvbuff_t *tvb, const gint offset)
{
	guint	abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	if (compute_offset_length(tvb->length, tvb->reported_length, offset, -1, &abs_offset, &abs_length, NULL)) {
		if (tvb->reported_length >= abs_offset)
			return tvb->reported_length - abs_offset;
		else
			return -1;
	}
	else {
		return -1;
	}
}

/* Set the reported length of a tvbuff to a given value; used for protocols
   whose headers contain an explicit length and where the calling
   dissector's payload may include padding as well as the packet for
   this protocol.

   Also adjusts the data length. */
void
tvb_set_reported_length(tvbuff_t* tvb, const guint reported_length)
{
	DISSECTOR_ASSERT(tvb && tvb->initialized);

	if (reported_length > tvb->reported_length)
		THROW(ReportedBoundsError);

	tvb->reported_length = reported_length;
	if (reported_length < tvb->length)
		tvb->length = reported_length;
}


#if 0
static const guint8*
first_real_data_ptr(tvbuff_t *tvb)
{
	tvbuff_t	*member;

	switch(tvb->type) {
		case TVBUFF_REAL_DATA:
			return tvb->real_data;
		case TVBUFF_SUBSET:
			member = tvb->tvbuffs.subset.tvb;
			return first_real_data_ptr(member);
		case TVBUFF_COMPOSITE:
			member = tvb->tvbuffs.composite.tvbs->data;
			return first_real_data_ptr(member);
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return NULL;
}
#endif

static guint
offset_from_real_beginning(const tvbuff_t *tvb, const guint counter)
{
	tvbuff_t	*member;

	switch(tvb->type) {
		case TVBUFF_REAL_DATA:
			return counter;
		case TVBUFF_SUBSET:
			member = tvb->tvbuffs.subset.tvb;
			return offset_from_real_beginning(member, counter + tvb->tvbuffs.subset.offset);
		case TVBUFF_COMPOSITE:
			member = tvb->tvbuffs.composite.tvbs->data;
			return offset_from_real_beginning(member, counter);
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return 0;
}

guint
tvb_offset_from_real_beginning(const tvbuff_t *tvb)
{
	return offset_from_real_beginning(tvb, 0);
}

static const guint8*
composite_ensure_contiguous_no_exception(tvbuff_t *tvb, const guint abs_offset,
		const guint abs_length)
{
	guint		i, num_members;
	tvb_comp_t	*composite;
	tvbuff_t	*member_tvb = NULL;
	guint		member_offset, member_length;
	GSList		*slist;

	DISSECTOR_ASSERT(tvb->type == TVBUFF_COMPOSITE);

	/* Maybe the range specified by offset/length
	 * is contiguous inside one of the member tvbuffs */
	composite = &tvb->tvbuffs.composite;
	num_members = g_slist_length(composite->tvbs);

	for (i = 0; i < num_members; i++) {
		if (abs_offset <= composite->end_offsets[i]) {
			slist = g_slist_nth(composite->tvbs, i);
			member_tvb = slist->data;
			break;
		}
	}
	DISSECTOR_ASSERT(member_tvb);

	if (check_offset_length_no_exception(member_tvb->length, member_tvb->reported_length, abs_offset - composite->start_offsets[i],
				abs_length, &member_offset, &member_length, NULL)) {

		/*
		 * The range is, in fact, contiguous within member_tvb.
		 */
		DISSECTOR_ASSERT(!tvb->real_data);
		return ensure_contiguous_no_exception(member_tvb, member_offset, member_length, NULL);
	}
	else {
		tvb->real_data = tvb_memdup(tvb, 0, -1);
		return tvb->real_data + abs_offset;
	}

	DISSECTOR_ASSERT_NOT_REACHED();
}

static const guint8*
ensure_contiguous_no_exception(tvbuff_t *tvb, const gint offset, const gint length,
		int *exception)
{
	guint	abs_offset, abs_length;

	if (!check_offset_length_no_exception(tvb->length, tvb->reported_length, offset, length,
		&abs_offset, &abs_length, exception)) {
		return NULL;
	}

	/*
	 * We know that all the data is present in the tvbuff, so
	 * no exceptions should be thrown.
	 */
	if (tvb->real_data) {
		return tvb->real_data + abs_offset;
	}
	else {
		switch(tvb->type) {
			case TVBUFF_REAL_DATA:
				DISSECTOR_ASSERT_NOT_REACHED();
			case TVBUFF_SUBSET:
				return ensure_contiguous_no_exception(tvb->tvbuffs.subset.tvb,
						abs_offset - tvb->tvbuffs.subset.offset,
						abs_length, NULL);
			case TVBUFF_COMPOSITE:
				return composite_ensure_contiguous_no_exception(tvb, abs_offset, abs_length);
		}
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return NULL;
}

static const guint8*
ensure_contiguous(tvbuff_t *tvb, const gint offset, const gint length)
{
	int exception;
	const guint8 *p;

	p = ensure_contiguous_no_exception(tvb, offset, length, &exception);
	if (p == NULL) {
		DISSECTOR_ASSERT(exception > 0);
		THROW(exception);
	}
	return p;
}

static const guint8*
fast_ensure_contiguous(tvbuff_t *tvb, const gint offset, const guint length)
{
	guint	end_offset;
	guint	u_offset;

	DISSECTOR_ASSERT(tvb && tvb->initialized);
	/* We don't check for overflow in this fast path so we only handle simple types */
	DISSECTOR_ASSERT(length <= 8);

	if (offset < 0 || !tvb->real_data) {
		return ensure_contiguous(tvb, offset, length);
	}

	u_offset = offset;
	end_offset = u_offset + length;

	if (end_offset <= tvb->length) {
		return tvb->real_data + u_offset;
	}

	if (end_offset > tvb->reported_length) {
		THROW(ReportedBoundsError);
	}
	THROW(BoundsError);
	/* not reached */
	return NULL;
}

static const guint8*
guint8_pbrk(const guint8* haystack, size_t haystacklen, const guint8 *needles, guchar *found_needle)
{
	gchar tmp[256] = { 0 };
	const guint8 *haystack_end;

	while (*needles)
		tmp[*needles++] = 1;

	haystack_end = haystack + haystacklen;
	while (haystack < haystack_end) {
		if (tmp[*haystack]) {
			if(found_needle)
				*found_needle = *haystack;
			return haystack;
		}
		haystack++;
	}

	return NULL;
}



/************** ACCESSORS **************/

static void*
composite_memcpy(tvbuff_t *tvb, guint8* target, guint abs_offset, size_t abs_length)
{
	guint		i, num_members;
	tvb_comp_t	*composite;
	tvbuff_t	*member_tvb = NULL;
	guint		member_offset, member_length;
	gboolean	retval;
	GSList		*slist;

	DISSECTOR_ASSERT(tvb->type == TVBUFF_COMPOSITE);

	/* Maybe the range specified by offset/length
	 * is contiguous inside one of the member tvbuffs */
	composite = &tvb->tvbuffs.composite;
	num_members = g_slist_length(composite->tvbs);

	for (i = 0; i < num_members; i++) {
		if (abs_offset <= composite->end_offsets[i]) {
			slist = g_slist_nth(composite->tvbs, i);
			member_tvb = slist->data;
			break;
		}
	}
	DISSECTOR_ASSERT(member_tvb);

	if (check_offset_length_no_exception(member_tvb->length, member_tvb->reported_length, abs_offset - composite->start_offsets[i],
				(gint) abs_length, &member_offset, &member_length, NULL)) {

		DISSECTOR_ASSERT(!tvb->real_data);
		return tvb_memcpy(member_tvb, target, member_offset, member_length);
	}
	else {
		/* The requested data is non-contiguous inside
		 * the member tvb. We have to memcpy() the part that's in the member tvb,
		 * then iterate across the other member tvb's, copying their portions
		 * until we have copied all data.
		 */
		retval = compute_offset_length(member_tvb->length, member_tvb->reported_length, abs_offset - composite->start_offsets[i], -1,
				&member_offset, &member_length, NULL);
		DISSECTOR_ASSERT(retval);

		tvb_memcpy(member_tvb, target, member_offset, member_length);
		abs_offset	+= member_length;
		abs_length	-= member_length;

		/* Recurse */
		if (abs_length > 0) {
			composite_memcpy(tvb, target + member_length, abs_offset, abs_length);
		}

		return target;
	}

	DISSECTOR_ASSERT_NOT_REACHED();
}

void*
tvb_memcpy(tvbuff_t *tvb, void* target, const gint offset, size_t length)
{
	guint	abs_offset, abs_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	/*
	 * XXX - we should eliminate the "length = -1 means 'to the end
	 * of the tvbuff'" convention, and use other means to achieve
	 * that; this would let us eliminate a bunch of checks for
	 * negative lengths in cases where the protocol has a 32-bit
	 * length field.
	 *
	 * Allowing -1 but throwing an assertion on other negative
	 * lengths is a bit more work with the length being a size_t;
	 * instead, we check for a length <= 2^31-1.
	 */
	DISSECTOR_ASSERT(length <= 0x7FFFFFFF);
	check_offset_length(tvb->length, tvb->reported_length, offset, (gint) length, &abs_offset, &abs_length);

	if (tvb->real_data) {
		return memcpy(target, tvb->real_data + abs_offset, abs_length);
	}

	switch(tvb->type) {
		case TVBUFF_REAL_DATA:
			DISSECTOR_ASSERT_NOT_REACHED();

		case TVBUFF_SUBSET:
			return tvb_memcpy(tvb->tvbuffs.subset.tvb, target,
					abs_offset - tvb->tvbuffs.subset.offset,
					abs_length);

		case TVBUFF_COMPOSITE:
			return composite_memcpy(tvb, target, offset, length);
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return NULL;
}


/*
 * XXX - this doesn't treat a length of -1 as an error.
 * If it did, this could replace some code that calls
 * "tvb_ensure_bytes_exist()" and then allocates a buffer and copies
 * data to it.
 *
 * "composite_ensure_contiguous_no_exception()" depends on -1 not being
 * an error; does anything else depend on this routine treating -1 as
 * meaning "to the end of the buffer"?
 */
void*
tvb_memdup(tvbuff_t *tvb, const gint offset, size_t length)
{
	guint	abs_offset, abs_length;
	void	*duped;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, (gint) length, &abs_offset, &abs_length);

	duped = g_malloc(abs_length);
	return tvb_memcpy(tvb, duped, abs_offset, abs_length);
}

/*
 * XXX - this doesn't treat a length of -1 as an error.
 * If it did, this could replace some code that calls
 * "tvb_ensure_bytes_exist()" and then allocates a buffer and copies
 * data to it.
 *
 * "composite_ensure_contiguous_no_exception()" depends on -1 not being
 * an error; does anything else depend on this routine treating -1 as
 * meaning "to the end of the buffer"?
 *
 * This function allocates memory from a buffer with packet lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts decoding the next packet.
 * Do not use this function if you want the allocated memory to be persistent
 * after the current packet has been dissected.
 */
void*
ep_tvb_memdup(tvbuff_t *tvb, const gint offset, size_t length)
{
	guint	abs_offset, abs_length;
	void	*duped;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, (gint) length, &abs_offset, &abs_length);

	duped = ep_alloc(abs_length);
	return tvb_memcpy(tvb, duped, abs_offset, abs_length);
}



const guint8*
tvb_get_ptr(tvbuff_t *tvb, const gint offset, const gint length)
{
	return ensure_contiguous(tvb, offset, length);
}

/* ---------------- */
guint8
tvb_get_guint8(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint8));
	return *ptr;
}

guint16
tvb_get_ntohs(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint16));
	return pntohs(ptr);
}

guint32
tvb_get_ntoh24(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 3);
	return pntoh24(ptr);
}

guint32
tvb_get_ntohl(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint32));
	return pntohl(ptr);
}

guint64
tvb_get_ntoh40(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 5);
	return pntoh40(ptr);
}

guint64
tvb_get_ntoh48(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 6);
	return pntoh48(ptr);
}

guint64
tvb_get_ntoh56(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 7);
	return pntoh56(ptr);
}

guint64
tvb_get_ntoh64(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint64));
	return pntoh64(ptr);
}

/*
 * Stuff for IEEE float handling on platforms that don't have IEEE
 * format as the native floating-point format.
 *
 * For now, we treat only the VAX as such a platform.
 *
 * XXX - other non-IEEE boxes that can run UNIX include some Crays,
 * and possibly other machines.
 *
 * It appears that the official Linux port to System/390 and
 * zArchitecture uses IEEE format floating point (not a
 * huge surprise).
 *
 * I don't know whether there are any other machines that
 * could run Wireshark and that don't use IEEE format.
 * As far as I know, all of the main commercial microprocessor
 * families on which OSes that support Wireshark can run
 * use IEEE format (x86, 68k, SPARC, MIPS, PA-RISC, Alpha,
 * IA-64, and so on).
 */

#if defined(vax)

#include <math.h>

/*
 * Single-precision.
 */
#define IEEE_SP_NUMBER_WIDTH	32	/* bits in number */
#define IEEE_SP_EXP_WIDTH	8	/* bits in exponent */
#define IEEE_SP_MANTISSA_WIDTH	23	/* IEEE_SP_NUMBER_WIDTH - 1 - IEEE_SP_EXP_WIDTH */

#define IEEE_SP_SIGN_MASK	0x80000000
#define IEEE_SP_EXPONENT_MASK	0x7F800000
#define IEEE_SP_MANTISSA_MASK	0x007FFFFF
#define IEEE_SP_INFINITY	IEEE_SP_EXPONENT_MASK

#define IEEE_SP_IMPLIED_BIT (1 << IEEE_SP_MANTISSA_WIDTH)
#define IEEE_SP_INFINITE ((1 << IEEE_SP_EXP_WIDTH) - 1)
#define IEEE_SP_BIAS ((1 << (IEEE_SP_EXP_WIDTH - 1)) - 1)

static int
ieee_float_is_zero(const guint32 w)
{
	return ((w & ~IEEE_SP_SIGN_MASK) == 0);
}

static gfloat
get_ieee_float(const guint32 w)
{
	long sign;
	long exponent;
	long mantissa;

	sign = w & IEEE_SP_SIGN_MASK;
	exponent = w & IEEE_SP_EXPONENT_MASK;
	mantissa = w & IEEE_SP_MANTISSA_MASK;

	if (ieee_float_is_zero(w)) {
		/* number is zero, unnormalized, or not-a-number */
		return 0.0;
	}
#if 0
	/*
	 * XXX - how to handle this?
	 */
	if (IEEE_SP_INFINITY == exponent) {
		/*
		 * number is positive or negative infinity, or a special value
		 */
		return (sign? MINUS_INFINITY: PLUS_INFINITY);
	}
#endif

	exponent = ((exponent >> IEEE_SP_MANTISSA_WIDTH) - IEEE_SP_BIAS) -
		IEEE_SP_MANTISSA_WIDTH;
	mantissa |= IEEE_SP_IMPLIED_BIT;

	if (sign)
		return -mantissa * pow(2, exponent);
	else
		return mantissa * pow(2, exponent);
}

/*
 * Double-precision.
 * We assume that if you don't have IEEE floating-point, you have a
 * compiler that understands 64-bit integral quantities.
 */
#define IEEE_DP_NUMBER_WIDTH	64	/* bits in number */
#define IEEE_DP_EXP_WIDTH	11	/* bits in exponent */
#define IEEE_DP_MANTISSA_WIDTH	52	/* IEEE_DP_NUMBER_WIDTH - 1 - IEEE_DP_EXP_WIDTH */

#define IEEE_DP_SIGN_MASK	0x8000000000000000LL
#define IEEE_DP_EXPONENT_MASK	0x7FF0000000000000LL
#define IEEE_DP_MANTISSA_MASK	0x000FFFFFFFFFFFFFLL
#define IEEE_DP_INFINITY	IEEE_DP_EXPONENT_MASK

#define IEEE_DP_IMPLIED_BIT (1LL << IEEE_DP_MANTISSA_WIDTH)
#define IEEE_DP_INFINITE ((1 << IEEE_DP_EXP_WIDTH) - 1)
#define IEEE_DP_BIAS ((1 << (IEEE_DP_EXP_WIDTH - 1)) - 1)

static int
ieee_double_is_zero(const guint64 w)
{
	return ((w & ~IEEE_SP_SIGN_MASK) == 0);
}

static gdouble
get_ieee_double(const guint64 w)
{
	gint64 sign;
	gint64 exponent;
	gint64 mantissa;

	sign = w & IEEE_DP_SIGN_MASK;
	exponent = w & IEEE_DP_EXPONENT_MASK;
	mantissa = w & IEEE_DP_MANTISSA_MASK;

	if (ieee_double_is_zero(w)) {
		/* number is zero, unnormalized, or not-a-number */
		return 0.0;
	}
#if 0
	/*
	 * XXX - how to handle this?
	 */
	if (IEEE_DP_INFINITY == exponent) {
		/*
		 * number is positive or negative infinity, or a special value
		 */
		return (sign? MINUS_INFINITY: PLUS_INFINITY);
	}
#endif

	exponent = ((exponent >> IEEE_DP_MANTISSA_WIDTH) - IEEE_DP_BIAS) -
		IEEE_DP_MANTISSA_WIDTH;
	mantissa |= IEEE_DP_IMPLIED_BIT;

	if (sign)
		return -mantissa * pow(2, exponent);
	else
		return mantissa * pow(2, exponent);
}
#endif

/*
 * Fetches an IEEE single-precision floating-point number, in
 * big-endian form, and returns a "float".
 *
 * XXX - should this be "double", in case there are IEEE single-
 * precision numbers that won't fit in some platform's native
 * "float" format?
 */
gfloat
tvb_get_ntohieee_float(tvbuff_t *tvb, const int offset)
{
#if defined(vax)
	return get_ieee_float(tvb_get_ntohl(tvb, offset));
#else
	union {
		gfloat f;
		guint32 w;
	} ieee_fp_union;

	ieee_fp_union.w = tvb_get_ntohl(tvb, offset);
	return ieee_fp_union.f;
#endif
}

/*
 * Fetches an IEEE double-precision floating-point number, in
 * big-endian form, and returns a "double".
 */
gdouble
tvb_get_ntohieee_double(tvbuff_t *tvb, const int offset)
{
#if defined(vax)
	union {
		guint32 w[2];
		guint64 dw;
	} ieee_fp_union;
#else
	union {
		gdouble d;
		guint32 w[2];
	} ieee_fp_union;
#endif

#ifdef WORDS_BIGENDIAN
	ieee_fp_union.w[0] = tvb_get_ntohl(tvb, offset);
	ieee_fp_union.w[1] = tvb_get_ntohl(tvb, offset+4);
#else
	ieee_fp_union.w[0] = tvb_get_ntohl(tvb, offset+4);
	ieee_fp_union.w[1] = tvb_get_ntohl(tvb, offset);
#endif
#if defined(vax)
	return get_ieee_double(ieee_fp_union.dw);
#else
	return ieee_fp_union.d;
#endif
}

guint16
tvb_get_letohs(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint16));
	return pletohs(ptr);
}

guint32
tvb_get_letoh24(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 3);
	return pletoh24(ptr);
}

guint32
tvb_get_letohl(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint32));
	return pletohl(ptr);
}

guint64
tvb_get_letoh40(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 5);
	return pletoh40(ptr);
}

guint64
tvb_get_letoh48(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 6);
	return pletoh48(ptr);
}

guint64
tvb_get_letoh56(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, 7);
	return pletoh56(ptr);
}

guint64
tvb_get_letoh64(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint64));
	return pletoh64(ptr);
}

/*
 * Fetches an IEEE single-precision floating-point number, in
 * little-endian form, and returns a "float".
 *
 * XXX - should this be "double", in case there are IEEE single-
 * precision numbers that won't fit in some platform's native
 * "float" format?
 */
gfloat
tvb_get_letohieee_float(tvbuff_t *tvb, const int offset)
{
#if defined(vax)
	return get_ieee_float(tvb_get_letohl(tvb, offset));
#else
	union {
		gfloat f;
		guint32 w;
	} ieee_fp_union;

	ieee_fp_union.w = tvb_get_letohl(tvb, offset);
	return ieee_fp_union.f;
#endif
}

/*
 * Fetches an IEEE double-precision floating-point number, in
 * little-endian form, and returns a "double".
 */
gdouble
tvb_get_letohieee_double(tvbuff_t *tvb, const int offset)
{
#if defined(vax)
	union {
		guint32 w[2];
		guint64 dw;
	} ieee_fp_union;
#else
	union {
		gdouble d;
		guint32 w[2];
	} ieee_fp_union;
#endif

#ifdef WORDS_BIGENDIAN
	ieee_fp_union.w[0] = tvb_get_letohl(tvb, offset+4);
	ieee_fp_union.w[1] = tvb_get_letohl(tvb, offset);
#else
	ieee_fp_union.w[0] = tvb_get_letohl(tvb, offset);
	ieee_fp_union.w[1] = tvb_get_letohl(tvb, offset+4);
#endif
#if defined(vax)
	return get_ieee_double(ieee_fp_union.dw);
#else
	return ieee_fp_union.d;
#endif
}

/* Fetch an IPv4 address, in network byte order.
 * We do *not* convert them to host byte order; we leave them in
 * network byte order. */
guint32
tvb_get_ipv4(tvbuff_t *tvb, const gint offset)
{
	const guint8* ptr;
	guint32 addr;

	ptr = fast_ensure_contiguous(tvb, offset, sizeof(guint32));
	memcpy(&addr, ptr, sizeof addr);
	return addr;
}

/* Fetch an IPv6 address. */
void
tvb_get_ipv6(tvbuff_t *tvb, const gint offset, struct e_in6_addr *addr)
{
	const guint8* ptr;

	ptr = ensure_contiguous(tvb, offset, sizeof(*addr));
	memcpy(addr, ptr, sizeof *addr);
}

/* Fetch a GUID. */
void
tvb_get_ntohguid(tvbuff_t *tvb, const gint offset, e_guid_t *guid)
{
	ensure_contiguous(tvb, offset, sizeof(*guid));
	guid->data1 = tvb_get_ntohl(tvb, offset);
	guid->data2 = tvb_get_ntohs(tvb, offset + 4);
	guid->data3 = tvb_get_ntohs(tvb, offset + 6);
	tvb_memcpy(tvb, guid->data4, offset + 8, sizeof guid->data4);
}

void
tvb_get_letohguid(tvbuff_t *tvb, const gint offset, e_guid_t *guid)
{
	ensure_contiguous(tvb, offset, sizeof(*guid));
	guid->data1 = tvb_get_letohl(tvb, offset);
	guid->data2 = tvb_get_letohs(tvb, offset + 4);
	guid->data3 = tvb_get_letohs(tvb, offset + 6);
	tvb_memcpy(tvb, guid->data4, offset + 8, sizeof guid->data4);
}

/*
 * NOTE: to support code written when proto_tree_add_item() took a
 * gboolean as its last argument, with FALSE meaning "big-endian"
 * and TRUE meaning "little-endian", we treat any non-zero value of
 * "representation" as meaning "little-endian".
 */
void
tvb_get_guid(tvbuff_t *tvb, const gint offset, e_guid_t *guid, const guint representation)
{
	if (representation) {
		tvb_get_letohguid(tvb, offset, guid);
	} else {
		tvb_get_ntohguid(tvb, offset, guid);
	}
}

static const guint8 bit_mask8[] = {
	0xff,
	0x7f,
	0x3f,
	0x1f,
	0x0f,
	0x07,
	0x03,
	0x01
};

/* Bit offset mask for number of bits = 8 - 16 */
static const guint16 bit_mask16[] = {
	0xffff,
	0x7fff,
	0x3fff,
	0x1fff,
	0x0fff,
	0x07ff,
	0x03ff,
	0x01ff
};

/* Get 1 - 8 bits */
guint8
tvb_get_bits8(tvbuff_t *tvb, gint bit_offset, const gint no_of_bits)
{
	gint	offset;
	guint16	value = 0;
	guint8	tot_no_bits;

	if (no_of_bits>8) {
		DISSECTOR_ASSERT_NOT_REACHED();
	}
	/* Byte align offset */
	offset = bit_offset>>3;

	/* Find out which mask to use for the most significant octet
	 * by convering bit_offset into the offset into the first
	 * fetched octet.
	 */
	bit_offset = bit_offset & 0x7;
	tot_no_bits = bit_offset+no_of_bits;
	if(tot_no_bits<=8){
		/* Read one octet, mask off bit_offset bits and left shift out the unused bits */
		value = tvb_get_guint8(tvb,offset) & bit_mask8[bit_offset];
		value = value >> (8-tot_no_bits);
	}else{
		/* Read two octets, mask off bit_offset bits and left shift out the unused bits */
		value = tvb_get_ntohs(tvb,offset) & bit_mask16[bit_offset];
		value = value >> (16 - tot_no_bits);
	}

	return (guint8)value;
}

void
tvb_get_bits_buf(tvbuff_t *tvb, gint bit_offset, gint no_of_bits, guint8 *buf, gboolean lsb0)
{
	guint8 bit_mask, bit_shift;
	/* Byte align offset */
	gint offset = bit_offset >> 3;
	bit_offset = bit_offset & 0x7;

	bit_mask = (lsb0) ? 0xff : bit_mask8[bit_offset];
	bit_shift = (lsb0) ? bit_offset : (8 - bit_offset);

	if (G_LIKELY(bit_offset != 0)) {
		guint16 value = (guint16) tvb_get_guint8(tvb, offset);

		while (no_of_bits >= 8) {
			offset++;
			value = ((value & bit_mask) << 8) | tvb_get_guint8(tvb, offset);

			if (lsb0)
				*buf++ = (guint8) (GUINT16_SWAP_LE_BE(value) >> bit_shift);
			else
				*buf++ = (guint8) (value >> bit_shift);
			no_of_bits -= 8;
		}

		/* something left? */
		if (no_of_bits > 0) {
			guint8 tot_no_bits = bit_offset+no_of_bits;

			/* Overlaps with next byte? Get next byte */
			if (tot_no_bits > 8) {
				offset++;
				value = ((value & bit_mask) << 8) | tvb_get_guint8(tvb, offset);
			}

			if (lsb0) {
				if (tot_no_bits > 8)
					value = (GUINT16_SWAP_LE_BE(value) >> bit_offset) & (bit_mask8[8-no_of_bits]);
				else
					value = (value >> bit_offset) & (bit_mask8[8-no_of_bits]);

				/* value = (value & ((1 << tot_no_bits)-1)) >> bit_offset; */

			} else {
				if (tot_no_bits > 8)
					value = value >> (16 - tot_no_bits);
				else
					value = (value & bit_mask) >> (8-tot_no_bits);
			}
			*buf = (guint8) value;
		}

	} else {
		/* fast code path for bit_offset == 0 */
		while (no_of_bits >= 8) {
			*buf++ = tvb_get_guint8(tvb, offset);
			offset++;
			no_of_bits -= 8;
		}

		/* something left? */
		if (no_of_bits > 0) {
			if (lsb0)
				*buf = tvb_get_guint8(tvb, offset) & bit_mask8[8-no_of_bits]; /* read: ((1 << no_of_bits)-1) */
			else
				*buf = tvb_get_guint8(tvb, offset) >> (8-no_of_bits);
		}
	}
}

guint8 *
ep_tvb_get_bits(tvbuff_t *tvb, gint bit_offset, gint no_of_bits, gboolean lsb0)
{
	gint no_of_bytes;
	guint8 *buf;

	/* XXX, no_of_bits == -1 -> to end of tvb? */

	if (no_of_bits < 0 || bit_offset < 0) {
		DISSECTOR_ASSERT_NOT_REACHED();
	}

	no_of_bytes = (no_of_bits >> 3) + ((no_of_bits & 0x7) != 0);	/* ceil(no_of_bits / 8.0) */
	buf = ep_alloc(no_of_bytes);
	tvb_get_bits_buf(tvb, bit_offset, no_of_bits, buf, lsb0);
	return buf;
}

/* Get 9 - 16 bits */
/* Bit offset mask for number of bits = 16 - 32 */
static const guint32 bit_mask32[] = {
	0xffffffff,
	0x7fffffff,
	0x3fffffff,
	0x1fffffff,
	0x0fffffff,
	0x07ffffff,
	0x03ffffff,
	0x01ffffff
};

guint16
tvb_get_bits16(tvbuff_t *tvb, gint bit_offset, const gint no_of_bits,const guint encoding)
{
	gint	offset;
	guint16	value = 0;
	guint16	tempval = 0;
	guint8	tot_no_bits;

	if ((no_of_bits<=8)||(no_of_bits>16)) {
		/* If bits <= 8 use tvb_get_bits8 */
		DISSECTOR_ASSERT_NOT_REACHED();
	}
	/*
	 * For backwards compatibility, treat all non-zero values as
	 * meaning "little-endian".
	 */
	if(encoding){
		DISSECTOR_ASSERT_NOT_REACHED();
		/* This part is not implemented yet */
	}

	/* Byte align offset */
	offset = bit_offset>>3;

	/* Find out which mask to use for the most significant octet
	 * by convering bit_offset into the offset into the first
	 * fetched octet.
	 */
	bit_offset = bit_offset & 0x7;
	tot_no_bits = bit_offset+no_of_bits;
	/* Read two octets and mask off bit_offset bits */
	value = tvb_get_ntohs(tvb,offset) & bit_mask16[bit_offset];
	if(tot_no_bits < 16){
		/* Left shift out the unused bits */
		value = value >> (16 - tot_no_bits);
	}else if(tot_no_bits > 16){
		/* Spans three octets, read next octet and shift as needed */
		value = value << (tot_no_bits - 16);
		tempval = tvb_get_guint8(tvb,offset+2);
		tempval = tempval >> (24-tot_no_bits);
		value = value | tempval;
	}

	return value;
}

/* Bit offset mask for number of bits = 32 - 64 */
static const guint64 bit_mask64[] = {
	G_GINT64_CONSTANT(0xffffffffffffffffU),
	G_GINT64_CONSTANT(0x7fffffffffffffffU),
	G_GINT64_CONSTANT(0x3fffffffffffffffU),
	G_GINT64_CONSTANT(0x1fffffffffffffffU),
	G_GINT64_CONSTANT(0x0fffffffffffffffU),
	G_GINT64_CONSTANT(0x07ffffffffffffffU),
	G_GINT64_CONSTANT(0x03ffffffffffffffU),
	G_GINT64_CONSTANT(0x01ffffffffffffffU)
};

guint32
tvb_get_bits32(tvbuff_t *tvb, gint bit_offset, const gint no_of_bits, const guint encoding)
{
	gint	offset;
	guint32	value = 0;
	guint32	tempval = 0;
	guint8	tot_no_bits;
	guint8	tot_no_octets = 0;
	guint8	i = 0;
	gint8	shift = 0;

	if ((no_of_bits<=16)||(no_of_bits>32)) {
		/* If bits <= 16 use tvb_get_bits8 or tvb_get_bits16 */
		DISSECTOR_ASSERT_NOT_REACHED();
	}
	/*
	 * For backwards compatibility, treat all non-zero values as
	 * meaning "little-endian".
	 */
	if(encoding){
		DISSECTOR_ASSERT_NOT_REACHED();
		/* This part is not implemented yet */
	}

	/* Byte align offset */
	offset = bit_offset>>3;

	bit_offset = bit_offset & 0x7;
	tot_no_bits = bit_offset+no_of_bits;
	tot_no_octets = tot_no_bits / 8;
	if (tot_no_bits % 8)
		tot_no_octets++;
	shift = no_of_bits - (8 - bit_offset);

	value = tvb_get_guint8(tvb, offset) & bit_mask8[bit_offset];
	value = value << shift;

	for (i = 1; i < tot_no_octets; i++)
	{
		shift = shift - 8;
		tempval = tvb_get_guint8(tvb, offset+i);
		if (shift >= 0)
		{
			tempval = tempval << shift;
		}
		else
		{
			tempval = tempval >> (- shift);
		}
		value = value | tempval;
	}

	return value;
}

guint64
tvb_get_bits64(tvbuff_t *tvb, gint bit_offset, const gint no_of_bits, const guint encoding)
{
	gint	offset;
	guint64	value = 0;
	guint64	tempval = 0;
	guint8	tot_no_bits;

	if ((no_of_bits<=32)||(no_of_bits>64)) {
		/* If bits <= 32 use tvb_get_bits8, tvb_get_bits16 or tvb_get_bits32 */
		DISSECTOR_ASSERT_NOT_REACHED();
	}
	/*
	 * For backwards compatibility, treat all non-zero values as
	 * meaning "little-endian".
	 */
	if(encoding){
		DISSECTOR_ASSERT_NOT_REACHED();
		/* This part is not implemented yet */
	}

	/* Byte align offset */
	offset = bit_offset>>3;

	/* Find out which mask to use for the most significant octet
	 * by convering bit_offset into the offset into the first
	 * fetched octet.
	 */
	bit_offset = bit_offset & 0x7;
	tot_no_bits = bit_offset+no_of_bits;
	/* Read eight octets and mask off bit_offset bits */
	value = tvb_get_ntoh64(tvb,offset) & bit_mask64[bit_offset];
	if (tot_no_bits < 64){
		/* Left shift out the unused bits */
		value = value >> (64 - tot_no_bits);
	}else if (tot_no_bits > 64){
		/* Spans nine octets, read next octet and shift as needed */
		value = value << (tot_no_bits - 64);
		tempval = tvb_get_guint8(tvb,offset+8);
		tempval = tempval >> (72-tot_no_bits);
		value = value | tempval;
	}

	return value;
}

guint32
tvb_get_bits(tvbuff_t *tvb, const gint bit_offset, const gint no_of_bits, const guint encoding)
{
	/* This function can handle only up to 32 requested bits */
	if (no_of_bits > 32)
		DISSECTOR_ASSERT_NOT_REACHED();

	if (no_of_bits == 0)
		return 0;

	/* Number of requested bits is in range [17, 32] */
	if (no_of_bits > 16)
		return tvb_get_bits32(tvb, bit_offset, no_of_bits, encoding);

	/* Number of requested bits is in range [9, 16] */
	if (no_of_bits > 8)
		return tvb_get_bits16(tvb, bit_offset, no_of_bits, encoding);

	/* Number of requested bits is in range [1, 8] */
	return tvb_get_bits8(tvb, bit_offset, no_of_bits);
}

/* Find first occurence of needle in tvbuff, starting at offset. Searches
 * at most maxlength number of bytes; if maxlength is -1, searches to
 * end of tvbuff.
 * Returns the offset of the found needle, or -1 if not found.
 * Will not throw an exception, even if maxlength exceeds boundary of tvbuff;
 * in that case, -1 will be returned if the boundary is reached before
 * finding needle. */
gint
tvb_find_guint8(tvbuff_t *tvb, const gint offset, const gint maxlength, const guint8 needle)
{
	const guint8	*result;
	guint		abs_offset, junk_length;
	guint		tvbufflen;
	guint		limit;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, 0, &abs_offset, &junk_length);

	/* Only search to end of tvbuff, w/o throwing exception. */
	tvbufflen = tvb_length_remaining(tvb, abs_offset);
	if (maxlength == -1) {
		/* No maximum length specified; search to end of tvbuff. */
		limit = tvbufflen;
	}
	else if (tvbufflen < (guint) maxlength) {
		/* Maximum length goes past end of tvbuff; search to end
		   of tvbuff. */
		limit = tvbufflen;
	}
	else {
		/* Maximum length doesn't go past end of tvbuff; search
		   to that value. */
		limit = maxlength;
	}

	/* If we have real data, perform our search now. */
	if (tvb->real_data) {
		result = memchr(tvb->real_data + abs_offset, needle, limit);
		if (result == NULL) {
			return -1;
		}
		else {
			return (gint) (result - tvb->real_data);
		}
	}

	switch(tvb->type) {
		case TVBUFF_REAL_DATA:
			DISSECTOR_ASSERT_NOT_REACHED();

		case TVBUFF_SUBSET:
			return tvb_find_guint8(tvb->tvbuffs.subset.tvb,
					abs_offset - tvb->tvbuffs.subset.offset,
					limit, needle);

		case TVBUFF_COMPOSITE:
			DISSECTOR_ASSERT_NOT_REACHED();
			/* XXX - return composite_find_guint8(tvb, offset, limit, needle); */
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return -1;
}

/* Find first occurence of any of the needles in tvbuff, starting at offset.
 * Searches at most maxlength number of bytes; if maxlength is -1, searches
 * to end of tvbuff.
 * Returns the offset of the found needle, or -1 if not found.
 * Will not throw an exception, even if maxlength exceeds boundary of tvbuff;
 * in that case, -1 will be returned if the boundary is reached before
 * finding needle. */
gint
tvb_pbrk_guint8(tvbuff_t *tvb, const gint offset, const gint maxlength, const guint8 *needles, guchar *found_needle)
{
	const guint8	*result;
	guint		abs_offset, junk_length;
	guint		tvbufflen;
	guint		limit;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, 0, &abs_offset, &junk_length);

	/* Only search to end of tvbuff, w/o throwing exception. */
	tvbufflen = tvb_length_remaining(tvb, abs_offset);
	if (maxlength == -1) {
		/* No maximum length specified; search to end of tvbuff. */
		limit = tvbufflen;
	}
	else if (tvbufflen < (guint) maxlength) {
		/* Maximum length goes past end of tvbuff; search to end
		   of tvbuff. */
		limit = tvbufflen;
	}
	else {
		/* Maximum length doesn't go past end of tvbuff; search
		   to that value. */
		limit = maxlength;
	}

	/* If we have real data, perform our search now. */
	if (tvb->real_data) {
		result = guint8_pbrk(tvb->real_data + abs_offset, limit, needles, found_needle);
		if (result == NULL) {
			return -1;
		}
		else {
			return (gint) (result - tvb->real_data);
		}
	}

	switch(tvb->type) {
		case TVBUFF_REAL_DATA:
			DISSECTOR_ASSERT_NOT_REACHED();

		case TVBUFF_SUBSET:
			return tvb_pbrk_guint8(tvb->tvbuffs.subset.tvb,
					abs_offset - tvb->tvbuffs.subset.offset,
					limit, needles, found_needle);

		case TVBUFF_COMPOSITE:
			DISSECTOR_ASSERT_NOT_REACHED();
			/* XXX - return composite_pbrk_guint8(tvb, offset, limit, needle); */
	}

	DISSECTOR_ASSERT_NOT_REACHED();
	return -1;
}

/* Find size of stringz (NUL-terminated string) by looking for terminating
 * NUL.  The size of the string includes the terminating NUL.
 *
 * If the NUL isn't found, it throws the appropriate exception.
 */
guint
tvb_strsize(tvbuff_t *tvb, const gint offset)
{
	guint	abs_offset, junk_length;
	gint	nul_offset;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, 0, &abs_offset, &junk_length);
	nul_offset = tvb_find_guint8(tvb, abs_offset, -1, 0);
	if (nul_offset == -1) {
		/*
		 * OK, we hit the end of the tvbuff, so we should throw
		 * an exception.
		 *
		 * Did we hit the end of the captured data, or the end
		 * of the actual data?	If there's less captured data
		 * than actual data, we presumably hit the end of the
		 * captured data, otherwise we hit the end of the actual
		 * data.
		 */
		if (tvb_length(tvb) < tvb_reported_length(tvb)) {
			THROW(BoundsError);
		} else {
			THROW(ReportedBoundsError);
		}
	}
	return (nul_offset - abs_offset) + 1;
}

/* Unicode (UTF-16) version of tvb_strsize */
/* Returns number of *UTF-16 characters* (not bytes) excluding the null terminator */
guint
tvb_unicode_strsize(tvbuff_t *tvb, const gint offset)
{
	guint      i = 0;
	gunichar2  uchar;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	do {
		/* Endianness doesn't matter when looking for null */
		uchar = tvb_get_ntohs(tvb, offset + i);
		i += 2;
	} while(uchar != 0);

	return i; /* Number of *UTF-16* characters */
}

/* Find length of string by looking for end of string ('\0'), up to
 * 'maxlength' characters'; if 'maxlength' is -1, searches to end
 * of tvbuff.
 * Returns -1 if 'maxlength' reached before finding EOS. */
gint
tvb_strnlen(tvbuff_t *tvb, const gint offset, const guint maxlength)
{
	gint	result_offset;
	guint	abs_offset, junk_length;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	check_offset_length(tvb->length, tvb->reported_length, offset, 0, &abs_offset, &junk_length);

	result_offset = tvb_find_guint8(tvb, abs_offset, maxlength, 0);

	if (result_offset == -1) {
		return -1;
	}
	else {
		return result_offset - abs_offset;
	}
}

/*
 * Implement strneql etc
 */

/*
 * Call strncmp after checking if enough chars left, returning 0 if
 * it returns 0 (meaning "equal") and -1 otherwise, otherwise return -1.
 */
gint
tvb_strneql(tvbuff_t *tvb, const gint offset, const gchar *str, const size_t size)
{
	const guint8 *ptr;

	ptr = ensure_contiguous_no_exception(tvb, offset, (gint)size, NULL);

	if (ptr) {
		int cmp = strncmp((const char *)ptr, str, size);

		/*
		 * Return 0 if equal, -1 otherwise.
		 */
		return (cmp == 0 ? 0 : -1);
	} else {
		/*
		 * Not enough characters in the tvbuff to match the
		 * string.
		 */
		return -1;
	}
}

/*
 * Call g_ascii_strncasecmp after checking if enough chars left, returning
 * 0 if it returns 0 (meaning "equal") and -1 otherwise, otherwise return -1.
 */
gint
tvb_strncaseeql(tvbuff_t *tvb, const gint offset, const gchar *str, const size_t size)
{
	const guint8 *ptr;

	ptr = ensure_contiguous_no_exception(tvb, offset, (gint)size, NULL);

	if (ptr) {
		int cmp = g_ascii_strncasecmp((const char *)ptr, str, size);

		/*
		 * Return 0 if equal, -1 otherwise.
		 */
		return (cmp == 0 ? 0 : -1);
	} else {
		/*
		 * Not enough characters in the tvbuff to match the
		 * string.
		 */
		return -1;
	}
}

/*
 * Call memcmp after checking if enough chars left, returning 0 if
 * it returns 0 (meaning "equal") and -1 otherwise, otherwise return -1.
 */
gint
tvb_memeql(tvbuff_t *tvb, const gint offset, const guint8 *str, size_t size)
{
	const guint8 *ptr;

	ptr = ensure_contiguous_no_exception(tvb, offset, (gint) size, NULL);

	if (ptr) {
		int cmp = memcmp(ptr, str, size);

		/*
		 * Return 0 if equal, -1 otherwise.
		 */
		return (cmp == 0 ? 0 : -1);
	} else {
		/*
		 * Not enough characters in the tvbuff to match the
		 * string.
		 */
		return -1;
	}
}

/* Convert a string from Unicode to ASCII.	At the moment we fake it by
 * replacing all non-ASCII characters with a '.' )-:  The caller must
 * free the result returned.  The len parameter is the number of guint16's
 * to convert from Unicode. */
/* XXX - this function has been superceded by tvb_get_unicode_string() */
char *
tvb_fake_unicode(tvbuff_t *tvb, int offset, const int len, const gboolean little_endian)
{
	char *buffer;
	int i;
	guint16 character;

	/* Make sure we have enough data before allocating the buffer,
	   so we don't blow up if the length is huge. */
	tvb_ensure_bytes_exist(tvb, offset, 2*len);

	/* We know we won't throw an exception, so we don't have to worry
	   about leaking this buffer. */
	buffer = g_malloc(len + 1);

	for (i = 0; i < len; i++) {
		character = little_endian ? tvb_get_letohs(tvb, offset)
					  : tvb_get_ntohs(tvb, offset);
		buffer[i] = character < 256 ? character : '.';
		offset += 2;
	}

	buffer[len] = 0;

	return buffer;
}

/* Convert a string from Unicode to ASCII.	At the moment we fake it by
 * replacing all non-ASCII characters with a '.' )-:   The len parameter is
 * the number of guint16's to convert from Unicode.
 *
 * This function allocates memory from a buffer with packet lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts decoding the next packet.
 */
/* XXX: This has been replaced by tvb_get_ephemeral_unicode_string() */
char *
tvb_get_ephemeral_faked_unicode(tvbuff_t *tvb, int offset, const int len, const gboolean little_endian)
{
	char *buffer;
	int i;
	guint16 character;

	/* Make sure we have enough data before allocating the buffer,
	   so we don't blow up if the length is huge. */
	tvb_ensure_bytes_exist(tvb, offset, 2*len);

	/* We know we won't throw an exception, so we don't have to worry
	   about leaking this buffer. */
	buffer = ep_alloc(len + 1);

	for (i = 0; i < len; i++) {
		character = little_endian ? tvb_get_letohs(tvb, offset)
					  : tvb_get_ntohs(tvb, offset);
		buffer[i] = character < 256 ? character : '.';
		offset += 2;
	}

	buffer[len] = 0;

	return buffer;
}

/*
 * Format the data in the tvb from offset for length ...
 */

gchar *
tvb_format_text(tvbuff_t *tvb, const gint offset, const gint size)
{
  const guint8 *ptr;
  gint len = size;

  if ((ptr = ensure_contiguous(tvb, offset, size)) == NULL) {
    len = tvb_length_remaining(tvb, offset);
    ptr = ensure_contiguous(tvb, offset, len);
  }

  return format_text(ptr, len);
}

/*
 * Format the data in the tvb from offset for length ...
 */

gchar *
tvb_format_text_wsp(tvbuff_t *tvb, const gint offset, const gint size)
{
  const guint8 *ptr;
  gint len = size;

  if ((ptr = ensure_contiguous(tvb, offset, size)) == NULL) {

	len = tvb_length_remaining(tvb, offset);
	ptr = ensure_contiguous(tvb, offset, len);

  }

  return format_text_wsp(ptr, len);

}

/*
 * Like "tvb_format_text()", but for null-padded strings; don't show
 * the null padding characters as "\000".
 */
gchar *
tvb_format_stringzpad(tvbuff_t *tvb, const gint offset, const gint size)
{
  const guint8 *ptr, *p;
  gint len = size;
  gint stringlen;

  if ((ptr = ensure_contiguous(tvb, offset, size)) == NULL) {

	len = tvb_length_remaining(tvb, offset);
	ptr = ensure_contiguous(tvb, offset, len);

  }

  for (p = ptr, stringlen = 0; stringlen < len && *p != '\0'; p++, stringlen++)
	;
  return format_text(ptr, stringlen);

}

/*
 * Like "tvb_format_text_wsp()", but for null-padded strings; don't show
 * the null padding characters as "\000".
 */
gchar *
tvb_format_stringzpad_wsp(tvbuff_t *tvb, const gint offset, const gint size)
{
  const guint8 *ptr, *p;
  gint len = size;
  gint stringlen;

  if ((ptr = ensure_contiguous(tvb, offset, size)) == NULL) {

	len = tvb_length_remaining(tvb, offset);
	ptr = ensure_contiguous(tvb, offset, len);

  }

  for (p = ptr, stringlen = 0; stringlen < len && *p != '\0'; p++, stringlen++)
	;
  return format_text_wsp(ptr, stringlen);

}

/*
 * Given a tvbuff, an offset, and a length, allocate a buffer big enough
 * to hold a non-null-terminated string of that length at that offset,
 * plus a trailing '\0', copy the string into it, and return a pointer
 * to the string.
 *
 * Throws an exception if the tvbuff ends before the string does.
 */
guint8 *
tvb_get_string(tvbuff_t *tvb, const gint offset, const gint length)
{
	const guint8 *ptr;
	guint8 *strbuf = NULL;

	tvb_ensure_bytes_exist(tvb, offset, length);

	ptr = ensure_contiguous(tvb, offset, length);
	strbuf = g_malloc(length + 1);
	if (length != 0) {
		memcpy(strbuf, ptr, length);
	}
	strbuf[length] = '\0';
	return strbuf;
}

/*
 * Unicode (UTF-16) version of tvb_get_string()
 *
 * Encoding paramter should be ENC_BIG_ENDIAN or ENC_LITTLE_ENDIAN
 *
 * Specify length in bytes
 *
 * Returns an UTF-8 string that must be freed by the caller
 */
gchar *
tvb_get_unicode_string(tvbuff_t *tvb, const gint offset, gint length, const guint encoding)
{
	gchar *tmpbuf = NULL;
	gunichar2 uchar;
	gint i; /* Byte counter for tvbuff */
	gint tmpbuf_len;
	GString *strbuf = NULL;

	strbuf = g_string_new(NULL);

	for(i = 0; i < length; i += 2) {

		if(encoding == ENC_BIG_ENDIAN)
			uchar = tvb_get_ntohs(tvb, offset + i);
		else
			uchar = tvb_get_letohs(tvb, offset + i);

		/* Calculate how much space is needed to store UTF-16 character
		 * in UTF-8 */
		tmpbuf_len = g_unichar_to_utf8(uchar, NULL);

		tmpbuf = g_malloc(tmpbuf_len + 1); /* + 1 to make room for null
						    * terminator */

		g_unichar_to_utf8(uchar, tmpbuf);

		/* NULL terminate the tmpbuf so g_string_append knows where
		 * to stop */
		tmpbuf[tmpbuf_len] = '\0';

		g_string_append(strbuf, tmpbuf);

		g_free(tmpbuf);
	}

	return g_string_free(strbuf, FALSE);
}

/*
 * Given a tvbuff, an offset, a length, and an encoding, allocate a
 * buffer big enough to hold a non-null-terminated string of that length
 * at that offset, plus a trailing '\0', copy the string into it, and
 * return a pointer to the string; if the encoding is EBCDIC, map
 * the string from EBCDIC to ASCII.
 *
 * Throws an exception if the tvbuff ends before the string does.
 *
 * This function allocates memory from a buffer with packet lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts decoding the next packet.
 * Do not use this function if you want the allocated memory to be persistent
 * after the current packet has been dissected.
 */
guint8 *
tvb_get_ephemeral_string_enc(tvbuff_t *tvb, const gint offset,
    const gint length, const gint encoding)
{
	const guint8 *ptr;
	guint8 *strbuf = NULL;

	tvb_ensure_bytes_exist(tvb, offset, length);

	ptr = ensure_contiguous(tvb, offset, length);
	strbuf = ep_alloc(length + 1);
	if (length != 0) {
		memcpy(strbuf, ptr, length);
	}
	if ((encoding & ENC_CHARENCODING_MASK) == ENC_EBCDIC)
		EBCDIC_to_ASCII(strbuf, length);
	strbuf[length] = '\0';
	return strbuf;
}

guint8 *
tvb_get_ephemeral_string(tvbuff_t *tvb, const gint offset,
    const gint length)
{
	return tvb_get_ephemeral_string_enc(tvb, offset, length, ENC_UTF_8|ENC_NA);
}

/*
 * Unicode (UTF-16) version of tvb_get_ephemeral_string()
 *
 * Encoding paramter should be ENC_BIG_ENDIAN or ENC_LITTLE_ENDIAN
 *
 * Specify length in bytes
 *
 * Returns an ep_ allocated UTF-8 string
 */
gchar *
tvb_get_ephemeral_unicode_string(tvbuff_t *tvb, const gint offset, gint length, const guint encoding)
{
	gchar *tmpbuf = NULL;
	gunichar2 uchar;
	gint i; /* Byte counter for tvbuff */
	gint tmpbuf_len;
	emem_strbuf_t *strbuf = NULL;

	strbuf = ep_strbuf_new(NULL);

	for(i = 0; i < length; i += 2) {

		if(encoding == ENC_BIG_ENDIAN)
			uchar = tvb_get_ntohs(tvb, offset + i);
		else
			uchar = tvb_get_letohs(tvb, offset + i);

		/* Calculate how much space is needed to store UTF-16 character
		 * in UTF-8 */
		tmpbuf_len = g_unichar_to_utf8(uchar, NULL);

		tmpbuf = g_malloc(tmpbuf_len + 1); /* + 1 to make room for null
						    * terminator */

		g_unichar_to_utf8(uchar, tmpbuf);

		/* NULL terminate the tmpbuf so ep_strbuf_append knows where
		 * to stop */
		tmpbuf[tmpbuf_len] = '\0';

		ep_strbuf_append(strbuf, tmpbuf);

		g_free(tmpbuf);
	}

	return strbuf->str;
}

/*
 * Given a tvbuff, an offset, and a length, allocate a buffer big enough
 * to hold a non-null-terminated string of that length at that offset,
 * plus a trailing '\0', copy the string into it, and return a pointer
 * to the string.
 *
 * Throws an exception if the tvbuff ends before the string does.
 *
 * This function allocates memory from a buffer with capture session lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts or opens a new capture.
 */
guint8 *
tvb_get_seasonal_string(tvbuff_t *tvb, const gint offset, const gint length)
{
	const guint8 *ptr;
	guint8 *strbuf = NULL;

	tvb_ensure_bytes_exist(tvb, offset, length);

	ptr = ensure_contiguous(tvb, offset, length);
	strbuf = se_alloc(length + 1);
	if (length != 0) {
		memcpy(strbuf, ptr, length);
	}
	strbuf[length] = '\0';
	return strbuf;
}

/*
 * Given a tvbuff, an offset, and an encoding, with the offset assumed
 * to refer to a null-terminated string, find the length of that string
 * (and throw an exception if the tvbuff ends before we find the null),
 * allocate a buffer big enough to hold the string, copy the string into
 * it, and return a pointer to the string; if the encoding is EBCDIC, map
 * the string from EBCDIC to ASCII.  Also return the length of the
 * string (including the terminating null) through a pointer.
 */
guint8 *
tvb_get_stringz_enc(tvbuff_t *tvb, const gint offset, gint *lengthp, gint encoding)
{
	guint size;
	guint8 *strptr;

	size = tvb_strsize(tvb, offset);
	strptr = g_malloc(size);
	tvb_memcpy(tvb, strptr, offset, size);
	if ((encoding & ENC_CHARENCODING_MASK) == ENC_EBCDIC)
		EBCDIC_to_ASCII(strptr, size);
	if (lengthp)
		*lengthp = size;
	return strptr;
}

guint8 *
tvb_get_stringz(tvbuff_t *tvb, const gint offset, gint *lengthp)
{
	return tvb_get_stringz_enc(tvb, offset, lengthp, ENC_UTF_8|ENC_NA);
}

/*
 * Given a tvbuff and an offset, with the offset assumed to refer to
 * a null-terminated string, find the length of that string (and throw
 * an exception if the tvbuff ends before we find the null), ensure that
 * the TVB is flat, and return a pointer to the string (in the TVB).
 * Also return the length of the string (including the terminating null)
 * through a pointer.
 *
 * As long as we aren't using composite TVBs, this saves the cycles used
 * (often unnecessariliy) in allocating a buffer and copying the string into
 * it.  (If we do start using composite TVBs, we may want to replace this
 * function with the _ephemeral versoin.)
 */
const guint8 *
tvb_get_const_stringz(tvbuff_t *tvb, const gint offset, gint *lengthp)
{
	guint size;
	const guint8 *strptr;

	size = tvb_strsize(tvb, offset);
	strptr = ensure_contiguous(tvb, offset, size);
	if (lengthp)
		*lengthp = size;
	return strptr;
}

/*
 * Given a tvbuff and an offset, with the offset assumed to refer to
 * a null-terminated string, find the length of that string (and throw
 * an exception if the tvbuff ends before we find the null), allocate
 * a buffer big enough to hold the string, copy the string into it,
 * and return a pointer to the string.	Also return the length of the
 * string (including the terminating null) through a pointer.
 *
 * This function allocates memory from a buffer with packet lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts decoding the next packet.
 * Do not use this function if you want the allocated memory to be persistent
 * after the current packet has been dissected.
 */
guint8 *
tvb_get_ephemeral_stringz_enc(tvbuff_t *tvb, const gint offset, gint *lengthp, gint encoding)
{
	guint size;
	guint8 *strptr;

	size = tvb_strsize(tvb, offset);
	strptr = ep_alloc(size);
	tvb_memcpy(tvb, strptr, offset, size);
	if ((encoding & ENC_CHARENCODING_MASK) == ENC_EBCDIC)
		EBCDIC_to_ASCII(strptr, size);
	if (lengthp)
		*lengthp = size;
	return strptr;
}

guint8 *
tvb_get_ephemeral_stringz(tvbuff_t *tvb, const gint offset, gint *lengthp)
{
	return tvb_get_ephemeral_stringz_enc(tvb, offset, lengthp, ENC_UTF_8|ENC_NA);
}

/*
 * Unicode (UTF-16) version of tvb_get_ephemeral_stringz()
 *
 * Encoding paramter should be ENC_BIG_ENDIAN or ENC_LITTLE_ENDIAN
 *
 * Returns an ep_ allocated UTF-8 string and updates lengthp pointer with length of string (in bytes)
 */
gchar *
tvb_get_ephemeral_unicode_stringz(tvbuff_t *tvb, const gint offset, gint *lengthp, const guint encoding)
{
	gchar *tmpbuf = NULL;
	gunichar2 uchar;
	gint size; /* Number of UTF-16 characters */
	gint i; /* Byte counter for tvbuff */
	gint tmpbuf_len;
	emem_strbuf_t *strbuf = NULL;

	strbuf = ep_strbuf_new(NULL);

	size = tvb_unicode_strsize(tvb, offset);

	for(i = 0; i < size; i += 2) {

		if(encoding == ENC_BIG_ENDIAN)
			uchar = tvb_get_ntohs(tvb, offset + i);
		else
			uchar = tvb_get_letohs(tvb, offset + i);

		/* Calculate how much space is needed to store UTF-16 character
		 * in UTF-8 */
		tmpbuf_len = g_unichar_to_utf8(uchar, NULL);

		tmpbuf = g_malloc(tmpbuf_len + 1); /* + 1 to make room for null
						    * terminator */

		g_unichar_to_utf8(uchar, tmpbuf);

		/* NULL terminate the tmpbuf so ep_strbuf_append knows where
		 * to stop */
		tmpbuf[tmpbuf_len] = '\0';

		ep_strbuf_append(strbuf, tmpbuf);

		g_free(tmpbuf);
	}

	if(lengthp)
		*lengthp = i; /* Number of *bytes* processed */

	return strbuf->str;
}

/*
 * Given a tvbuff and an offset, with the offset assumed to refer to
 * a null-terminated string, find the length of that string (and throw
 * an exception if the tvbuff ends before we find the null), allocate
 * a buffer big enough to hold the string, copy the string into it,
 * and return a pointer to the string.	Also return the length of the
 * string (including the terminating null) through a pointer.
 *
 * This function allocates memory from a buffer with capture session lifetime.
 * You do not have to free this buffer, it will be automatically freed
 * when wireshark starts or opens a new capture.
 */
guint8 *
tvb_get_seasonal_stringz(tvbuff_t *tvb, const gint offset, gint *lengthp)
{
	guint size;
	guint8 *strptr;

	size = tvb_strsize(tvb, offset);
	strptr = se_alloc(size);
	tvb_memcpy(tvb, strptr, offset, size);
	if (lengthp)
		*lengthp = size;
	return strptr;
}

/* Looks for a stringz (NUL-terminated string) in tvbuff and copies
 * no more than bufsize number of bytes, including terminating NUL, to buffer.
 * Returns length of string (not including terminating NUL), or -1 if the string was
 * truncated in the buffer due to not having reached the terminating NUL.
 * In this way, it acts like g_snprintf().
 *
 * bufsize MUST be greater than 0.
 *
 * When processing a packet where the remaining number of bytes is less
 * than bufsize, an exception is not thrown if the end of the packet
 * is reached before the NUL is found. If no NUL is found before reaching
 * the end of the short packet, -1 is still returned, and the string
 * is truncated with a NUL, albeit not at buffer[bufsize - 1], but
 * at the correct spot, terminating the string.
 *
 * *bytes_copied will contain the number of bytes actually copied,
 * including the terminating-NUL.
 */
static gint
_tvb_get_nstringz(tvbuff_t *tvb, const gint offset, const guint bufsize, guint8* buffer,
		gint *bytes_copied)
{
	gint	stringlen;
	guint	abs_offset, junk_length;
	gint	limit, len;
	gboolean decreased_max = FALSE;

	check_offset_length(tvb->length, tvb->reported_length, offset, 0, &abs_offset, &junk_length);

	/* There must at least be room for the terminating NUL. */
	DISSECTOR_ASSERT(bufsize != 0);

	/* If there's no room for anything else, just return the NUL. */
	if (bufsize == 1) {
		buffer[0] = 0;
		*bytes_copied = 1;
		return 0;
	}

	/* Only read to end of tvbuff, w/o throwing exception. */
	len = tvb_length_remaining(tvb, abs_offset);

	/* check_offset_length() won't throw an exception if we're
	 * looking at the byte immediately after the end of the tvbuff. */
	if (len == 0) {
		THROW(ReportedBoundsError);
	}

	/* This should not happen because check_offset_length() would
	 * have already thrown an exception if 'offset' were out-of-bounds.
	 */
	DISSECTOR_ASSERT(len != -1);

	/*
	 * If we've been passed a negative number, bufsize will
	 * be huge.
	 */
	DISSECTOR_ASSERT(bufsize <= G_MAXINT);

	if ((guint)len < bufsize) {
		limit = len;
		decreased_max = TRUE;
	}
	else {
		limit = bufsize;
	}

	stringlen = tvb_strnlen(tvb, abs_offset, limit - 1);
	/* If NUL wasn't found, copy the data and return -1 */
	if (stringlen == -1) {
		tvb_memcpy(tvb, buffer, abs_offset, limit);
		if (decreased_max) {
			buffer[limit] = 0;
			/* Add 1 for the extra NUL that we set at buffer[limit],
			 * pretending that it was copied as part of the string. */
			*bytes_copied = limit + 1;
		}
		else {
			*bytes_copied = limit;
		}
		return -1;
	}

	/* Copy the string to buffer */
	tvb_memcpy(tvb, buffer, abs_offset, stringlen + 1);
	*bytes_copied = stringlen + 1;
	return stringlen;
}

/* Looks for a stringz (NUL-terminated string) in tvbuff and copies
 * no more than bufsize number of bytes, including terminating NUL, to buffer.
 * Returns length of string (not including terminating NUL), or -1 if the string was
 * truncated in the buffer due to not having reached the terminating NUL.
 * In this way, it acts like g_snprintf().
 *
 * When processing a packet where the remaining number of bytes is less
 * than bufsize, an exception is not thrown if the end of the packet
 * is reached before the NUL is found. If no NUL is found before reaching
 * the end of the short packet, -1 is still returned, and the string
 * is truncated with a NUL, albeit not at buffer[bufsize - 1], but
 * at the correct spot, terminating the string.
 */
gint
tvb_get_nstringz(tvbuff_t *tvb, const gint offset, const guint bufsize, guint8* buffer)
{
	gint bytes_copied;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	return _tvb_get_nstringz(tvb, offset, bufsize, buffer, &bytes_copied);
}

/* Like tvb_get_nstringz(), but never returns -1. The string is guaranteed to
 * have a terminating NUL. If the string was truncated when copied into buffer,
 * a NUL is placed at the end of buffer to terminate it.
 */
gint
tvb_get_nstringz0(tvbuff_t *tvb, const gint offset, const guint bufsize, guint8* buffer)
{
	gint	len, bytes_copied;

	DISSECTOR_ASSERT(tvb && tvb->initialized);

	len = _tvb_get_nstringz(tvb, offset, bufsize, buffer, &bytes_copied);

	if (len == -1) {
		buffer[bufsize - 1] = 0;
		return bytes_copied - 1;
	}
	else {
		return len;
	}
}

/*
 * Given a tvbuff, an offset into the tvbuff, and a length that starts
 * at that offset (which may be -1 for "all the way to the end of the
 * tvbuff"), find the end of the (putative) line that starts at the
 * specified offset in the tvbuff, going no further than the specified
 * length.
 *
 * Return the length of the line (not counting the line terminator at
 * the end), or, if we don't find a line terminator:
 *
 *	if "deseg" is true, return -1;
 *
 *	if "deseg" is false, return the amount of data remaining in
 *	the buffer.
 *
 * Set "*next_offset" to the offset of the character past the line
 * terminator, or past the end of the buffer if we don't find a line
 * terminator.	(It's not set if we return -1.)
 */
gint
tvb_find_line_end(tvbuff_t *tvb, const gint offset, int len, gint *next_offset,
	const gboolean desegment)
{
	gint eob_offset;
	gint eol_offset;
	int linelen;
	guchar found_needle = 0;

	if (len == -1)
		len = tvb_length_remaining(tvb, offset);
	/*
	 * XXX - what if "len" is still -1, meaning "offset is past the
	 * end of the tvbuff"?
	 */
	eob_offset = offset + len;

	/*
	 * Look either for a CR or an LF.
	 */
	eol_offset = tvb_pbrk_guint8(tvb, offset, len, (const guint8 *)"\r\n", &found_needle);
	if (eol_offset == -1) {
		/*
		 * No CR or LF - line is presumably continued in next packet.
		 */
		if (desegment) {
			/*
			 * Tell our caller we saw no EOL, so they can
			 * try to desegment and get the entire line
			 * into one tvbuff.
			 */
			return -1;
		} else {
			/*
			 * Pretend the line runs to the end of the tvbuff.
			 */
			linelen = eob_offset - offset;
			if (next_offset)
				*next_offset = eob_offset;
		}
	} else {
		/*
		 * Find the number of bytes between the starting offset
		 * and the CR or LF.
		 */
		linelen = eol_offset - offset;

		/*
		 * Is it a CR?
		 */
		if (found_needle == '\r') {
			/*
			 * Yes - is it followed by an LF?
			 */
			if (eol_offset + 1 >= eob_offset) {
				/*
				 * Dunno - the next byte isn't in this
				 * tvbuff.
				 */
				if (desegment) {
					/*
					 * We'll return -1, although that
					 * runs the risk that if the line
					 * really *is* terminated with a CR,
					 * we won't properly dissect this
					 * tvbuff.
					 *
					 * It's probably more likely that
					 * the line ends with CR-LF than
					 * that it ends with CR by itself.
					 */
					return -1;
				}
			} else {
				/*
				 * Well, we can at least look at the next
				 * byte.
				 */
				if (tvb_get_guint8(tvb, eol_offset + 1) == '\n') {
					/*
					 * It's an LF; skip over the CR.
					 */
					eol_offset++;
				}
			}
		}

		/*
		 * Return the offset of the character after the last
		 * character in the line, skipping over the last character
		 * in the line terminator.
		 */
		if (next_offset)
			*next_offset = eol_offset + 1;
	}
	return linelen;
}

/*
 * Given a tvbuff, an offset into the tvbuff, and a length that starts
 * at that offset (which may be -1 for "all the way to the end of the
 * tvbuff"), find the end of the (putative) line that starts at the
 * specified offset in the tvbuff, going no further than the specified
 * length.
 *
 * However, treat quoted strings inside the buffer specially - don't
 * treat newlines in quoted strings as line terminators.
 *
 * Return the length of the line (not counting the line terminator at
 * the end), or the amount of data remaining in the buffer if we don't
 * find a line terminator.
 *
 * Set "*next_offset" to the offset of the character past the line
 * terminator, or past the end of the buffer if we don't find a line
 * terminator.
 */
gint
tvb_find_line_end_unquoted(tvbuff_t *tvb, const gint offset, int len,
	gint *next_offset)
{
	gint cur_offset, char_offset;
	gboolean is_quoted;
	guchar c = 0;
	gint eob_offset;
	int linelen;

	if (len == -1)
		len = tvb_length_remaining(tvb, offset);
	/*
	 * XXX - what if "len" is still -1, meaning "offset is past the
	 * end of the tvbuff"?
	 */
	eob_offset = offset + len;

	cur_offset = offset;
	is_quoted = FALSE;
	for (;;) {
			/*
		 * Is this part of the string quoted?
		 */
		if (is_quoted) {
			/*
			 * Yes - look only for the terminating quote.
			 */
			char_offset = tvb_find_guint8(tvb, cur_offset, len,
				'"');
		} else {
			/*
			 * Look either for a CR, an LF, or a '"'.
			 */
			char_offset = tvb_pbrk_guint8(tvb, cur_offset, len,
				(const guint8 *)"\r\n\"", &c);
		}
		if (char_offset == -1) {
			/*
			 * Not found - line is presumably continued in
			 * next packet.
			 * We pretend the line runs to the end of the tvbuff.
			 */
			linelen = eob_offset - offset;
			if (next_offset)
				*next_offset = eob_offset;
			break;
		}

		if (is_quoted) {
			/*
			 * We're processing a quoted string.
			 * We only looked for ", so we know it's a ";
			 * as we're processing a quoted string, it's a
			 * closing quote.
			 */
			is_quoted = FALSE;
		} else {
			/*
			 * OK, what is it?
			 */
			if (c == '"') {
				/*
				 * Un-quoted "; it begins a quoted
				 * string.
				 */
				is_quoted = TRUE;
			} else {
				/*
				 * It's a CR or LF; we've found a line
				 * terminator.
				 *
				 * Find the number of bytes between the
				 * starting offset and the CR or LF.
				 */
				linelen = char_offset - offset;

				/*
				 * Is it a CR?
				 */
				if (c == '\r') {
					/*
					 * Yes; is it followed by an LF?
					 */
					if (char_offset + 1 < eob_offset &&
						tvb_get_guint8(tvb, char_offset + 1)
						  == '\n') {
						/*
						 * Yes; skip over the CR.
						 */
						char_offset++;
					}
				}

				/*
				 * Return the offset of the character after
				 * the last character in the line, skipping
				 * over the last character in the line
				 * terminator, and quit.
				 */
				if (next_offset)
					*next_offset = char_offset + 1;
				break;
			}
		}

		/*
		 * Step past the character we found.
		 */
		cur_offset = char_offset + 1;
		if (cur_offset >= eob_offset) {
			/*
			 * The character we found was the last character
			 * in the tvbuff - line is presumably continued in
			 * next packet.
			 * We pretend the line runs to the end of the tvbuff.
			 */
			linelen = eob_offset - offset;
			if (next_offset)
				*next_offset = eob_offset;
			break;
		}
	}
	return linelen;
}

/*
 * Copied from the mgcp dissector. (This function should be moved to /epan )
 * tvb_skip_wsp - Returns the position in tvb of the first non-whitespace
 *				  character following offset or offset + maxlength -1 whichever
 *				  is smaller.
 *
 * Parameters:
 * tvb - The tvbuff in which we are skipping whitespace.
 * offset - The offset in tvb from which we begin trying to skip whitespace.
 * maxlength - The maximum distance from offset that we may try to skip
 * whitespace.
 *
 * Returns: The position in tvb of the first non-whitespace
 *			character following offset or offset + maxlength -1 whichever
 *			is smaller.
 */
gint tvb_skip_wsp(tvbuff_t* tvb, const gint offset, const gint maxlength)
{
	gint counter = offset;
	gint end, tvb_len;
	guint8 tempchar;

	/* Get the length remaining */
	tvb_len = tvb_length(tvb);
	end = offset + maxlength;
	if (end >= tvb_len)
	{
		end = tvb_len;
	}

	/* Skip past spaces, tabs, CRs and LFs until run out or meet something else */
	for (counter = offset;
		 counter < end &&
		  ((tempchar = tvb_get_guint8(tvb,counter)) == ' ' ||
		  tempchar == '\t' || tempchar == '\r' || tempchar == '\n');
		 counter++);

	return (counter);
}

gint tvb_skip_wsp_return(tvbuff_t* tvb, const gint offset){
	gint counter = offset;
	gint end;
	guint8 tempchar;
	end = 0;

	for(counter = offset; counter > end &&
		((tempchar = tvb_get_guint8(tvb,counter)) == ' ' ||
		tempchar == '\t' || tempchar == '\n' || tempchar == '\r'); counter--);
	counter++;
	return (counter);
}


/*
 * Format a bunch of data from a tvbuff as bytes, returning a pointer
 * to the string with the formatted data, with "punct" as a byte
 * separator.
 */
gchar *
tvb_bytes_to_str_punct(tvbuff_t *tvb, const gint offset, const gint len, const gchar punct)
{
	return bytes_to_str_punct(ensure_contiguous(tvb, offset, len), len, punct);
}


/*
 * Given a tvbuff, an offset into the tvbuff, and a length that starts
 * at that offset (which may be -1 for "all the way to the end of the
 * tvbuff"), fetch BCD encoded digits from a tvbuff starting from either
 * the low or high half byte, formating the digits according to an input digit set,
 * if NUll a default digit set of 0-9 returning "?" for overdecadic digits will be used.
 * A pointer to the EP allocated string will be returned.
 * Note a tvbuff content of 0xf is considered a 'filler' and will end the conversion.
 */
static dgt_set_t Dgt1_9_bcd = {
    {
  /*  0   1   2   3   4   5   6   7   8   9   a   b   c   d   e */
     '0','1','2','3','4','5','6','7','8','9','?','?','?','?','?'
    }
};
const gchar *
tvb_bcd_dig_to_ep_str(tvbuff_t *tvb, const gint offset, const gint len, dgt_set_t *dgt, gboolean skip_first)
{
	int length;
	guint8 octet;
	int i=0;
	char *digit_str;
	gint t_offset = offset;

	if (!dgt)
		dgt = &Dgt1_9_bcd;

	if( len == -1){
		length = tvb_length(tvb);
		if (length < offset){
			return "";
		}
	}else{
		length = offset + len;
	}
	digit_str = ep_alloc((length - offset)*2+1);

	while ( t_offset < length ){

		octet = tvb_get_guint8(tvb,t_offset);
		if (!skip_first){
			digit_str[i] = dgt->out[octet & 0x0f];
			i++;
		}
		skip_first = FALSE;

		/*
		 * unpack second value in byte
		 */
		octet = octet >> 4;

		if (octet == 0x0f)	/* odd number bytes - hit filler */
			break;

		digit_str[i] = dgt->out[octet & 0x0f];
		i++;
		t_offset++;

	}
	digit_str[i]= '\0';
	return digit_str;

}

/*
 * Format a bunch of data from a tvbuff as bytes, returning a pointer
 * to the string with the formatted data.
 */
gchar *
tvb_bytes_to_str(tvbuff_t *tvb, const gint offset, const gint len)
{
	return bytes_to_str(ensure_contiguous(tvb, offset, len), len);
}

/* Find a needle tvbuff within a haystack tvbuff. */
gint
tvb_find_tvb(tvbuff_t *haystack_tvb, tvbuff_t *needle_tvb, const gint haystack_offset)
{
	guint		haystack_abs_offset, haystack_abs_length;
	const guint8	*haystack_data;
	const guint8	*needle_data;
	const guint 	needle_len = needle_tvb->length;
	const guint8	*location;

	DISSECTOR_ASSERT(haystack_tvb && haystack_tvb->initialized);

	if (haystack_tvb->length < 1 || needle_tvb->length < 1) {
		return -1;
	}

	/* Get pointers to the tvbuffs' data. */
	haystack_data = ensure_contiguous(haystack_tvb, 0, -1);
	needle_data = ensure_contiguous(needle_tvb, 0, -1);

	check_offset_length(haystack_tvb->length, haystack_tvb->reported_length, haystack_offset, -1,
			&haystack_abs_offset, &haystack_abs_length);

	location = epan_memmem(haystack_data + haystack_abs_offset, haystack_abs_length,
			needle_data, needle_len);

	if (location) {
		return (gint) (location - haystack_data);
	}

	return -1;
}

#ifdef HAVE_LIBZ
/*
 * Uncompresses a zlib compressed packet inside a message of tvb at offset with
 * length comprlen.  Returns an uncompressed tvbuffer if uncompression
 * succeeded or NULL if uncompression failed.
 */
#define TVB_Z_MIN_BUFSIZ 32768
#define TVB_Z_MAX_BUFSIZ 1048576 * 10
/* #define TVB_Z_DEBUG 1 */
#undef TVB_Z_DEBUG

tvbuff_t *
tvb_uncompress(tvbuff_t *tvb, const int offset, int comprlen)
{
	gint err = Z_OK;
	guint bytes_out = 0;
	guint8 *compr = NULL;
	guint8 *uncompr = NULL;
	tvbuff_t *uncompr_tvb = NULL;
	z_streamp strm = NULL;
	Bytef *strmbuf = NULL;
	guint inits_done = 0;
	gint wbits = MAX_WBITS;
	guint8 *next = NULL;
	guint bufsiz = TVB_Z_MIN_BUFSIZ;
#ifdef TVB_Z_DEBUG
	guint inflate_passes = 0;
	guint bytes_in = tvb_length_remaining(tvb, offset);
#endif

	if (tvb == NULL) {
		return NULL;
	}

	compr = tvb_memdup(tvb, offset, comprlen);

	if (!compr)
		return NULL;

	/*
	 * Assume that the uncompressed data is at least twice as big as
	 * the compressed size.
	 */
	bufsiz = tvb_length_remaining(tvb, offset) * 2;
	bufsiz = CLAMP(bufsiz, TVB_Z_MIN_BUFSIZ, TVB_Z_MAX_BUFSIZ);

#ifdef TVB_Z_DEBUG
	printf("bufsiz: %u bytes\n", bufsiz);
#endif

	next = compr;

	strm = g_new0(z_stream, 1);
	strm->next_in = next;
	strm->avail_in = comprlen;

	strmbuf = g_malloc0(bufsiz);
	strm->next_out = strmbuf;
	strm->avail_out = bufsiz;

	err = inflateInit2(strm, wbits);
	inits_done = 1;
	if (err != Z_OK) {
		inflateEnd(strm);
		g_free(strm);
		g_free(compr);
		g_free(strmbuf);
		return NULL;
	}

	while (1) {
		memset(strmbuf, '\0', bufsiz);
		strm->next_out = strmbuf;
		strm->avail_out = bufsiz;

		err = inflate(strm, Z_SYNC_FLUSH);

		if (err == Z_OK || err == Z_STREAM_END) {
			guint bytes_pass = bufsiz - strm->avail_out;

#ifdef TVB_Z_DEBUG
			++inflate_passes;
#endif

			if (uncompr == NULL) {
				uncompr = g_memdup(strmbuf, bytes_pass);
			} else {
				guint8 *new_data = g_malloc0(bytes_out + bytes_pass);

				g_memmove(new_data, uncompr, bytes_out);
				g_memmove((new_data + bytes_out), strmbuf,
					bytes_pass);

				g_free(uncompr);
				uncompr = new_data;
			}

			bytes_out += bytes_pass;

			if ( err == Z_STREAM_END) {
				inflateEnd(strm);
				g_free(strm);
				g_free(strmbuf);
				break;
			}
		} else if (err == Z_BUF_ERROR) {
			/*
			 * It's possible that not enough frames were captured
			 * to decompress this fully, so return what we've done
			 * so far, if any.
			 */
			inflateEnd(strm);
			g_free(strm);
			g_free(strmbuf);

			if (uncompr != NULL) {
				break;
			} else {
				g_free(compr);
				return NULL;
			}

		} else if (err == Z_DATA_ERROR && inits_done == 1
			&& uncompr == NULL && (*compr  == 0x1f) &&
			(*(compr + 1) == 0x8b)) {
			/*
			 * inflate() is supposed to handle both gzip and deflate
			 * streams automatically, but in reality it doesn't
			 * seem to handle either (at least not within the
			 * context of an HTTP response.)  We have to try
			 * several tweaks, depending on the type of data and
			 * version of the library installed.
			 */

			/*
			 * Gzip file format.  Skip past the header, since the
			 * fix to make it work (setting windowBits to 31)
			 * doesn't work with all versions of the library.
			 */
			Bytef *c = compr + 2;
			Bytef flags = 0;

			if (*c == Z_DEFLATED) {
				c++;
			} else {
				inflateEnd(strm);
				g_free(strm);
				g_free(compr);
				g_free(strmbuf);
				return NULL;
			}

			flags = *c;

			/* Skip past the MTIME, XFL, and OS fields. */
			c += 7;

			if (flags & (1 << 2)) {
				/* An Extra field is present. */
				gint xsize = (gint)(*c |
					(*(c + 1) << 8));

				c += xsize;
			}

			if (flags & (1 << 3)) {
				/* A null terminated filename */

				while ((c - compr) < comprlen && *c != '\0') {
					c++;
				}

				c++;
			}

			if (flags & (1 << 4)) {
				/* A null terminated comment */

				while ((c - compr) < comprlen && *c != '\0') {
					c++;
				}

				c++;
			}


			inflateReset(strm);
			next = c;
			strm->next_in = next;
			if (c - compr > comprlen) {
				inflateEnd(strm);
				g_free(strm);
				g_free(compr);
				g_free(strmbuf);
				return NULL;
			}
			comprlen -= (int) (c - compr);

			inflateEnd(strm);
			err = inflateInit2(strm, wbits);
			inits_done++;
		} else if (err == Z_DATA_ERROR && uncompr == NULL &&
			inits_done <= 3) {

			/*
			 * Re-init the stream with a negative
			 * MAX_WBITS. This is necessary due to
			 * some servers (Apache) not sending
			 * the deflate header with the
			 * content-encoded response.
			 */
			wbits = -MAX_WBITS;

			inflateReset(strm);

			strm->next_in = next;
			strm->avail_in = comprlen;

			inflateEnd(strm);
			memset(strmbuf, '\0', bufsiz);
			strm->next_out = strmbuf;
			strm->avail_out = bufsiz;

			err = inflateInit2(strm, wbits);

			inits_done++;

			if (err != Z_OK) {
				g_free(strm);
				g_free(strmbuf);
				g_free(compr);
				g_free(uncompr);

				return NULL;
			}
		} else {
			inflateEnd(strm);
			g_free(strm);
			g_free(strmbuf);

			if (uncompr == NULL) {
				g_free(compr);
				return NULL;
			}

			break;
		}
	}

#ifdef TVB_Z_DEBUG
	printf("inflate() total passes: %u\n", inflate_passes);
	printf("bytes  in: %u\nbytes out: %u\n\n", bytes_in, bytes_out);
#endif

	if (uncompr != NULL) {
		uncompr_tvb =  tvb_new_real_data((guint8*) uncompr, bytes_out,
			bytes_out);
		tvb_set_free_cb(uncompr_tvb, g_free);
	}
	g_free(compr);
	return uncompr_tvb;
}
#else
tvbuff_t *
tvb_uncompress(tvbuff_t *tvb _U_, const int offset _U_, int comprlen _U_)
{
	return NULL;
}
#endif

tvbuff_t *
tvb_child_uncompress(tvbuff_t *parent, tvbuff_t *tvb, const int offset, int comprlen)
{
	tvbuff_t *new_tvb = tvb_uncompress(tvb, offset, comprlen);
	if (new_tvb)
		tvb_set_child_real_data_tvbuff (parent, new_tvb);
	return new_tvb;
}

gint
tvb_raw_offset(tvbuff_t *tvb)
{
	return ((tvb->raw_offset==-1)?(tvb->raw_offset = tvb_offset_from_real_beginning(tvb)):tvb->raw_offset);
}

struct tvbuff *
tvb_get_ds_tvb(tvbuff_t *tvb)
{
	return(tvb->ds_tvb);
}
/* tvbuff-int.h
 *
 * Structures that most TVB users should not be accessing directly.
 *
 * $Id: tvbuff-int.h 37422 2011-05-27 03:06:50Z morriss $
 *
 * Wireshark - Network traffic analyzer
 * By Gerald Combs <gerald@xxxxxxxxxxxxx>
 * Copyright 1998 Gerald Combs
 *
 * 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.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 */

#ifndef __TVBUFF_INT_H__
#define __TVBUFF_INT_H__

typedef struct {
	/** The backing tvbuff_t */
	struct tvbuff	*tvb;

	/** The offset of 'tvb' to which I'm privy */
	guint		offset;
	/** The length of 'tvb' to which I'm privy */
	guint		length;

} tvb_backing_t;

typedef struct {
	GSList		*tvbs;

	/* Used for quick testing to see if this
	 * is the tvbuff that a COMPOSITE is
	 * interested in. */
	guint		*start_offsets;
	guint		*end_offsets;

} tvb_comp_t;

struct tvbuff {
	/* Doubly linked list pointers */
	tvbuff_t                *next;
	tvbuff_t                *previous;

	/* Record-keeping */
	tvbuff_type		type;
	gboolean		initialized;
	struct tvbuff		*ds_tvb;  /**< data source top-level tvbuff */

	/** TVBUFF_SUBSET and TVBUFF_COMPOSITE keep track
	 * of the other tvbuff's they use */
	union {
		tvb_backing_t	subset;
		tvb_comp_t	composite;
	} tvbuffs;

	/** We're either a TVBUFF_REAL_DATA or a
	 * TVBUFF_SUBSET that has a backing buffer that
	 * has real_data != NULL, or a TVBUFF_COMPOSITE
	 * which has flattened its data due to a call
	 * to tvb_get_ptr().
	 */
	const guint8		*real_data;

	/** Length of virtual buffer (and/or real_data). */
	guint			length;

	/** Reported length. */
	guint			reported_length;

	/* Offset from beginning of first TVBUFF_REAL. */
	gint			raw_offset;

	/** Func to call when actually freed */
	tvbuff_free_cb_t	free_cb;
};

#endif