mirror of
https://github.com/AsahiLinux/u-boot
synced 2024-12-24 03:53:31 +00:00
b606a6cfa6
GCC recognizes /* fallthrough */ if -Wimplicit-fallthrough=3 is enabled. Let's use it consistently. Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
1242 lines
31 KiB
C
1242 lines
31 KiB
C
// SPDX-License-Identifier: GPL-2.0+
|
|
/*
|
|
* Copyright (c) 2013, Google Inc.
|
|
* Written by Simon Glass <sjg@chromium.org>
|
|
*
|
|
* Perform a grep of an FDT either displaying the source subset or producing
|
|
* a new .dtb subset which can be used as required.
|
|
*/
|
|
|
|
#include <assert.h>
|
|
#include <ctype.h>
|
|
#include <errno.h>
|
|
#include <getopt.h>
|
|
#include <fcntl.h>
|
|
#include <stdbool.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <unistd.h>
|
|
#include <fdt_region.h>
|
|
|
|
#include "fdt_host.h"
|
|
#include "libfdt_internal.h"
|
|
|
|
/* Define DEBUG to get some debugging output on stderr */
|
|
#ifdef DEBUG
|
|
#define debug(a, b...) fprintf(stderr, a, ## b)
|
|
#else
|
|
#define debug(a, b...)
|
|
#endif
|
|
|
|
/* A linked list of values we are grepping for */
|
|
struct value_node {
|
|
int type; /* Types this value matches (FDT_IS... mask) */
|
|
int include; /* 1 to include matches, 0 to exclude */
|
|
const char *string; /* String to match */
|
|
struct value_node *next; /* Pointer to next node, or NULL */
|
|
};
|
|
|
|
/* Output formats we support */
|
|
enum output_t {
|
|
OUT_DTS, /* Device tree source */
|
|
OUT_DTB, /* Valid device tree binary */
|
|
OUT_BIN, /* Fragment of .dtb, for hashing */
|
|
};
|
|
|
|
/* Holds information which controls our output and options */
|
|
struct display_info {
|
|
enum output_t output; /* Output format */
|
|
int add_aliases; /* Add aliases node to output */
|
|
int all; /* Display all properties/nodes */
|
|
int colour; /* Display output in ANSI colour */
|
|
int region_list; /* Output a region list */
|
|
int flags; /* Flags (FDT_REG_...) */
|
|
int list_strings; /* List strings in string table */
|
|
int show_offset; /* Show offset */
|
|
int show_addr; /* Show address */
|
|
int header; /* Output an FDT header */
|
|
int diff; /* Show +/- diff markers */
|
|
int include_root; /* Include the root node and all properties */
|
|
int remove_strings; /* Remove unused strings */
|
|
int show_dts_version; /* Put '/dts-v1/;' on the first line */
|
|
int types_inc; /* Mask of types that we include (FDT_IS...) */
|
|
int types_exc; /* Mask of types that we exclude (FDT_IS...) */
|
|
int invert; /* Invert polarity of match */
|
|
struct value_node *value_head; /* List of values to match */
|
|
const char *output_fname; /* Output filename */
|
|
FILE *fout; /* File to write dts/dtb output */
|
|
};
|
|
|
|
static void report_error(const char *where, int err)
|
|
{
|
|
fprintf(stderr, "Error at '%s': %s\n", where, fdt_strerror(err));
|
|
}
|
|
|
|
/* Supported ANSI colours */
|
|
enum {
|
|
COL_BLACK,
|
|
COL_RED,
|
|
COL_GREEN,
|
|
COL_YELLOW,
|
|
COL_BLUE,
|
|
COL_MAGENTA,
|
|
COL_CYAN,
|
|
COL_WHITE,
|
|
|
|
COL_NONE = -1,
|
|
};
|
|
|
|
/**
|
|
* print_ansi_colour() - Print out the ANSI sequence for a colour
|
|
*
|
|
* @fout: Output file
|
|
* @col: Colour to output (COL_...), or COL_NONE to reset colour
|
|
*/
|
|
static void print_ansi_colour(FILE *fout, int col)
|
|
{
|
|
if (col == COL_NONE)
|
|
fprintf(fout, "\033[0m");
|
|
else
|
|
fprintf(fout, "\033[1;%dm", col + 30);
|
|
}
|
|
|
|
|
|
/**
|
|
* value_add() - Add a new value to our list of things to grep for
|
|
*
|
|
* @disp: Display structure, holding info about our options
|
|
* @headp: Pointer to header pointer of list
|
|
* @type: Type of this value (FDT_IS_...)
|
|
* @include: 1 if we want to include matches, 0 to exclude
|
|
* @str: String value to match
|
|
*/
|
|
static int value_add(struct display_info *disp, struct value_node **headp,
|
|
int type, int include, const char *str)
|
|
{
|
|
struct value_node *node;
|
|
|
|
/*
|
|
* Keep track of which types we are excluding/including. We don't
|
|
* allow both including and excluding things, because it doesn't make
|
|
* sense. 'Including' means that everything not mentioned is
|
|
* excluded. 'Excluding' means that everything not mentioned is
|
|
* included. So using the two together would be meaningless.
|
|
*/
|
|
if (include)
|
|
disp->types_inc |= type;
|
|
else
|
|
disp->types_exc |= type;
|
|
if (disp->types_inc & disp->types_exc & type) {
|
|
fprintf(stderr,
|
|
"Cannot use both include and exclude for '%s'\n", str);
|
|
return -1;
|
|
}
|
|
|
|
str = strdup(str);
|
|
if (!str)
|
|
goto err_mem;
|
|
node = malloc(sizeof(*node));
|
|
if (!node)
|
|
goto err_mem;
|
|
node->next = *headp;
|
|
node->type = type;
|
|
node->include = include;
|
|
node->string = str;
|
|
*headp = node;
|
|
|
|
return 0;
|
|
err_mem:
|
|
fprintf(stderr, "Out of memory\n");
|
|
return -1;
|
|
}
|
|
|
|
static bool util_is_printable_string(const void *data, int len)
|
|
{
|
|
const char *s = data;
|
|
const char *ss, *se;
|
|
|
|
/* zero length is not */
|
|
if (len == 0)
|
|
return 0;
|
|
|
|
/* must terminate with zero */
|
|
if (s[len - 1] != '\0')
|
|
return 0;
|
|
|
|
se = s + len;
|
|
|
|
while (s < se) {
|
|
ss = s;
|
|
while (s < se && *s && isprint((unsigned char)*s))
|
|
s++;
|
|
|
|
/* not zero, or not done yet */
|
|
if (*s != '\0' || s == ss)
|
|
return 0;
|
|
|
|
s++;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
static void utilfdt_print_data(const char *data, int len)
|
|
{
|
|
int i;
|
|
const char *p = data;
|
|
const char *s;
|
|
|
|
/* no data, don't print */
|
|
if (len == 0)
|
|
return;
|
|
|
|
if (util_is_printable_string(data, len)) {
|
|
printf(" = ");
|
|
|
|
s = data;
|
|
do {
|
|
printf("\"%s\"", s);
|
|
s += strlen(s) + 1;
|
|
if (s < data + len)
|
|
printf(", ");
|
|
} while (s < data + len);
|
|
|
|
} else if ((len % 4) == 0) {
|
|
const uint32_t *cell = (const uint32_t *)data;
|
|
|
|
printf(" = <");
|
|
for (i = 0, len /= 4; i < len; i++)
|
|
printf("0x%08x%s", fdt32_to_cpu(cell[i]),
|
|
i < (len - 1) ? " " : "");
|
|
printf(">");
|
|
} else {
|
|
printf(" = [");
|
|
for (i = 0; i < len; i++)
|
|
printf("%02x%s", *p++, i < len - 1 ? " " : "");
|
|
printf("]");
|
|
}
|
|
}
|
|
|
|
/**
|
|
* display_fdt_by_regions() - Display regions of an FDT source
|
|
*
|
|
* This dumps an FDT as source, but only certain regions of it. This is the
|
|
* final stage of the grep - we have a list of regions we want to display,
|
|
* and this function displays them.
|
|
*
|
|
* @disp: Display structure, holding info about our options
|
|
* @blob: FDT blob to display
|
|
* @region: List of regions to display
|
|
* @count: Number of regions
|
|
*/
|
|
static int display_fdt_by_regions(struct display_info *disp, const void *blob,
|
|
struct fdt_region region[], int count)
|
|
{
|
|
struct fdt_region *reg = region, *reg_end = region + count;
|
|
uint32_t off_mem_rsvmap = fdt_off_mem_rsvmap(blob);
|
|
int base = fdt_off_dt_struct(blob);
|
|
int version = fdt_version(blob);
|
|
int offset, nextoffset;
|
|
int tag, depth, shift;
|
|
FILE *f = disp->fout;
|
|
uint64_t addr, size;
|
|
int in_region;
|
|
int file_ofs;
|
|
int i;
|
|
|
|
if (disp->show_dts_version)
|
|
fprintf(f, "/dts-v1/;\n");
|
|
|
|
if (disp->header) {
|
|
fprintf(f, "// magic:\t\t0x%x\n", fdt_magic(blob));
|
|
fprintf(f, "// totalsize:\t\t0x%x (%d)\n", fdt_totalsize(blob),
|
|
fdt_totalsize(blob));
|
|
fprintf(f, "// off_dt_struct:\t0x%x\n",
|
|
fdt_off_dt_struct(blob));
|
|
fprintf(f, "// off_dt_strings:\t0x%x\n",
|
|
fdt_off_dt_strings(blob));
|
|
fprintf(f, "// off_mem_rsvmap:\t0x%x\n", off_mem_rsvmap);
|
|
fprintf(f, "// version:\t\t%d\n", version);
|
|
fprintf(f, "// last_comp_version:\t%d\n",
|
|
fdt_last_comp_version(blob));
|
|
if (version >= 2) {
|
|
fprintf(f, "// boot_cpuid_phys:\t0x%x\n",
|
|
fdt_boot_cpuid_phys(blob));
|
|
}
|
|
if (version >= 3) {
|
|
fprintf(f, "// size_dt_strings:\t0x%x\n",
|
|
fdt_size_dt_strings(blob));
|
|
}
|
|
if (version >= 17) {
|
|
fprintf(f, "// size_dt_struct:\t0x%x\n",
|
|
fdt_size_dt_struct(blob));
|
|
}
|
|
fprintf(f, "\n");
|
|
}
|
|
|
|
if (disp->flags & FDT_REG_ADD_MEM_RSVMAP) {
|
|
const struct fdt_reserve_entry *p_rsvmap;
|
|
|
|
p_rsvmap = (const struct fdt_reserve_entry *)
|
|
((const char *)blob + off_mem_rsvmap);
|
|
for (i = 0; ; i++) {
|
|
addr = fdt64_to_cpu(p_rsvmap[i].address);
|
|
size = fdt64_to_cpu(p_rsvmap[i].size);
|
|
if (addr == 0 && size == 0)
|
|
break;
|
|
|
|
fprintf(f, "/memreserve/ %llx %llx;\n",
|
|
(unsigned long long)addr,
|
|
(unsigned long long)size);
|
|
}
|
|
}
|
|
|
|
depth = 0;
|
|
nextoffset = 0;
|
|
shift = 4; /* 4 spaces per indent */
|
|
do {
|
|
const struct fdt_property *prop;
|
|
const char *name;
|
|
int show;
|
|
int len;
|
|
|
|
offset = nextoffset;
|
|
|
|
/*
|
|
* Work out the file offset of this offset, and decide
|
|
* whether it is in the region list or not
|
|
*/
|
|
file_ofs = base + offset;
|
|
if (reg < reg_end && file_ofs >= reg->offset + reg->size)
|
|
reg++;
|
|
in_region = reg < reg_end && file_ofs >= reg->offset &&
|
|
file_ofs < reg->offset + reg->size;
|
|
tag = fdt_next_tag(blob, offset, &nextoffset);
|
|
|
|
if (tag == FDT_END)
|
|
break;
|
|
show = in_region || disp->all;
|
|
if (show && disp->diff)
|
|
fprintf(f, "%c", in_region ? '+' : '-');
|
|
|
|
if (!show) {
|
|
/* Do this here to avoid 'if (show)' in every 'case' */
|
|
if (tag == FDT_BEGIN_NODE)
|
|
depth++;
|
|
else if (tag == FDT_END_NODE)
|
|
depth--;
|
|
continue;
|
|
}
|
|
if (tag != FDT_END) {
|
|
if (disp->show_addr)
|
|
fprintf(f, "%4x: ", file_ofs);
|
|
if (disp->show_offset)
|
|
fprintf(f, "%4x: ", file_ofs - base);
|
|
}
|
|
|
|
/* Green means included, red means excluded */
|
|
if (disp->colour)
|
|
print_ansi_colour(f, in_region ? COL_GREEN : COL_RED);
|
|
|
|
switch (tag) {
|
|
case FDT_PROP:
|
|
prop = fdt_get_property_by_offset(blob, offset, NULL);
|
|
name = fdt_string(blob, fdt32_to_cpu(prop->nameoff));
|
|
fprintf(f, "%*s%s", depth * shift, "", name);
|
|
utilfdt_print_data(prop->data,
|
|
fdt32_to_cpu(prop->len));
|
|
fprintf(f, ";");
|
|
break;
|
|
|
|
case FDT_NOP:
|
|
fprintf(f, "%*s// [NOP]", depth * shift, "");
|
|
break;
|
|
|
|
case FDT_BEGIN_NODE:
|
|
name = fdt_get_name(blob, offset, &len);
|
|
fprintf(f, "%*s%s {", depth++ * shift, "",
|
|
*name ? name : "/");
|
|
break;
|
|
|
|
case FDT_END_NODE:
|
|
fprintf(f, "%*s};", --depth * shift, "");
|
|
break;
|
|
}
|
|
|
|
/* Reset colour back to normal before end of line */
|
|
if (disp->colour)
|
|
print_ansi_colour(f, COL_NONE);
|
|
fprintf(f, "\n");
|
|
} while (1);
|
|
|
|
/* Print a list of strings if requested */
|
|
if (disp->list_strings) {
|
|
const char *str;
|
|
int str_base = fdt_off_dt_strings(blob);
|
|
|
|
for (offset = 0; offset < fdt_size_dt_strings(blob);
|
|
offset += strlen(str) + 1) {
|
|
str = fdt_string(blob, offset);
|
|
int len = strlen(str) + 1;
|
|
int show;
|
|
|
|
/* Only print strings that are in the region */
|
|
file_ofs = str_base + offset;
|
|
in_region = reg < reg_end &&
|
|
file_ofs >= reg->offset &&
|
|
file_ofs + len < reg->offset +
|
|
reg->size;
|
|
show = in_region || disp->all;
|
|
if (show && disp->diff)
|
|
printf("%c", in_region ? '+' : '-');
|
|
if (disp->show_addr)
|
|
printf("%4x: ", file_ofs);
|
|
if (disp->show_offset)
|
|
printf("%4x: ", offset);
|
|
printf("%s\n", str);
|
|
}
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
/**
|
|
* dump_fdt_regions() - Dump regions of an FDT as binary data
|
|
*
|
|
* This dumps an FDT as binary, but only certain regions of it. This is the
|
|
* final stage of the grep - we have a list of regions we want to dump,
|
|
* and this function dumps them.
|
|
*
|
|
* The output of this function may or may not be a valid FDT. To ensure it
|
|
* is, these disp->flags must be set:
|
|
*
|
|
* FDT_REG_SUPERNODES: ensures that subnodes are preceded by their
|
|
* parents. Without this option, fragments of subnode data may be
|
|
* output without the supernodes above them. This is useful for
|
|
* hashing but cannot produce a valid FDT.
|
|
* FDT_REG_ADD_STRING_TAB: Adds a string table to the end of the FDT.
|
|
* Without this none of the properties will have names
|
|
* FDT_REG_ADD_MEM_RSVMAP: Adds a mem_rsvmap table - an FDT is invalid
|
|
* without this.
|
|
*
|
|
* @disp: Display structure, holding info about our options
|
|
* @blob: FDT blob to display
|
|
* @region: List of regions to display
|
|
* @count: Number of regions
|
|
* @out: Output destination
|
|
*/
|
|
static int dump_fdt_regions(struct display_info *disp, const void *blob,
|
|
struct fdt_region region[], int count, char *out)
|
|
{
|
|
struct fdt_header *fdt;
|
|
int size, struct_start;
|
|
int ptr;
|
|
int i;
|
|
|
|
/* Set up a basic header (even if we don't actually write it) */
|
|
fdt = (struct fdt_header *)out;
|
|
memset(fdt, '\0', sizeof(*fdt));
|
|
fdt_set_magic(fdt, FDT_MAGIC);
|
|
struct_start = FDT_ALIGN(sizeof(struct fdt_header),
|
|
sizeof(struct fdt_reserve_entry));
|
|
fdt_set_off_mem_rsvmap(fdt, struct_start);
|
|
fdt_set_version(fdt, FDT_LAST_SUPPORTED_VERSION);
|
|
fdt_set_last_comp_version(fdt, FDT_FIRST_SUPPORTED_VERSION);
|
|
|
|
/*
|
|
* Calculate the total size of the regions we are writing out. The
|
|
* first will be the mem_rsvmap if the FDT_REG_ADD_MEM_RSVMAP flag
|
|
* is set. The last will be the string table if FDT_REG_ADD_STRING_TAB
|
|
* is set.
|
|
*/
|
|
for (i = size = 0; i < count; i++)
|
|
size += region[i].size;
|
|
|
|
/* Bring in the mem_rsvmap section from the old file if requested */
|
|
if (count > 0 && (disp->flags & FDT_REG_ADD_MEM_RSVMAP)) {
|
|
struct_start += region[0].size;
|
|
size -= region[0].size;
|
|
}
|
|
fdt_set_off_dt_struct(fdt, struct_start);
|
|
|
|
/* Update the header to have the correct offsets/sizes */
|
|
if (count >= 2 && (disp->flags & FDT_REG_ADD_STRING_TAB)) {
|
|
int str_size;
|
|
|
|
str_size = region[count - 1].size;
|
|
fdt_set_size_dt_struct(fdt, size - str_size);
|
|
fdt_set_off_dt_strings(fdt, struct_start + size - str_size);
|
|
fdt_set_size_dt_strings(fdt, str_size);
|
|
fdt_set_totalsize(fdt, struct_start + size);
|
|
}
|
|
|
|
/* Write the header if required */
|
|
ptr = 0;
|
|
if (disp->header) {
|
|
ptr = sizeof(*fdt);
|
|
while (ptr < fdt_off_mem_rsvmap(fdt))
|
|
out[ptr++] = '\0';
|
|
}
|
|
|
|
/* Output all the nodes including any mem_rsvmap/string table */
|
|
for (i = 0; i < count; i++) {
|
|
struct fdt_region *reg = ®ion[i];
|
|
|
|
memcpy(out + ptr, (const char *)blob + reg->offset, reg->size);
|
|
ptr += reg->size;
|
|
}
|
|
|
|
return ptr;
|
|
}
|
|
|
|
/**
|
|
* show_region_list() - Print out a list of regions
|
|
*
|
|
* The list includes the region offset (absolute offset from start of FDT
|
|
* blob in bytes) and size
|
|
*
|
|
* @reg: List of regions to print
|
|
* @count: Number of regions
|
|
*/
|
|
static void show_region_list(struct fdt_region *reg, int count)
|
|
{
|
|
int i;
|
|
|
|
printf("Regions: %d\n", count);
|
|
for (i = 0; i < count; i++, reg++) {
|
|
printf("%d: %-10x %-10x\n", i, reg->offset,
|
|
reg->offset + reg->size);
|
|
}
|
|
}
|
|
|
|
static int check_type_include(void *priv, int type, const char *data, int size)
|
|
{
|
|
struct display_info *disp = priv;
|
|
struct value_node *val;
|
|
int match, none_match = FDT_IS_ANY;
|
|
|
|
/* If none of our conditions mention this type, we know nothing */
|
|
debug("type=%x, data=%s\n", type, data ? data : "(null)");
|
|
if (!((disp->types_inc | disp->types_exc) & type)) {
|
|
debug(" - not in any condition\n");
|
|
return -1;
|
|
}
|
|
|
|
/*
|
|
* Go through the list of conditions. For inclusive conditions, we
|
|
* return 1 at the first match. For exclusive conditions, we must
|
|
* check that there are no matches.
|
|
*/
|
|
if (data) {
|
|
for (val = disp->value_head; val; val = val->next) {
|
|
if (!(type & val->type))
|
|
continue;
|
|
match = fdt_stringlist_contains(data, size,
|
|
val->string);
|
|
debug(" - val->type=%x, str='%s', match=%d\n",
|
|
val->type, val->string, match);
|
|
if (match && val->include) {
|
|
debug(" - match inc %s\n", val->string);
|
|
return 1;
|
|
}
|
|
if (match)
|
|
none_match &= ~val->type;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* If this is an exclusive condition, and nothing matches, then we
|
|
* should return 1.
|
|
*/
|
|
if ((type & disp->types_exc) && (none_match & type)) {
|
|
debug(" - match exc\n");
|
|
/*
|
|
* Allow FDT_IS_COMPAT to make the final decision in the
|
|
* case where there is no specific type
|
|
*/
|
|
if (type == FDT_IS_NODE && disp->types_exc == FDT_ANY_GLOBAL) {
|
|
debug(" - supressed exc node\n");
|
|
return -1;
|
|
}
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* Allow FDT_IS_COMPAT to make the final decision in the
|
|
* case where there is no specific type (inclusive)
|
|
*/
|
|
if (type == FDT_IS_NODE && disp->types_inc == FDT_ANY_GLOBAL)
|
|
return -1;
|
|
|
|
debug(" - no match, types_inc=%x, types_exc=%x, none_match=%x\n",
|
|
disp->types_inc, disp->types_exc, none_match);
|
|
|
|
return 0;
|
|
}
|
|
|
|
/**
|
|
* h_include() - Include handler function for fdt_find_regions()
|
|
*
|
|
* This function decides whether to include or exclude a node, property or
|
|
* compatible string. The function is defined by fdt_find_regions().
|
|
*
|
|
* The algorithm is documented in the code - disp->invert is 0 for normal
|
|
* operation, and 1 to invert the sense of all matches.
|
|
*
|
|
* See
|
|
*/
|
|
static int h_include(void *priv, const void *fdt, int offset, int type,
|
|
const char *data, int size)
|
|
{
|
|
struct display_info *disp = priv;
|
|
int inc, len;
|
|
|
|
inc = check_type_include(priv, type, data, size);
|
|
if (disp->include_root && type == FDT_IS_PROP && offset == 0 && inc)
|
|
return 1;
|
|
|
|
/*
|
|
* If the node name does not tell us anything, check the
|
|
* compatible string
|
|
*/
|
|
if (inc == -1 && type == FDT_IS_NODE) {
|
|
debug(" - checking compatible2\n");
|
|
data = fdt_getprop(fdt, offset, "compatible", &len);
|
|
inc = check_type_include(priv, FDT_IS_COMPAT, data, len);
|
|
}
|
|
|
|
/* If we still have no idea, check for properties in the node */
|
|
if (inc != 1 && type == FDT_IS_NODE &&
|
|
(disp->types_inc & FDT_NODE_HAS_PROP)) {
|
|
debug(" - checking node '%s'\n",
|
|
fdt_get_name(fdt, offset, NULL));
|
|
for (offset = fdt_first_property_offset(fdt, offset);
|
|
offset > 0 && inc != 1;
|
|
offset = fdt_next_property_offset(fdt, offset)) {
|
|
const struct fdt_property *prop;
|
|
const char *str;
|
|
|
|
prop = fdt_get_property_by_offset(fdt, offset, NULL);
|
|
if (!prop)
|
|
continue;
|
|
str = fdt_string(fdt, fdt32_to_cpu(prop->nameoff));
|
|
inc = check_type_include(priv, FDT_NODE_HAS_PROP, str,
|
|
strlen(str));
|
|
}
|
|
if (inc == -1)
|
|
inc = 0;
|
|
}
|
|
|
|
switch (inc) {
|
|
case 1:
|
|
inc = !disp->invert;
|
|
break;
|
|
case 0:
|
|
inc = disp->invert;
|
|
break;
|
|
}
|
|
debug(" - returning %d\n", inc);
|
|
|
|
return inc;
|
|
}
|
|
|
|
static int h_cmp_region(const void *v1, const void *v2)
|
|
{
|
|
const struct fdt_region *region1 = v1, *region2 = v2;
|
|
|
|
return region1->offset - region2->offset;
|
|
}
|
|
|
|
static int fdtgrep_find_regions(const void *fdt,
|
|
int (*include_func)(void *priv, const void *fdt, int offset,
|
|
int type, const char *data, int size),
|
|
struct display_info *disp, struct fdt_region *region,
|
|
int max_regions, char *path, int path_len, int flags)
|
|
{
|
|
struct fdt_region_state state;
|
|
int count;
|
|
int ret;
|
|
|
|
count = 0;
|
|
ret = fdt_first_region(fdt, include_func, disp,
|
|
®ion[count++], path, path_len,
|
|
disp->flags, &state);
|
|
while (ret == 0) {
|
|
ret = fdt_next_region(fdt, include_func, disp,
|
|
count < max_regions ? ®ion[count] : NULL,
|
|
path, path_len, disp->flags, &state);
|
|
if (!ret)
|
|
count++;
|
|
}
|
|
if (ret && ret != -FDT_ERR_NOTFOUND)
|
|
return ret;
|
|
|
|
/* Find all the aliases and add those regions back in */
|
|
if (disp->add_aliases && count < max_regions) {
|
|
int new_count;
|
|
|
|
new_count = fdt_add_alias_regions(fdt, region, count,
|
|
max_regions, &state);
|
|
if (new_count == -FDT_ERR_NOTFOUND) {
|
|
/* No alias node found */
|
|
} else if (new_count < 0) {
|
|
return new_count;
|
|
} else if (new_count <= max_regions) {
|
|
/*
|
|
* The alias regions will now be at the end of the list.
|
|
* Sort the regions by offset to get things into the
|
|
* right order
|
|
*/
|
|
count = new_count;
|
|
qsort(region, count, sizeof(struct fdt_region),
|
|
h_cmp_region);
|
|
}
|
|
}
|
|
|
|
return count;
|
|
}
|
|
|
|
int utilfdt_read_err_len(const char *filename, char **buffp, off_t *len)
|
|
{
|
|
int fd = 0; /* assume stdin */
|
|
char *buf = NULL;
|
|
off_t bufsize = 1024, offset = 0;
|
|
int ret = 0;
|
|
|
|
*buffp = NULL;
|
|
if (strcmp(filename, "-") != 0) {
|
|
fd = open(filename, O_RDONLY);
|
|
if (fd < 0)
|
|
return errno;
|
|
}
|
|
|
|
/* Loop until we have read everything */
|
|
buf = malloc(bufsize);
|
|
if (!buf)
|
|
return -ENOMEM;
|
|
do {
|
|
/* Expand the buffer to hold the next chunk */
|
|
if (offset == bufsize) {
|
|
bufsize *= 2;
|
|
buf = realloc(buf, bufsize);
|
|
if (!buf)
|
|
return -ENOMEM;
|
|
}
|
|
|
|
ret = read(fd, &buf[offset], bufsize - offset);
|
|
if (ret < 0) {
|
|
ret = errno;
|
|
break;
|
|
}
|
|
offset += ret;
|
|
} while (ret != 0);
|
|
|
|
/* Clean up, including closing stdin; return errno on error */
|
|
close(fd);
|
|
if (ret)
|
|
free(buf);
|
|
else
|
|
*buffp = buf;
|
|
*len = bufsize;
|
|
return ret;
|
|
}
|
|
|
|
int utilfdt_read_err(const char *filename, char **buffp)
|
|
{
|
|
off_t len;
|
|
return utilfdt_read_err_len(filename, buffp, &len);
|
|
}
|
|
|
|
char *utilfdt_read_len(const char *filename, off_t *len)
|
|
{
|
|
char *buff;
|
|
int ret = utilfdt_read_err_len(filename, &buff, len);
|
|
|
|
if (ret) {
|
|
fprintf(stderr, "Couldn't open blob from '%s': %s\n", filename,
|
|
strerror(ret));
|
|
return NULL;
|
|
}
|
|
/* Successful read */
|
|
return buff;
|
|
}
|
|
|
|
char *utilfdt_read(const char *filename)
|
|
{
|
|
off_t len;
|
|
return utilfdt_read_len(filename, &len);
|
|
}
|
|
|
|
/**
|
|
* Run the main fdtgrep operation, given a filename and valid arguments
|
|
*
|
|
* @param disp Display information / options
|
|
* @param filename Filename of blob file
|
|
* @param return 0 if ok, -ve on error
|
|
*/
|
|
static int do_fdtgrep(struct display_info *disp, const char *filename)
|
|
{
|
|
struct fdt_region *region = NULL;
|
|
int max_regions;
|
|
int count = 100;
|
|
char path[1024];
|
|
char *blob;
|
|
int i, ret;
|
|
|
|
blob = utilfdt_read(filename);
|
|
if (!blob)
|
|
return -1;
|
|
ret = fdt_check_header(blob);
|
|
if (ret) {
|
|
fprintf(stderr, "Error: %s\n", fdt_strerror(ret));
|
|
return ret;
|
|
}
|
|
|
|
/* Allow old files, but they are untested */
|
|
if (fdt_version(blob) < 17 && disp->value_head) {
|
|
fprintf(stderr,
|
|
"Warning: fdtgrep does not fully support version %d files\n",
|
|
fdt_version(blob));
|
|
}
|
|
|
|
/*
|
|
* We do two passes, since we don't know how many regions we need.
|
|
* The first pass will count the regions, but if it is too many,
|
|
* we do another pass to actually record them.
|
|
*/
|
|
for (i = 0; i < 2; i++) {
|
|
region = realloc(region, count * sizeof(struct fdt_region));
|
|
if (!region) {
|
|
fprintf(stderr, "Out of memory for %d regions\n",
|
|
count);
|
|
return -1;
|
|
}
|
|
max_regions = count;
|
|
count = fdtgrep_find_regions(blob,
|
|
h_include, disp,
|
|
region, max_regions, path, sizeof(path),
|
|
disp->flags);
|
|
if (count < 0) {
|
|
report_error("fdt_find_regions", count);
|
|
free(region);
|
|
return -1;
|
|
}
|
|
if (count <= max_regions)
|
|
break;
|
|
}
|
|
if (count > max_regions) {
|
|
free(region);
|
|
fprintf(stderr, "Internal error with fdtgrep_find_region()\n");
|
|
return -1;
|
|
}
|
|
|
|
/* Optionally print a list of regions */
|
|
if (disp->region_list)
|
|
show_region_list(region, count);
|
|
|
|
/* Output either source .dts or binary .dtb */
|
|
if (disp->output == OUT_DTS) {
|
|
ret = display_fdt_by_regions(disp, blob, region, count);
|
|
} else {
|
|
void *fdt;
|
|
/* Allow reserved memory section to expand slightly */
|
|
int size = fdt_totalsize(blob) + 16;
|
|
|
|
fdt = malloc(size);
|
|
if (!fdt) {
|
|
fprintf(stderr, "Out_of_memory\n");
|
|
ret = -1;
|
|
goto err;
|
|
}
|
|
size = dump_fdt_regions(disp, blob, region, count, fdt);
|
|
if (disp->remove_strings) {
|
|
void *out;
|
|
|
|
out = malloc(size);
|
|
if (!out) {
|
|
fprintf(stderr, "Out_of_memory\n");
|
|
ret = -1;
|
|
goto err;
|
|
}
|
|
ret = fdt_remove_unused_strings(fdt, out);
|
|
if (ret < 0) {
|
|
fprintf(stderr,
|
|
"Failed to remove unused strings: err=%d\n",
|
|
ret);
|
|
goto err;
|
|
}
|
|
free(fdt);
|
|
fdt = out;
|
|
ret = fdt_pack(fdt);
|
|
if (ret < 0) {
|
|
fprintf(stderr, "Failed to pack: err=%d\n",
|
|
ret);
|
|
goto err;
|
|
}
|
|
size = fdt_totalsize(fdt);
|
|
}
|
|
|
|
if (size != fwrite(fdt, 1, size, disp->fout)) {
|
|
fprintf(stderr, "Write failure, %d bytes\n", size);
|
|
free(fdt);
|
|
ret = 1;
|
|
goto err;
|
|
}
|
|
free(fdt);
|
|
}
|
|
err:
|
|
free(blob);
|
|
free(region);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static const char usage_synopsis[] =
|
|
"fdtgrep - extract portions from device tree\n"
|
|
"\n"
|
|
"Usage:\n"
|
|
" fdtgrep <options> <dt file>|-\n\n"
|
|
"Output formats are:\n"
|
|
"\tdts - device tree soure text\n"
|
|
"\tdtb - device tree blob (sets -Hmt automatically)\n"
|
|
"\tbin - device tree fragment (may not be a valid .dtb)";
|
|
|
|
/* Helper for usage_short_opts string constant */
|
|
#define USAGE_COMMON_SHORT_OPTS "hV"
|
|
|
|
/* Helper for aligning long_opts array */
|
|
#define a_argument required_argument
|
|
|
|
/* Helper for usage_long_opts option array */
|
|
#define USAGE_COMMON_LONG_OPTS \
|
|
{"help", no_argument, NULL, 'h'}, \
|
|
{"version", no_argument, NULL, 'V'}, \
|
|
{NULL, no_argument, NULL, 0x0}
|
|
|
|
/* Helper for usage_opts_help array */
|
|
#define USAGE_COMMON_OPTS_HELP \
|
|
"Print this help and exit", \
|
|
"Print version and exit", \
|
|
NULL
|
|
|
|
/* Helper for getopt case statements */
|
|
#define case_USAGE_COMMON_FLAGS \
|
|
case 'h': usage(NULL); \
|
|
/* fallthrough */ \
|
|
case 'V': util_version(); \
|
|
/* fallthrough */ \
|
|
case '?': usage("unknown option");
|
|
|
|
static const char usage_short_opts[] =
|
|
"haAc:b:C:defg:G:HIlLmn:N:o:O:p:P:rRsStTv"
|
|
USAGE_COMMON_SHORT_OPTS;
|
|
static struct option const usage_long_opts[] = {
|
|
{"show-address", no_argument, NULL, 'a'},
|
|
{"colour", no_argument, NULL, 'A'},
|
|
{"include-node-with-prop", a_argument, NULL, 'b'},
|
|
{"include-compat", a_argument, NULL, 'c'},
|
|
{"exclude-compat", a_argument, NULL, 'C'},
|
|
{"diff", no_argument, NULL, 'd'},
|
|
{"enter-node", no_argument, NULL, 'e'},
|
|
{"show-offset", no_argument, NULL, 'f'},
|
|
{"include-match", a_argument, NULL, 'g'},
|
|
{"exclude-match", a_argument, NULL, 'G'},
|
|
{"show-header", no_argument, NULL, 'H'},
|
|
{"show-version", no_argument, NULL, 'I'},
|
|
{"list-regions", no_argument, NULL, 'l'},
|
|
{"list-strings", no_argument, NULL, 'L'},
|
|
{"include-mem", no_argument, NULL, 'm'},
|
|
{"include-node", a_argument, NULL, 'n'},
|
|
{"exclude-node", a_argument, NULL, 'N'},
|
|
{"include-prop", a_argument, NULL, 'p'},
|
|
{"exclude-prop", a_argument, NULL, 'P'},
|
|
{"remove-strings", no_argument, NULL, 'r'},
|
|
{"include-root", no_argument, NULL, 'R'},
|
|
{"show-subnodes", no_argument, NULL, 's'},
|
|
{"skip-supernodes", no_argument, NULL, 'S'},
|
|
{"show-stringtab", no_argument, NULL, 't'},
|
|
{"show-aliases", no_argument, NULL, 'T'},
|
|
{"out", a_argument, NULL, 'o'},
|
|
{"out-format", a_argument, NULL, 'O'},
|
|
{"invert-match", no_argument, NULL, 'v'},
|
|
USAGE_COMMON_LONG_OPTS,
|
|
};
|
|
static const char * const usage_opts_help[] = {
|
|
"Display address",
|
|
"Show all nodes/tags, colour those that match",
|
|
"Include contains containing property",
|
|
"Compatible nodes to include in grep",
|
|
"Compatible nodes to exclude in grep",
|
|
"Diff: Mark matching nodes with +, others with -",
|
|
"Enter direct subnode names of matching nodes",
|
|
"Display offset",
|
|
"Node/property/compatible string to include in grep",
|
|
"Node/property/compatible string to exclude in grep",
|
|
"Output a header",
|
|
"Put \"/dts-v1/;\" on first line of dts output",
|
|
"Output a region list",
|
|
"List strings in string table",
|
|
"Include mem_rsvmap section in binary output",
|
|
"Node to include in grep",
|
|
"Node to exclude in grep",
|
|
"Property to include in grep",
|
|
"Property to exclude in grep",
|
|
"Remove unused strings from string table",
|
|
"Include root node and all properties",
|
|
"Show all subnodes matching nodes",
|
|
"Don't include supernodes of matching nodes",
|
|
"Include string table in binary output",
|
|
"Include matching aliases in output",
|
|
"-o <output file>",
|
|
"-O <output format>",
|
|
"Invert the sense of matching (select non-matching lines)",
|
|
USAGE_COMMON_OPTS_HELP
|
|
};
|
|
|
|
/**
|
|
* Call getopt_long() with standard options
|
|
*
|
|
* Since all util code runs getopt in the same way, provide a helper.
|
|
*/
|
|
#define util_getopt_long() getopt_long(argc, argv, usage_short_opts, \
|
|
usage_long_opts, NULL)
|
|
|
|
void util_usage(const char *errmsg, const char *synopsis,
|
|
const char *short_opts, struct option const long_opts[],
|
|
const char * const opts_help[])
|
|
{
|
|
FILE *fp = errmsg ? stderr : stdout;
|
|
const char a_arg[] = "<arg>";
|
|
size_t a_arg_len = strlen(a_arg) + 1;
|
|
size_t i;
|
|
int optlen;
|
|
|
|
fprintf(fp,
|
|
"Usage: %s\n"
|
|
"\n"
|
|
"Options: -[%s]\n", synopsis, short_opts);
|
|
|
|
/* prescan the --long opt length to auto-align */
|
|
optlen = 0;
|
|
for (i = 0; long_opts[i].name; ++i) {
|
|
/* +1 is for space between --opt and help text */
|
|
int l = strlen(long_opts[i].name) + 1;
|
|
if (long_opts[i].has_arg == a_argument)
|
|
l += a_arg_len;
|
|
if (optlen < l)
|
|
optlen = l;
|
|
}
|
|
|
|
for (i = 0; long_opts[i].name; ++i) {
|
|
/* helps when adding new applets or options */
|
|
assert(opts_help[i] != NULL);
|
|
|
|
/* first output the short flag if it has one */
|
|
if (long_opts[i].val > '~')
|
|
fprintf(fp, " ");
|
|
else
|
|
fprintf(fp, " -%c, ", long_opts[i].val);
|
|
|
|
/* then the long flag */
|
|
if (long_opts[i].has_arg == no_argument) {
|
|
fprintf(fp, "--%-*s", optlen, long_opts[i].name);
|
|
} else {
|
|
fprintf(fp, "--%s %s%*s", long_opts[i].name, a_arg,
|
|
(int)(optlen - strlen(long_opts[i].name) -
|
|
a_arg_len), "");
|
|
}
|
|
|
|
/* finally the help text */
|
|
fprintf(fp, "%s\n", opts_help[i]);
|
|
}
|
|
|
|
if (errmsg) {
|
|
fprintf(fp, "\nError: %s\n", errmsg);
|
|
exit(EXIT_FAILURE);
|
|
} else {
|
|
exit(EXIT_SUCCESS);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Show usage and exit
|
|
*
|
|
* If you name all your usage variables with usage_xxx, then you can call this
|
|
* help macro rather than expanding all arguments yourself.
|
|
*
|
|
* @param errmsg If non-NULL, an error message to display
|
|
*/
|
|
#define usage(errmsg) \
|
|
util_usage(errmsg, usage_synopsis, usage_short_opts, \
|
|
usage_long_opts, usage_opts_help)
|
|
|
|
void util_version(void)
|
|
{
|
|
printf("Version: %s\n", "(U-Boot)");
|
|
exit(0);
|
|
}
|
|
|
|
static void scan_args(struct display_info *disp, int argc, char *argv[])
|
|
{
|
|
int opt;
|
|
|
|
while ((opt = util_getopt_long()) != EOF) {
|
|
int type = 0;
|
|
int inc = 1;
|
|
|
|
switch (opt) {
|
|
case_USAGE_COMMON_FLAGS
|
|
/* fallthrough */
|
|
case 'a':
|
|
disp->show_addr = 1;
|
|
break;
|
|
case 'A':
|
|
disp->all = 1;
|
|
break;
|
|
case 'b':
|
|
type = FDT_NODE_HAS_PROP;
|
|
break;
|
|
case 'C':
|
|
inc = 0;
|
|
/* fallthrough */
|
|
case 'c':
|
|
type = FDT_IS_COMPAT;
|
|
break;
|
|
case 'd':
|
|
disp->diff = 1;
|
|
break;
|
|
case 'e':
|
|
disp->flags |= FDT_REG_DIRECT_SUBNODES;
|
|
break;
|
|
case 'f':
|
|
disp->show_offset = 1;
|
|
break;
|
|
case 'G':
|
|
inc = 0;
|
|
/* fallthrough */
|
|
case 'g':
|
|
type = FDT_ANY_GLOBAL;
|
|
break;
|
|
case 'H':
|
|
disp->header = 1;
|
|
break;
|
|
case 'l':
|
|
disp->region_list = 1;
|
|
break;
|
|
case 'L':
|
|
disp->list_strings = 1;
|
|
break;
|
|
case 'm':
|
|
disp->flags |= FDT_REG_ADD_MEM_RSVMAP;
|
|
break;
|
|
case 'N':
|
|
inc = 0;
|
|
/* fallthrough */
|
|
case 'n':
|
|
type = FDT_IS_NODE;
|
|
break;
|
|
case 'o':
|
|
disp->output_fname = optarg;
|
|
break;
|
|
case 'O':
|
|
if (!strcmp(optarg, "dtb"))
|
|
disp->output = OUT_DTB;
|
|
else if (!strcmp(optarg, "dts"))
|
|
disp->output = OUT_DTS;
|
|
else if (!strcmp(optarg, "bin"))
|
|
disp->output = OUT_BIN;
|
|
else
|
|
usage("Unknown output format");
|
|
break;
|
|
case 'P':
|
|
inc = 0;
|
|
/* fallthrough */
|
|
case 'p':
|
|
type = FDT_IS_PROP;
|
|
break;
|
|
case 'r':
|
|
disp->remove_strings = 1;
|
|
break;
|
|
case 'R':
|
|
disp->include_root = 1;
|
|
break;
|
|
case 's':
|
|
disp->flags |= FDT_REG_ALL_SUBNODES;
|
|
break;
|
|
case 'S':
|
|
disp->flags &= ~FDT_REG_SUPERNODES;
|
|
break;
|
|
case 't':
|
|
disp->flags |= FDT_REG_ADD_STRING_TAB;
|
|
break;
|
|
case 'T':
|
|
disp->add_aliases = 1;
|
|
break;
|
|
case 'v':
|
|
disp->invert = 1;
|
|
break;
|
|
case 'I':
|
|
disp->show_dts_version = 1;
|
|
break;
|
|
}
|
|
|
|
if (type && value_add(disp, &disp->value_head, type, inc,
|
|
optarg))
|
|
usage("Cannot add value");
|
|
}
|
|
|
|
if (disp->invert && disp->types_exc)
|
|
usage("-v has no meaning when used with 'exclude' conditions");
|
|
}
|
|
|
|
int main(int argc, char *argv[])
|
|
{
|
|
char *filename = NULL;
|
|
struct display_info disp;
|
|
int ret;
|
|
|
|
/* set defaults */
|
|
memset(&disp, '\0', sizeof(disp));
|
|
disp.flags = FDT_REG_SUPERNODES; /* Default flags */
|
|
|
|
scan_args(&disp, argc, argv);
|
|
|
|
/* Show matched lines in colour if we can */
|
|
disp.colour = disp.all && isatty(0);
|
|
|
|
/* Any additional arguments can match anything, just like -g */
|
|
while (optind < argc - 1) {
|
|
if (value_add(&disp, &disp.value_head, FDT_IS_ANY, 1,
|
|
argv[optind++]))
|
|
usage("Cannot add value");
|
|
}
|
|
|
|
if (optind < argc)
|
|
filename = argv[optind++];
|
|
if (!filename)
|
|
usage("Missing filename");
|
|
|
|
/* If a valid .dtb is required, set flags to ensure we get one */
|
|
if (disp.output == OUT_DTB) {
|
|
disp.header = 1;
|
|
disp.flags |= FDT_REG_ADD_MEM_RSVMAP | FDT_REG_ADD_STRING_TAB;
|
|
}
|
|
|
|
if (disp.output_fname) {
|
|
disp.fout = fopen(disp.output_fname, "w");
|
|
if (!disp.fout)
|
|
usage("Cannot open output file");
|
|
} else {
|
|
disp.fout = stdout;
|
|
}
|
|
|
|
/* Run the grep and output the results */
|
|
ret = do_fdtgrep(&disp, filename);
|
|
if (disp.output_fname)
|
|
fclose(disp.fout);
|
|
if (ret)
|
|
return 1;
|
|
|
|
return 0;
|
|
}
|