1916 lines
48 KiB
C
1916 lines
48 KiB
C
/* SPDX-License-Identifier: GPL-2.0-only */
|
|
|
|
#include <inttypes.h>
|
|
#include <stdbool.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <unistd.h>
|
|
#include <getopt.h>
|
|
#include <dirent.h>
|
|
#include <errno.h>
|
|
#include <fcntl.h>
|
|
#include <ctype.h>
|
|
#include <arpa/inet.h>
|
|
#include <sys/types.h>
|
|
#include <sys/stat.h>
|
|
#include <sys/mman.h>
|
|
#include <libgen.h>
|
|
#include <assert.h>
|
|
#include <regex.h>
|
|
#include <commonlib/bsd/cbmem_id.h>
|
|
#include <commonlib/bsd/tpm_log_defs.h>
|
|
#include <commonlib/loglevel.h>
|
|
#include <commonlib/timestamp_serialized.h>
|
|
#include <commonlib/tpm_log_serialized.h>
|
|
#include <commonlib/coreboot_tables.h>
|
|
|
|
#ifdef __OpenBSD__
|
|
#include <sys/param.h>
|
|
#include <sys/sysctl.h>
|
|
#endif
|
|
|
|
#if defined(__i386__) || defined(__x86_64__)
|
|
#include <x86intrin.h>
|
|
#endif
|
|
|
|
typedef uint8_t u8;
|
|
typedef uint16_t u16;
|
|
typedef uint32_t u32;
|
|
typedef uint64_t u64;
|
|
|
|
/* Return < 0 on error, 0 on success. */
|
|
static int parse_cbtable(u64 address, size_t table_size);
|
|
|
|
struct mapping {
|
|
void *virt;
|
|
size_t offset;
|
|
size_t virt_size;
|
|
unsigned long long phys;
|
|
size_t size;
|
|
};
|
|
|
|
#define CBMEM_VERSION "1.1"
|
|
|
|
/* verbose output? */
|
|
static int verbose = 0;
|
|
#define debug(x...) if(verbose) printf(x)
|
|
|
|
/* File handle used to access /dev/mem */
|
|
static int mem_fd;
|
|
static struct mapping lbtable_mapping;
|
|
|
|
/* TSC frequency from the LB_TAG_TSC_INFO record. 0 if not present. */
|
|
static uint32_t tsc_freq_khz = 0;
|
|
|
|
static void die(const char *msg)
|
|
{
|
|
if (msg)
|
|
fputs(msg, stderr);
|
|
exit(1);
|
|
}
|
|
|
|
static unsigned long long system_page_size(void)
|
|
{
|
|
static unsigned long long page_size;
|
|
|
|
if (!page_size)
|
|
page_size = getpagesize();
|
|
|
|
return page_size;
|
|
}
|
|
|
|
static inline size_t size_to_mib(size_t sz)
|
|
{
|
|
return sz >> 20;
|
|
}
|
|
|
|
/* Return mapping of physical address requested. */
|
|
static void *mapping_virt(const struct mapping *mapping)
|
|
{
|
|
char *v = mapping->virt;
|
|
|
|
if (v == NULL)
|
|
return NULL;
|
|
|
|
return v + mapping->offset;
|
|
}
|
|
|
|
/* Returns virtual address on success, NULL on error. mapping is filled in. */
|
|
static void *map_memory_with_prot(struct mapping *mapping,
|
|
unsigned long long phys, size_t sz, int prot)
|
|
{
|
|
void *v;
|
|
unsigned long long page_size;
|
|
|
|
page_size = system_page_size();
|
|
|
|
mapping->virt = NULL;
|
|
mapping->offset = phys % page_size;
|
|
mapping->virt_size = sz + mapping->offset;
|
|
mapping->size = sz;
|
|
mapping->phys = phys;
|
|
|
|
if (size_to_mib(mapping->virt_size) == 0) {
|
|
debug("Mapping %zuB of physical memory at 0x%llx (requested 0x%llx).\n",
|
|
mapping->virt_size, phys - mapping->offset, phys);
|
|
} else {
|
|
debug("Mapping %zuMB of physical memory at 0x%llx (requested 0x%llx).\n",
|
|
size_to_mib(mapping->virt_size), phys - mapping->offset,
|
|
phys);
|
|
}
|
|
|
|
v = mmap(NULL, mapping->virt_size, prot, MAP_SHARED, mem_fd,
|
|
phys - mapping->offset);
|
|
|
|
if (v == MAP_FAILED) {
|
|
debug("Mapping failed %zuB of physical memory at 0x%llx.\n",
|
|
mapping->virt_size, phys - mapping->offset);
|
|
return NULL;
|
|
}
|
|
|
|
mapping->virt = v;
|
|
|
|
if (mapping->offset != 0)
|
|
debug(" ... padding virtual address with 0x%zx bytes.\n",
|
|
mapping->offset);
|
|
|
|
return mapping_virt(mapping);
|
|
}
|
|
|
|
/* Convenience helper for the common case of read-only mappings. */
|
|
static const void *map_memory(struct mapping *mapping, unsigned long long phys,
|
|
size_t sz)
|
|
{
|
|
return map_memory_with_prot(mapping, phys, sz, PROT_READ);
|
|
}
|
|
|
|
|
|
/* Returns 0 on success, < 0 on error. mapping is cleared if successful. */
|
|
static int unmap_memory(struct mapping *mapping)
|
|
{
|
|
if (mapping->virt == NULL)
|
|
return -1;
|
|
|
|
munmap(mapping->virt, mapping->virt_size);
|
|
mapping->virt = NULL;
|
|
mapping->offset = 0;
|
|
mapping->virt_size = 0;
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* Return size of physical address mapping requested. */
|
|
static size_t mapping_size(const struct mapping *mapping)
|
|
{
|
|
if (mapping->virt == NULL)
|
|
return 0;
|
|
|
|
return mapping->size;
|
|
}
|
|
|
|
/*
|
|
* Some architectures map /dev/mem memory in a way that doesn't support
|
|
* unaligned accesses. Most normal libc memcpy()s aren't safe to use in this
|
|
* case, so build our own which makes sure to never do unaligned accesses on
|
|
* *src (*dest is fine since we never map /dev/mem for writing).
|
|
*/
|
|
static void *aligned_memcpy(void *dest, const void *src, size_t n)
|
|
{
|
|
u8 *d = dest;
|
|
const volatile u8 *s = src; /* volatile to prevent optimization */
|
|
|
|
while ((uintptr_t)s & (sizeof(size_t) - 1)) {
|
|
if (n-- == 0)
|
|
return dest;
|
|
*d++ = *s++;
|
|
}
|
|
|
|
while (n >= sizeof(size_t)) {
|
|
*(size_t *)d = *(const volatile size_t *)s;
|
|
d += sizeof(size_t);
|
|
s += sizeof(size_t);
|
|
n -= sizeof(size_t);
|
|
}
|
|
|
|
while (n-- > 0)
|
|
*d++ = *s++;
|
|
|
|
return dest;
|
|
}
|
|
|
|
/*
|
|
* calculate ip checksum (16 bit quantities) on a passed in buffer. In case
|
|
* the buffer length is odd last byte is excluded from the calculation
|
|
*/
|
|
static u16 ipchcksum(const void *addr, unsigned size)
|
|
{
|
|
const u16 *p = addr;
|
|
unsigned i, n = size / 2; /* don't expect odd sized blocks */
|
|
u32 sum = 0;
|
|
|
|
for (i = 0; i < n; i++)
|
|
sum += p[i];
|
|
|
|
sum = (sum >> 16) + (sum & 0xffff);
|
|
sum += (sum >> 16);
|
|
sum = ~sum & 0xffff;
|
|
return (u16) sum;
|
|
}
|
|
|
|
/* Find the first cbmem entry filling in the details. */
|
|
static int find_cbmem_entry(uint32_t id, uint64_t *addr, size_t *size)
|
|
{
|
|
const uint8_t *table;
|
|
size_t offset;
|
|
int ret = -1;
|
|
|
|
table = mapping_virt(&lbtable_mapping);
|
|
|
|
if (table == NULL)
|
|
return -1;
|
|
|
|
offset = 0;
|
|
|
|
while (offset < mapping_size(&lbtable_mapping)) {
|
|
const struct lb_record *lbr;
|
|
struct lb_cbmem_entry lbe;
|
|
|
|
lbr = (const void *)(table + offset);
|
|
offset += lbr->size;
|
|
|
|
if (lbr->tag != LB_TAG_CBMEM_ENTRY)
|
|
continue;
|
|
|
|
aligned_memcpy(&lbe, lbr, sizeof(lbe));
|
|
if (lbe.id != id)
|
|
continue;
|
|
|
|
*addr = lbe.address;
|
|
*size = lbe.entry_size;
|
|
ret = 0;
|
|
break;
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Try finding the timestamp table and coreboot cbmem console starting from the
|
|
* passed in memory offset. Could be called recursively in case a forwarding
|
|
* entry is found.
|
|
*
|
|
* Returns pointer to a memory buffer containing the timestamp table or zero if
|
|
* none found.
|
|
*/
|
|
|
|
static struct lb_cbmem_ref timestamps;
|
|
static struct lb_cbmem_ref console;
|
|
static struct lb_cbmem_ref tpm_cb_log;
|
|
static struct lb_memory_range cbmem;
|
|
|
|
/* This is a work-around for a nasty problem introduced by initially having
|
|
* pointer sized entries in the lb_cbmem_ref structures. This caused problems
|
|
* on 64bit x86 systems because coreboot is 32bit on those systems.
|
|
* When the problem was found, it was corrected, but there are a lot of
|
|
* systems out there with a firmware that does not produce the right
|
|
* lb_cbmem_ref structure. Hence we try to autocorrect this issue here.
|
|
*/
|
|
static struct lb_cbmem_ref parse_cbmem_ref(const struct lb_cbmem_ref *cbmem_ref)
|
|
{
|
|
struct lb_cbmem_ref ret;
|
|
|
|
aligned_memcpy(&ret, cbmem_ref, sizeof(ret));
|
|
|
|
if (cbmem_ref->size < sizeof(*cbmem_ref))
|
|
ret.cbmem_addr = (uint32_t)ret.cbmem_addr;
|
|
|
|
debug(" cbmem_addr = %" PRIx64 "\n", ret.cbmem_addr);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static void parse_memory_tags(const struct lb_memory *mem)
|
|
{
|
|
int num_entries;
|
|
int i;
|
|
|
|
/* Peel off the header size and calculate the number of entries. */
|
|
num_entries = (mem->size - sizeof(*mem)) / sizeof(mem->map[0]);
|
|
|
|
for (i = 0; i < num_entries; i++) {
|
|
if (mem->map[i].type != LB_MEM_TABLE)
|
|
continue;
|
|
debug(" LB_MEM_TABLE found.\n");
|
|
/* The last one found is CBMEM */
|
|
aligned_memcpy(&cbmem, &mem->map[i], sizeof(cbmem));
|
|
}
|
|
}
|
|
|
|
/* Return < 0 on error, 0 on success, 1 if forwarding table entry found. */
|
|
static int parse_cbtable_entries(const struct mapping *table_mapping)
|
|
{
|
|
size_t i;
|
|
const struct lb_record *lbr_p;
|
|
size_t table_size = mapping_size(table_mapping);
|
|
const void *lbtable = mapping_virt(table_mapping);
|
|
int forwarding_table_found = 0;
|
|
|
|
for (i = 0; i < table_size; i += lbr_p->size) {
|
|
lbr_p = lbtable + i;
|
|
debug(" coreboot table entry 0x%02x\n", lbr_p->tag);
|
|
switch (lbr_p->tag) {
|
|
case LB_TAG_MEMORY:
|
|
debug(" Found memory map.\n");
|
|
parse_memory_tags(lbtable + i);
|
|
continue;
|
|
case LB_TAG_TIMESTAMPS: {
|
|
debug(" Found timestamp table.\n");
|
|
timestamps =
|
|
parse_cbmem_ref((struct lb_cbmem_ref *)lbr_p);
|
|
continue;
|
|
}
|
|
case LB_TAG_CBMEM_CONSOLE: {
|
|
debug(" Found cbmem console.\n");
|
|
console = parse_cbmem_ref((struct lb_cbmem_ref *)lbr_p);
|
|
continue;
|
|
}
|
|
case LB_TAG_TPM_CB_LOG: {
|
|
debug(" Found TPM CB log table.\n");
|
|
tpm_cb_log =
|
|
parse_cbmem_ref((struct lb_cbmem_ref *)lbr_p);
|
|
continue;
|
|
}
|
|
case LB_TAG_TSC_INFO:
|
|
debug(" Found TSC info.\n");
|
|
tsc_freq_khz = ((struct lb_tsc_info *)lbr_p)->freq_khz;
|
|
continue;
|
|
case LB_TAG_FORWARD: {
|
|
int ret;
|
|
/*
|
|
* This is a forwarding entry - repeat the
|
|
* search at the new address.
|
|
*/
|
|
struct lb_forward lbf_p =
|
|
*(const struct lb_forward *)lbr_p;
|
|
debug(" Found forwarding entry.\n");
|
|
ret = parse_cbtable(lbf_p.forward, 0);
|
|
|
|
/* Assume the forwarding entry is valid. If this fails
|
|
* then there's a total failure. */
|
|
if (ret < 0)
|
|
return -1;
|
|
forwarding_table_found = 1;
|
|
}
|
|
default:
|
|
break;
|
|
}
|
|
}
|
|
|
|
return forwarding_table_found;
|
|
}
|
|
|
|
/* Return < 0 on error, 0 on success. */
|
|
static int parse_cbtable(u64 address, size_t table_size)
|
|
{
|
|
const void *buf;
|
|
struct mapping header_mapping;
|
|
size_t req_size;
|
|
size_t i;
|
|
|
|
req_size = table_size;
|
|
/* Default to 4 KiB search space. */
|
|
if (req_size == 0)
|
|
req_size = 4 * 1024;
|
|
|
|
debug("Looking for coreboot table at %" PRIx64 " %zd bytes.\n",
|
|
address, req_size);
|
|
|
|
buf = map_memory(&header_mapping, address, req_size);
|
|
|
|
if (!buf)
|
|
return -1;
|
|
|
|
/* look at every 16 bytes */
|
|
for (i = 0; i <= req_size - sizeof(struct lb_header); i += 16) {
|
|
int ret;
|
|
const struct lb_header *lbh;
|
|
struct mapping table_mapping;
|
|
|
|
lbh = buf + i;
|
|
if (memcmp(lbh->signature, "LBIO", sizeof(lbh->signature)) ||
|
|
!lbh->header_bytes ||
|
|
ipchcksum(lbh, sizeof(*lbh))) {
|
|
continue;
|
|
}
|
|
|
|
/* Map in the whole table to parse. */
|
|
if (!map_memory(&table_mapping, address + i + lbh->header_bytes,
|
|
lbh->table_bytes)) {
|
|
debug("Couldn't map in table\n");
|
|
continue;
|
|
}
|
|
|
|
if (ipchcksum(mapping_virt(&table_mapping), lbh->table_bytes) !=
|
|
lbh->table_checksum) {
|
|
debug("Signature found, but wrong checksum.\n");
|
|
unmap_memory(&table_mapping);
|
|
continue;
|
|
}
|
|
|
|
debug("Found!\n");
|
|
|
|
ret = parse_cbtable_entries(&table_mapping);
|
|
|
|
/* Table parsing failed. */
|
|
if (ret < 0) {
|
|
unmap_memory(&table_mapping);
|
|
continue;
|
|
}
|
|
|
|
/* Succeeded in parsing the table. Header not needed anymore. */
|
|
unmap_memory(&header_mapping);
|
|
|
|
/*
|
|
* Table parsing succeeded. If forwarding table not found update
|
|
* coreboot table mapping for future use.
|
|
*/
|
|
if (ret == 0)
|
|
lbtable_mapping = table_mapping;
|
|
else
|
|
unmap_memory(&table_mapping);
|
|
|
|
return 0;
|
|
}
|
|
|
|
unmap_memory(&header_mapping);
|
|
|
|
return -1;
|
|
}
|
|
|
|
#if defined(linux) && (defined(__i386__) || defined(__x86_64__))
|
|
/*
|
|
* read CPU frequency from a sysfs file, return an frequency in Megahertz as
|
|
* an int or exit on any error.
|
|
*/
|
|
static unsigned long arch_tick_frequency(void)
|
|
{
|
|
FILE *cpuf;
|
|
char freqs[100];
|
|
int size;
|
|
char *endp;
|
|
u64 rv;
|
|
|
|
const char* freq_file =
|
|
"/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq";
|
|
|
|
cpuf = fopen(freq_file, "r");
|
|
if (!cpuf) {
|
|
fprintf(stderr, "Could not open %s: %s\n",
|
|
freq_file, strerror(errno));
|
|
exit(1);
|
|
}
|
|
|
|
memset(freqs, 0, sizeof(freqs));
|
|
size = fread(freqs, 1, sizeof(freqs), cpuf);
|
|
if (!size || (size == sizeof(freqs))) {
|
|
fprintf(stderr, "Wrong number of bytes(%d) read from %s\n",
|
|
size, freq_file);
|
|
exit(1);
|
|
}
|
|
fclose(cpuf);
|
|
rv = strtoull(freqs, &endp, 10);
|
|
|
|
if (*endp == '\0' || *endp == '\n')
|
|
/* cpuinfo_max_freq is in kHz. Convert it to MHz. */
|
|
return rv / 1000;
|
|
fprintf(stderr, "Wrong formatted value ^%s^ read from %s\n",
|
|
freqs, freq_file);
|
|
exit(1);
|
|
}
|
|
#elif defined(__OpenBSD__) && (defined(__i386__) || defined(__x86_64__))
|
|
static unsigned long arch_tick_frequency(void)
|
|
{
|
|
int mib[2] = { CTL_HW, HW_CPUSPEED };
|
|
static int value = 0;
|
|
size_t value_len = sizeof(value);
|
|
|
|
/* Return 1 MHz when sysctl fails. */
|
|
if ((value == 0) && (sysctl(mib, 2, &value, &value_len, NULL, 0) == -1))
|
|
return 1;
|
|
|
|
return value;
|
|
}
|
|
#else
|
|
static unsigned long arch_tick_frequency(void)
|
|
{
|
|
/* 1 MHz = 1us. */
|
|
return 1;
|
|
}
|
|
#endif
|
|
|
|
static unsigned long tick_freq_mhz;
|
|
|
|
static void timestamp_set_tick_freq(unsigned long table_tick_freq_mhz)
|
|
{
|
|
tick_freq_mhz = table_tick_freq_mhz;
|
|
|
|
/* Honor table frequency if present. */
|
|
if (!tick_freq_mhz)
|
|
tick_freq_mhz = arch_tick_frequency();
|
|
|
|
if (!tick_freq_mhz) {
|
|
fprintf(stderr, "Cannot determine timestamp tick frequency.\n");
|
|
exit(1);
|
|
}
|
|
|
|
debug("Timestamp tick frequency: %ld MHz\n", tick_freq_mhz);
|
|
}
|
|
|
|
static u64 arch_convert_raw_ts_entry(u64 ts)
|
|
{
|
|
return ts / tick_freq_mhz;
|
|
}
|
|
|
|
/*
|
|
* Print an integer in 'normalized' form - with commas separating every three
|
|
* decimal orders.
|
|
*/
|
|
static void print_norm(u64 v)
|
|
{
|
|
if (v >= 1000) {
|
|
/* print the higher order sections first */
|
|
print_norm(v / 1000);
|
|
printf(",%3.3u", (u32)(v % 1000));
|
|
} else {
|
|
printf("%u", (u32)(v % 1000));
|
|
}
|
|
}
|
|
|
|
static uint64_t timestamp_get(uint64_t table_tick_freq_mhz)
|
|
{
|
|
#if defined(__i386__) || defined(__x86_64__)
|
|
uint64_t tsc = __rdtsc();
|
|
|
|
/* No tick frequency specified means raw TSC values. */
|
|
if (!table_tick_freq_mhz)
|
|
return tsc;
|
|
|
|
if (tsc_freq_khz)
|
|
return tsc * table_tick_freq_mhz * 1000 / tsc_freq_khz;
|
|
#else
|
|
(void)table_tick_freq_mhz;
|
|
#endif
|
|
die("Don't know how to obtain timestamps on this platform.\n");
|
|
return 0;
|
|
}
|
|
|
|
static const char *timestamp_name(uint32_t id)
|
|
{
|
|
for (size_t i = 0; i < ARRAY_SIZE(timestamp_ids); i++) {
|
|
if (timestamp_ids[i].id == id)
|
|
return timestamp_ids[i].name;
|
|
}
|
|
return "<unknown>";
|
|
}
|
|
|
|
static uint32_t timestamp_enum_name_to_id(const char *name)
|
|
{
|
|
for (size_t i = 0; i < ARRAY_SIZE(timestamp_ids); i++) {
|
|
if (!strcmp(timestamp_ids[i].enum_name, name))
|
|
return timestamp_ids[i].id;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static uint64_t timestamp_print_parseable_entry(uint32_t id, uint64_t stamp,
|
|
uint64_t prev_stamp)
|
|
{
|
|
const char *name;
|
|
uint64_t step_time;
|
|
|
|
name = timestamp_name(id);
|
|
|
|
step_time = arch_convert_raw_ts_entry(stamp - prev_stamp);
|
|
|
|
/* ID<tab>absolute time<tab>relative time<tab>description */
|
|
printf("%d\t", id);
|
|
printf("%llu\t", (long long)arch_convert_raw_ts_entry(stamp));
|
|
printf("%llu\t", (long long)step_time);
|
|
printf("%s\n", name);
|
|
|
|
return step_time;
|
|
}
|
|
|
|
static uint64_t timestamp_print_entry(uint32_t id, uint64_t stamp, uint64_t prev_stamp)
|
|
{
|
|
const char *name;
|
|
uint64_t step_time;
|
|
|
|
name = timestamp_name(id);
|
|
|
|
printf("%4d:", id);
|
|
printf("%-50s", name);
|
|
print_norm(arch_convert_raw_ts_entry(stamp));
|
|
step_time = arch_convert_raw_ts_entry(stamp - prev_stamp);
|
|
if (prev_stamp) {
|
|
printf(" (");
|
|
print_norm(step_time);
|
|
printf(")");
|
|
}
|
|
printf("\n");
|
|
|
|
return step_time;
|
|
}
|
|
|
|
static int compare_timestamp_entries(const void *a, const void *b)
|
|
{
|
|
const struct timestamp_entry *tse_a = (struct timestamp_entry *)a;
|
|
const struct timestamp_entry *tse_b = (struct timestamp_entry *)b;
|
|
|
|
if (tse_a->entry_stamp > tse_b->entry_stamp)
|
|
return 1;
|
|
else if (tse_a->entry_stamp < tse_b->entry_stamp)
|
|
return -1;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int find_matching_end(struct timestamp_table *sorted_tst_p, uint32_t start, uint32_t end)
|
|
{
|
|
uint32_t id = sorted_tst_p->entries[start].entry_id;
|
|
uint32_t possible_match = 0;
|
|
|
|
for (uint32_t i = 0; i < ARRAY_SIZE(timestamp_ids); ++i) {
|
|
if (timestamp_ids[i].id == id) {
|
|
possible_match = timestamp_ids[i].id_end;
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* No match found or timestamp not defined in IDs table */
|
|
if (!possible_match)
|
|
return -1;
|
|
|
|
for (uint32_t i = start + 1; i < end; i++)
|
|
if (sorted_tst_p->entries[i].entry_id == possible_match)
|
|
return i;
|
|
|
|
return -1;
|
|
}
|
|
|
|
static const char *get_timestamp_name(const uint32_t id)
|
|
{
|
|
for (uint32_t i = 0; i < ARRAY_SIZE(timestamp_ids); i++)
|
|
if (timestamp_ids[i].id == id)
|
|
return timestamp_ids[i].enum_name;
|
|
|
|
return "UNKNOWN";
|
|
}
|
|
|
|
struct ts_range_stack {
|
|
const char *name;
|
|
const char *end_name;
|
|
uint32_t end;
|
|
};
|
|
|
|
static void print_with_path(struct ts_range_stack *range_stack, const int stacklvl,
|
|
const uint64_t stamp, const char *last_part)
|
|
{
|
|
for (int i = 1; i <= stacklvl; ++i) {
|
|
printf("%s -> %s", range_stack[i].name, range_stack[i].end_name);
|
|
if (i < stacklvl || last_part)
|
|
putchar(';');
|
|
}
|
|
if (last_part)
|
|
printf("%s", last_part);
|
|
printf(" %llu\n", (long long)arch_convert_raw_ts_entry(stamp));
|
|
}
|
|
|
|
enum timestamps_print_type {
|
|
TIMESTAMPS_PRINT_NONE,
|
|
TIMESTAMPS_PRINT_NORMAL,
|
|
TIMESTAMPS_PRINT_MACHINE_READABLE,
|
|
TIMESTAMPS_PRINT_STACKED,
|
|
};
|
|
|
|
/* dump the timestamp table */
|
|
static void dump_timestamps(enum timestamps_print_type output_type)
|
|
{
|
|
const struct timestamp_table *tst_p;
|
|
struct timestamp_table *sorted_tst_p;
|
|
size_t size;
|
|
uint64_t prev_stamp = 0;
|
|
uint64_t total_time = 0;
|
|
struct mapping timestamp_mapping;
|
|
|
|
if (timestamps.tag != LB_TAG_TIMESTAMPS) {
|
|
fprintf(stderr, "No timestamps found in coreboot table.\n");
|
|
return;
|
|
}
|
|
|
|
size = sizeof(*tst_p);
|
|
tst_p = map_memory(×tamp_mapping, timestamps.cbmem_addr, size);
|
|
if (!tst_p)
|
|
die("Unable to map timestamp header\n");
|
|
|
|
timestamp_set_tick_freq(tst_p->tick_freq_mhz);
|
|
|
|
if (output_type == TIMESTAMPS_PRINT_NORMAL)
|
|
printf("%d entries total:\n\n", tst_p->num_entries);
|
|
size += tst_p->num_entries * sizeof(tst_p->entries[0]);
|
|
|
|
unmap_memory(×tamp_mapping);
|
|
|
|
tst_p = map_memory(×tamp_mapping, timestamps.cbmem_addr, size);
|
|
if (!tst_p)
|
|
die("Unable to map full timestamp table\n");
|
|
|
|
sorted_tst_p = malloc(size + sizeof(struct timestamp_entry));
|
|
if (!sorted_tst_p)
|
|
die("Failed to allocate memory");
|
|
aligned_memcpy(sorted_tst_p, tst_p, size);
|
|
|
|
/*
|
|
* Insert a timestamp to represent the base time (start of coreboot),
|
|
* in case we have to rebase for negative timestamps below.
|
|
*/
|
|
sorted_tst_p->entries[tst_p->num_entries].entry_id = 0;
|
|
sorted_tst_p->entries[tst_p->num_entries].entry_stamp = 0;
|
|
sorted_tst_p->num_entries += 1;
|
|
|
|
qsort(&sorted_tst_p->entries[0], sorted_tst_p->num_entries,
|
|
sizeof(struct timestamp_entry), compare_timestamp_entries);
|
|
|
|
/*
|
|
* If there are negative timestamp entries, rebase all of the
|
|
* timestamps to the lowest one in the list.
|
|
*/
|
|
if (sorted_tst_p->entries[0].entry_stamp < 0) {
|
|
sorted_tst_p->base_time = -sorted_tst_p->entries[0].entry_stamp;
|
|
prev_stamp = 0;
|
|
} else {
|
|
prev_stamp = tst_p->base_time;
|
|
}
|
|
|
|
struct ts_range_stack range_stack[20];
|
|
range_stack[0].end = sorted_tst_p->num_entries;
|
|
int stacklvl = 0;
|
|
|
|
for (uint32_t i = 0; i < sorted_tst_p->num_entries; i++) {
|
|
uint64_t stamp;
|
|
const struct timestamp_entry *tse = &sorted_tst_p->entries[i];
|
|
|
|
/* Make all timestamps absolute. */
|
|
stamp = tse->entry_stamp + sorted_tst_p->base_time;
|
|
if (output_type == TIMESTAMPS_PRINT_MACHINE_READABLE) {
|
|
timestamp_print_parseable_entry(tse->entry_id, stamp, prev_stamp);
|
|
} else if (output_type == TIMESTAMPS_PRINT_NORMAL) {
|
|
total_time += timestamp_print_entry(tse->entry_id, stamp, prev_stamp);
|
|
} else if (output_type == TIMESTAMPS_PRINT_STACKED) {
|
|
bool end_of_range = false;
|
|
/* Iterate over stacked entries to pop all ranges, which are closed by
|
|
current element. For example, assuming two ranges: (TS_A, TS_C),
|
|
(TS_B, TS_C) it will pop all of them instead of just last one. */
|
|
while (stacklvl > 0 && range_stack[stacklvl].end == i) {
|
|
end_of_range = true;
|
|
stacklvl--;
|
|
}
|
|
|
|
int match =
|
|
find_matching_end(sorted_tst_p, i, range_stack[stacklvl].end);
|
|
if (match != -1) {
|
|
const uint64_t match_stamp =
|
|
sorted_tst_p->entries[match].entry_stamp
|
|
+ sorted_tst_p->base_time;
|
|
stacklvl++;
|
|
assert(stacklvl < (int)ARRAY_SIZE(range_stack));
|
|
range_stack[stacklvl].name = get_timestamp_name(tse->entry_id);
|
|
range_stack[stacklvl].end_name = get_timestamp_name(
|
|
sorted_tst_p->entries[match].entry_id);
|
|
range_stack[stacklvl].end = match;
|
|
print_with_path(range_stack, stacklvl, match_stamp - stamp,
|
|
NULL);
|
|
} else if (!end_of_range) {
|
|
print_with_path(range_stack, stacklvl, stamp - prev_stamp,
|
|
get_timestamp_name(tse->entry_id));
|
|
}
|
|
/* else: No match && end_of_range == true */
|
|
}
|
|
prev_stamp = stamp;
|
|
}
|
|
|
|
if (output_type == TIMESTAMPS_PRINT_NORMAL) {
|
|
printf("\nTotal Time: ");
|
|
print_norm(total_time);
|
|
printf("\n");
|
|
}
|
|
|
|
unmap_memory(×tamp_mapping);
|
|
free(sorted_tst_p);
|
|
}
|
|
|
|
/* add a timestamp entry */
|
|
static void timestamp_add_now(uint32_t timestamp_id)
|
|
{
|
|
struct timestamp_table *tst_p;
|
|
struct mapping timestamp_mapping;
|
|
|
|
if (timestamps.tag != LB_TAG_TIMESTAMPS) {
|
|
die("No timestamps found in coreboot table.\n");
|
|
}
|
|
|
|
tst_p = map_memory_with_prot(×tamp_mapping, timestamps.cbmem_addr,
|
|
timestamps.size, PROT_READ | PROT_WRITE);
|
|
if (!tst_p)
|
|
die("Unable to map timestamp table\n");
|
|
|
|
/*
|
|
* Note that coreboot sizes the cbmem entry in the table according to
|
|
* max_entries, so it's OK to just add more entries if there's room.
|
|
*/
|
|
if (tst_p->num_entries >= tst_p->max_entries) {
|
|
die("Not enough space to add timestamp.\n");
|
|
} else {
|
|
int64_t time =
|
|
timestamp_get(tst_p->tick_freq_mhz) - tst_p->base_time;
|
|
tst_p->entries[tst_p->num_entries].entry_id = timestamp_id;
|
|
tst_p->entries[tst_p->num_entries].entry_stamp = time;
|
|
tst_p->num_entries += 1;
|
|
}
|
|
|
|
unmap_memory(×tamp_mapping);
|
|
}
|
|
|
|
static bool can_print(const uint8_t *data, size_t len)
|
|
{
|
|
unsigned int i;
|
|
for (i = 0; i < len; i++) {
|
|
if (!isprint(data[i]) && !isspace(data[i])) {
|
|
/* If printable prefix is followed by zeroes, this is a valid string */
|
|
for (; i < len; i++) {
|
|
if (data[i] != 0)
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
static void print_hex_string(const uint8_t *hex, size_t len)
|
|
{
|
|
unsigned int i;
|
|
for (i = 0; i < len; i++)
|
|
printf("%02x", hex[i]);
|
|
}
|
|
|
|
static void print_hex_line(const uint8_t *hex, size_t len)
|
|
{
|
|
print_hex_string(hex, len);
|
|
printf("\n");
|
|
}
|
|
|
|
static void print_event_type(uint32_t event_type)
|
|
{
|
|
unsigned int known_event_count = ARRAY_SIZE(tpm_event_types);
|
|
if (event_type >= known_event_count)
|
|
printf("Unknown (0x%x >= %u)", event_type, known_event_count);
|
|
else
|
|
printf("%s", tpm_event_types[event_type]);
|
|
}
|
|
|
|
static void parse_tpm12_log(const struct tcpa_spec_entry *spec_log)
|
|
{
|
|
const uint8_t zero_block[sizeof(struct tcpa_spec_entry)] = {0};
|
|
|
|
uintptr_t current;
|
|
uint32_t counter = 0;
|
|
|
|
printf("TCPA log:\n");
|
|
printf("\tSpecification: %d.%d%d\n",
|
|
spec_log->spec_version_major,
|
|
spec_log->spec_version_minor,
|
|
spec_log->spec_errata);
|
|
printf("\tPlatform class: %s\n",
|
|
le32toh(spec_log->platform_class) == 0 ? "PC Client" :
|
|
le32toh(spec_log->platform_class) == 1 ? "Server" : "Unknown");
|
|
|
|
current = (uintptr_t)&spec_log->vendor_info[spec_log->vendor_info_size];
|
|
while (memcmp((const void *)current, (const void *)zero_block, sizeof(zero_block))) {
|
|
uint32_t len;
|
|
struct tcpa_log_entry *log_entry = (void *)current;
|
|
uint32_t event_type = le32toh(log_entry->event_type);
|
|
|
|
printf("TCPA log entry %u:\n", ++counter);
|
|
printf("\tPCR: %d\n", le32toh(log_entry->pcr));
|
|
printf("\tEvent type: ");
|
|
print_event_type(event_type);
|
|
printf("\n");
|
|
printf("\tDigest: ");
|
|
print_hex_line(log_entry->digest, SHA1_DIGEST_SIZE);
|
|
current += sizeof(struct tcpa_log_entry);
|
|
len = le32toh(log_entry->event_data_size);
|
|
if (len != 0) {
|
|
current += len;
|
|
printf("\tEvent data: ");
|
|
if (can_print(log_entry->event, len))
|
|
printf("%.*s\n", len, log_entry->event);
|
|
else
|
|
print_hex_line(log_entry->event, len);
|
|
} else {
|
|
printf("\tEvent data not provided\n");
|
|
}
|
|
}
|
|
}
|
|
|
|
static uint32_t print_tpm2_digests(struct tcg_pcr_event2_header *log_entry)
|
|
{
|
|
unsigned int i;
|
|
uintptr_t current = (uintptr_t)log_entry->digests;
|
|
|
|
for (i = 0; i < le32toh(log_entry->digest_count); i++) {
|
|
struct tpm_hash_algorithm *hash = (struct tpm_hash_algorithm *)current;
|
|
switch (le16toh(hash->hashAlg)) {
|
|
case TPM2_ALG_SHA1:
|
|
printf("\t\t SHA1: ");
|
|
print_hex_line(hash->digest.sha1, SHA1_DIGEST_SIZE);
|
|
current += sizeof(hash->hashAlg) + SHA1_DIGEST_SIZE;
|
|
break;
|
|
case TPM2_ALG_SHA256:
|
|
printf("\t\t SHA256: ");
|
|
print_hex_line(hash->digest.sha256, SHA256_DIGEST_SIZE);
|
|
current += sizeof(hash->hashAlg) + SHA256_DIGEST_SIZE;
|
|
break;
|
|
case TPM2_ALG_SHA384:
|
|
printf("\t\t SHA384: ");
|
|
print_hex_line(hash->digest.sha384, SHA384_DIGEST_SIZE);
|
|
current += sizeof(hash->hashAlg) + SHA384_DIGEST_SIZE;
|
|
break;
|
|
case TPM2_ALG_SHA512:
|
|
printf("\t\t SHA512: ");
|
|
print_hex_line(hash->digest.sha512, SHA512_DIGEST_SIZE);
|
|
current += sizeof(hash->hashAlg) + SHA512_DIGEST_SIZE;
|
|
break;
|
|
case TPM2_ALG_SM3_256:
|
|
printf("\t\t SM3: ");
|
|
print_hex_line(hash->digest.sm3_256, SM3_256_DIGEST_SIZE);
|
|
current += sizeof(hash->hashAlg) + SM3_256_DIGEST_SIZE;
|
|
break;
|
|
default:
|
|
die("Unknown hash algorithm\n");
|
|
}
|
|
}
|
|
|
|
return current - (uintptr_t)&log_entry->digest_count;
|
|
}
|
|
|
|
static void parse_tpm2_log(const struct tcg_efi_spec_id_event *tpm2_log)
|
|
{
|
|
const uint8_t zero_block[12] = {0}; /* Only PCR index, event type and digest count */
|
|
|
|
uintptr_t current;
|
|
uint32_t counter = 0;
|
|
|
|
printf("TPM2 log:\n");
|
|
printf("\tSpecification: %d.%d%d\n",
|
|
tpm2_log->spec_version_major,
|
|
tpm2_log->spec_version_minor,
|
|
tpm2_log->spec_errata);
|
|
printf("\tPlatform class: %s\n",
|
|
le32toh(tpm2_log->platform_class) == 0 ? "PC Client" :
|
|
le32toh(tpm2_log->platform_class) == 1 ? "Server" : "Unknown");
|
|
|
|
/* Start after the first variable-sized part of the header */
|
|
current = (uintptr_t)&tpm2_log->digest_sizes[le32toh(tpm2_log->num_of_algorithms)];
|
|
/* current is at `uint8_t vendor_info_size` here */
|
|
current += 1 + *(uint8_t *)current;
|
|
|
|
while (memcmp((const void *)current, (const void *)zero_block, sizeof(zero_block))) {
|
|
uint32_t len;
|
|
struct tcg_pcr_event2_header *log_entry = (void *)current;
|
|
uint32_t event_type = le32toh(log_entry->event_type);
|
|
|
|
printf("TPM2 log entry %u:\n", ++counter);
|
|
printf("\tPCR: %d\n", le32toh(log_entry->pcr_index));
|
|
printf("\tEvent type: ");
|
|
print_event_type(event_type);
|
|
printf("\n");
|
|
|
|
current = (uintptr_t)&log_entry->digest_count;
|
|
if (le32toh(log_entry->digest_count) > 0) {
|
|
printf("\tDigests:\n");
|
|
current += print_tpm2_digests(log_entry);
|
|
} else {
|
|
printf("\tNo digests in this log entry\n");
|
|
current += sizeof(log_entry->digest_count);
|
|
}
|
|
/* Now event size and event are left to be parsed */
|
|
len = le32toh(*(uint32_t *)current);
|
|
current += sizeof(uint32_t);
|
|
if (len != 0) {
|
|
printf("\tEvent data: %.*s\n", len, (const char *)current);
|
|
current += len;
|
|
} else {
|
|
printf("\tEvent data not provided\n");
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Dump the TPM log table in format defined by specifications */
|
|
static void dump_tpm_std_log(uint64_t addr, size_t size)
|
|
{
|
|
const void *event_log;
|
|
const struct tcpa_spec_entry *tspec_entry;
|
|
const struct tcg_efi_spec_id_event *tcg_spec_entry;
|
|
struct mapping log_mapping;
|
|
|
|
event_log = map_memory(&log_mapping, addr, size);
|
|
if (!event_log)
|
|
die("Unable to map TPM eventlog\n");
|
|
|
|
tspec_entry = event_log;
|
|
if (!strcmp((const char *)tspec_entry->signature, TCPA_SPEC_ID_EVENT_SIGNATURE)) {
|
|
if (tspec_entry->spec_version_major == 1 &&
|
|
tspec_entry->spec_version_minor == 2 &&
|
|
tspec_entry->spec_errata >= 1 &&
|
|
le32toh(tspec_entry->entry.event_type) == EV_NO_ACTION) {
|
|
parse_tpm12_log(tspec_entry);
|
|
} else {
|
|
fprintf(stderr, "Unknown TPM1.2 log specification\n");
|
|
}
|
|
unmap_memory(&log_mapping);
|
|
return;
|
|
}
|
|
|
|
tcg_spec_entry = event_log;
|
|
if (!strcmp((const char *)tcg_spec_entry->signature, TCG_EFI_SPEC_ID_EVENT_SIGNATURE)) {
|
|
if (tcg_spec_entry->spec_version_major == 2 &&
|
|
tcg_spec_entry->spec_version_minor == 0 &&
|
|
le32toh(tcg_spec_entry->event_type) == EV_NO_ACTION) {
|
|
parse_tpm2_log(tcg_spec_entry);
|
|
} else {
|
|
fprintf(stderr, "Unknown TPM2 log specification.\n");
|
|
}
|
|
unmap_memory(&log_mapping);
|
|
return;
|
|
}
|
|
|
|
fprintf(stderr, "Unknown TPM log specification: %.*s\n",
|
|
(int)sizeof(tcg_spec_entry->signature),
|
|
(const char *)tcg_spec_entry->signature);
|
|
|
|
unmap_memory(&log_mapping);
|
|
}
|
|
|
|
/* dump the TPM CB log table */
|
|
static void dump_tpm_cb_log(void)
|
|
{
|
|
const struct tpm_cb_log_table *tclt_p;
|
|
size_t size;
|
|
struct mapping log_mapping;
|
|
|
|
if (tpm_cb_log.tag != LB_TAG_TPM_CB_LOG) {
|
|
fprintf(stderr, "No TPM log found in coreboot table.\n");
|
|
return;
|
|
}
|
|
|
|
size = sizeof(*tclt_p);
|
|
tclt_p = map_memory(&log_mapping, tpm_cb_log.cbmem_addr, size);
|
|
if (!tclt_p)
|
|
die("Unable to map TPM log header\n");
|
|
|
|
size += tclt_p->num_entries * sizeof(tclt_p->entries[0]);
|
|
|
|
unmap_memory(&log_mapping);
|
|
|
|
tclt_p = map_memory(&log_mapping, tpm_cb_log.cbmem_addr, size);
|
|
if (!tclt_p)
|
|
die("Unable to map full TPM log table\n");
|
|
|
|
printf("coreboot TPM log:\n\n");
|
|
|
|
for (uint16_t i = 0; i < tclt_p->num_entries; i++) {
|
|
const struct tpm_cb_log_entry *tce = &tclt_p->entries[i];
|
|
|
|
printf(" PCR-%u ", tce->pcr);
|
|
print_hex_string(tce->digest, tce->digest_length);
|
|
printf(" %s [%s]\n", tce->digest_type, tce->name);
|
|
}
|
|
|
|
unmap_memory(&log_mapping);
|
|
}
|
|
|
|
static void dump_tpm_log(void)
|
|
{
|
|
uint64_t start;
|
|
size_t size;
|
|
|
|
if (!find_cbmem_entry(CBMEM_ID_TCPA_TCG_LOG, &start, &size) ||
|
|
!find_cbmem_entry(CBMEM_ID_TPM2_TCG_LOG, &start, &size))
|
|
dump_tpm_std_log(start, size);
|
|
else
|
|
dump_tpm_cb_log();
|
|
}
|
|
|
|
struct cbmem_console {
|
|
u32 size;
|
|
u32 cursor;
|
|
u8 body[];
|
|
} __attribute__ ((__packed__));
|
|
|
|
#define CBMC_CURSOR_MASK ((1 << 28) - 1)
|
|
#define CBMC_OVERFLOW (1 << 31)
|
|
|
|
enum console_print_type {
|
|
CONSOLE_PRINT_FULL = 0,
|
|
CONSOLE_PRINT_LAST,
|
|
CONSOLE_PRINT_PREVIOUS,
|
|
};
|
|
|
|
static int parse_loglevel(char *arg, int *print_unknown_logs)
|
|
{
|
|
if (arg[0] == '+') {
|
|
*print_unknown_logs = 1;
|
|
arg++;
|
|
} else {
|
|
*print_unknown_logs = 0;
|
|
}
|
|
|
|
char *endptr;
|
|
int loglevel = strtol(arg, &endptr, 0);
|
|
if (*endptr == '\0' && loglevel >= BIOS_EMERG && loglevel <= BIOS_LOG_PREFIX_MAX_LEVEL)
|
|
return loglevel;
|
|
|
|
/* Only match first 3 characters so `NOTE` and `NOTICE` both match. */
|
|
for (int i = BIOS_EMERG; i <= BIOS_LOG_PREFIX_MAX_LEVEL; i++)
|
|
if (!strncasecmp(arg, bios_log_prefix[i], 3))
|
|
return i;
|
|
|
|
*print_unknown_logs = 1;
|
|
return BIOS_NEVER;
|
|
}
|
|
|
|
/* dump the cbmem console */
|
|
static void dump_console(enum console_print_type type, int max_loglevel, int print_unknown_logs)
|
|
{
|
|
const struct cbmem_console *console_p;
|
|
char *console_c;
|
|
size_t size, cursor, previous;
|
|
struct mapping console_mapping;
|
|
|
|
if (console.tag != LB_TAG_CBMEM_CONSOLE) {
|
|
fprintf(stderr, "No console found in coreboot table.\n");
|
|
return;
|
|
}
|
|
|
|
size = sizeof(*console_p);
|
|
console_p = map_memory(&console_mapping, console.cbmem_addr, size);
|
|
if (!console_p)
|
|
die("Unable to map console object.\n");
|
|
|
|
cursor = console_p->cursor & CBMC_CURSOR_MASK;
|
|
if (!(console_p->cursor & CBMC_OVERFLOW) && cursor < console_p->size)
|
|
size = cursor;
|
|
else
|
|
size = console_p->size;
|
|
unmap_memory(&console_mapping);
|
|
|
|
console_c = malloc(size + 1);
|
|
if (!console_c) {
|
|
fprintf(stderr, "Not enough memory for console.\n");
|
|
exit(1);
|
|
}
|
|
console_c[size] = '\0';
|
|
|
|
console_p = map_memory(&console_mapping, console.cbmem_addr,
|
|
size + sizeof(*console_p));
|
|
|
|
if (!console_p)
|
|
die("Unable to map full console object.\n");
|
|
|
|
if (console_p->cursor & CBMC_OVERFLOW) {
|
|
if (cursor >= size) {
|
|
printf("cbmem: ERROR: CBMEM console struct is illegal, "
|
|
"output may be corrupt or out of order!\n\n");
|
|
cursor = 0;
|
|
}
|
|
aligned_memcpy(console_c, console_p->body + cursor,
|
|
size - cursor);
|
|
aligned_memcpy(console_c + size - cursor,
|
|
console_p->body, cursor);
|
|
} else {
|
|
aligned_memcpy(console_c, console_p->body, size);
|
|
}
|
|
|
|
/* Slight memory corruption may occur between reboots and give us a few
|
|
unprintable characters like '\0'. Replace them with '?' on output. */
|
|
for (cursor = 0; cursor < size; cursor++)
|
|
if (!isprint(console_c[cursor]) && !isspace(console_c[cursor])
|
|
&& !BIOS_LOG_IS_MARKER(console_c[cursor]))
|
|
console_c[cursor] = '?';
|
|
|
|
/* We detect the reboot cutoff by looking for a bootblock, romstage or
|
|
ramstage banner, in that order (to account for platforms without
|
|
CONFIG_BOOTBLOCK_CONSOLE and/or CONFIG_EARLY_CONSOLE). Once we find
|
|
a banner, store the last two matches for that stage and stop. */
|
|
cursor = previous = 0;
|
|
if (type != CONSOLE_PRINT_FULL) {
|
|
#define BANNER_REGEX(stage) \
|
|
"\n\n.?coreboot-[^\n]* " stage " starting.*\\.\\.\\.\n"
|
|
#define OVERFLOW_REGEX(stage) "\n.?\\*\\*\\* Pre-CBMEM " stage " console overflow"
|
|
const char *regex[] = { BANNER_REGEX("verstage-before-bootblock"),
|
|
BANNER_REGEX("bootblock"),
|
|
BANNER_REGEX("verstage"),
|
|
OVERFLOW_REGEX("romstage"),
|
|
BANNER_REGEX("romstage"),
|
|
OVERFLOW_REGEX("ramstage"),
|
|
BANNER_REGEX("ramstage") };
|
|
|
|
for (size_t i = 0; !cursor && i < ARRAY_SIZE(regex); i++) {
|
|
regex_t re;
|
|
regmatch_t match;
|
|
int res = regcomp(&re, regex[i], REG_EXTENDED | REG_NEWLINE);
|
|
assert(res == 0);
|
|
|
|
/* Keep looking for matches so we find the last one. */
|
|
while (!regexec(&re, console_c + cursor, 1, &match, 0)) {
|
|
previous = cursor;
|
|
cursor += match.rm_so + 1;
|
|
}
|
|
regfree(&re);
|
|
}
|
|
}
|
|
|
|
if (type == CONSOLE_PRINT_PREVIOUS) {
|
|
console_c[cursor] = '\0';
|
|
cursor = previous;
|
|
}
|
|
|
|
char c;
|
|
int suppressed = 0;
|
|
int tty = isatty(fileno(stdout));
|
|
while ((c = console_c[cursor++])) {
|
|
if (BIOS_LOG_IS_MARKER(c)) {
|
|
int lvl = BIOS_LOG_MARKER_TO_LEVEL(c);
|
|
if (lvl > max_loglevel) {
|
|
suppressed = 1;
|
|
continue;
|
|
}
|
|
suppressed = 0;
|
|
if (tty)
|
|
printf(BIOS_LOG_ESCAPE_PATTERN, bios_log_escape[lvl]);
|
|
printf(BIOS_LOG_PREFIX_PATTERN, bios_log_prefix[lvl]);
|
|
} else {
|
|
if (!suppressed)
|
|
putchar(c);
|
|
if (c == '\n') {
|
|
if (tty && !suppressed)
|
|
printf(BIOS_LOG_ESCAPE_RESET);
|
|
suppressed = !print_unknown_logs;
|
|
}
|
|
}
|
|
}
|
|
if (tty)
|
|
printf(BIOS_LOG_ESCAPE_RESET);
|
|
|
|
free(console_c);
|
|
unmap_memory(&console_mapping);
|
|
}
|
|
|
|
static void hexdump(unsigned long memory, int length)
|
|
{
|
|
int i;
|
|
const uint8_t *m;
|
|
int all_zero = 0;
|
|
struct mapping hexdump_mapping;
|
|
|
|
m = map_memory(&hexdump_mapping, memory, length);
|
|
if (!m)
|
|
die("Unable to map hexdump memory.\n");
|
|
|
|
for (i = 0; i < length; i += 16) {
|
|
int j;
|
|
|
|
all_zero++;
|
|
for (j = 0; j < 16; j++) {
|
|
if(m[i+j] != 0) {
|
|
all_zero = 0;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (all_zero < 2) {
|
|
printf("%08lx:", memory + i);
|
|
for (j = 0; j < 16; j++)
|
|
printf(" %02x", m[i+j]);
|
|
printf(" ");
|
|
for (j = 0; j < 16; j++)
|
|
printf("%c", isprint(m[i+j]) ? m[i+j] : '.');
|
|
printf("\n");
|
|
} else if (all_zero == 2) {
|
|
printf("...\n");
|
|
}
|
|
}
|
|
|
|
unmap_memory(&hexdump_mapping);
|
|
}
|
|
|
|
static void dump_cbmem_hex(void)
|
|
{
|
|
if (cbmem.type != LB_MEM_TABLE) {
|
|
fprintf(stderr, "No coreboot CBMEM area found!\n");
|
|
return;
|
|
}
|
|
|
|
hexdump(cbmem.start, cbmem.size);
|
|
}
|
|
|
|
static void rawdump(uint64_t base, uint64_t size)
|
|
{
|
|
const uint8_t *m;
|
|
struct mapping dump_mapping;
|
|
|
|
m = map_memory(&dump_mapping, base, size);
|
|
if (!m)
|
|
die("Unable to map rawdump memory\n");
|
|
|
|
for (uint64_t i = 0 ; i < size; i++)
|
|
printf("%c", m[i]);
|
|
|
|
unmap_memory(&dump_mapping);
|
|
}
|
|
|
|
static void dump_cbmem_raw(unsigned int id)
|
|
{
|
|
const uint8_t *table;
|
|
size_t offset;
|
|
uint64_t base = 0;
|
|
uint64_t size = 0;
|
|
|
|
table = mapping_virt(&lbtable_mapping);
|
|
|
|
if (table == NULL)
|
|
return;
|
|
|
|
offset = 0;
|
|
|
|
while (offset < mapping_size(&lbtable_mapping)) {
|
|
const struct lb_record *lbr;
|
|
struct lb_cbmem_entry lbe;
|
|
|
|
lbr = (const void *)(table + offset);
|
|
offset += lbr->size;
|
|
|
|
if (lbr->tag != LB_TAG_CBMEM_ENTRY)
|
|
continue;
|
|
|
|
aligned_memcpy(&lbe, lbr, sizeof(lbe));
|
|
if (lbe.id == id) {
|
|
debug("found id for raw dump %0x", lbe.id);
|
|
base = lbe.address;
|
|
size = lbe.entry_size;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (!base)
|
|
fprintf(stderr, "id %0x not found in cbtable\n", id);
|
|
else
|
|
rawdump(base, size);
|
|
}
|
|
|
|
struct cbmem_id_to_name {
|
|
uint32_t id;
|
|
const char *name;
|
|
};
|
|
static const struct cbmem_id_to_name cbmem_ids[] = { CBMEM_ID_TO_NAME_TABLE };
|
|
|
|
#define MAX_STAGEx 10
|
|
static void cbmem_print_entry(int n, uint32_t id, uint64_t base, uint64_t size)
|
|
{
|
|
const char *name;
|
|
char stage_x[20];
|
|
|
|
name = NULL;
|
|
for (size_t i = 0; i < ARRAY_SIZE(cbmem_ids); i++) {
|
|
if (cbmem_ids[i].id == id) {
|
|
name = cbmem_ids[i].name;
|
|
break;
|
|
}
|
|
if (id >= CBMEM_ID_STAGEx_META &&
|
|
id < CBMEM_ID_STAGEx_META + MAX_STAGEx) {
|
|
snprintf(stage_x, sizeof(stage_x), "STAGE%d META",
|
|
(id - CBMEM_ID_STAGEx_META));
|
|
name = stage_x;
|
|
}
|
|
if (id >= CBMEM_ID_STAGEx_CACHE &&
|
|
id < CBMEM_ID_STAGEx_CACHE + MAX_STAGEx) {
|
|
snprintf(stage_x, sizeof(stage_x), "STAGE%d $ ",
|
|
(id - CBMEM_ID_STAGEx_CACHE));
|
|
name = stage_x;
|
|
}
|
|
}
|
|
|
|
printf("%2d. ", n);
|
|
if (name == NULL)
|
|
name = "(unknown)";
|
|
printf("%-20s %08x", name, id);
|
|
printf(" %08" PRIx64 " ", base);
|
|
printf(" %08" PRIx64 "\n", size);
|
|
}
|
|
|
|
static void dump_cbmem_toc(void)
|
|
{
|
|
int i;
|
|
const uint8_t *table;
|
|
size_t offset;
|
|
|
|
table = mapping_virt(&lbtable_mapping);
|
|
|
|
if (table == NULL)
|
|
return;
|
|
|
|
printf("CBMEM table of contents:\n");
|
|
printf(" %-20s %-8s %-8s %-8s\n", "NAME", "ID", "START",
|
|
"LENGTH");
|
|
|
|
i = 0;
|
|
offset = 0;
|
|
|
|
while (offset < mapping_size(&lbtable_mapping)) {
|
|
const struct lb_record *lbr;
|
|
struct lb_cbmem_entry lbe;
|
|
|
|
lbr = (const void *)(table + offset);
|
|
offset += lbr->size;
|
|
|
|
if (lbr->tag != LB_TAG_CBMEM_ENTRY)
|
|
continue;
|
|
|
|
aligned_memcpy(&lbe, lbr, sizeof(lbe));
|
|
cbmem_print_entry(i, lbe.id, lbe.address, lbe.entry_size);
|
|
i++;
|
|
}
|
|
}
|
|
|
|
#define COVERAGE_MAGIC 0x584d4153
|
|
struct file {
|
|
uint32_t magic;
|
|
uint32_t next;
|
|
uint32_t filename;
|
|
uint32_t data;
|
|
int offset;
|
|
int len;
|
|
};
|
|
|
|
static int mkpath(char *path, mode_t mode)
|
|
{
|
|
assert (path && *path);
|
|
char *p;
|
|
for (p = strchr(path+1, '/'); p; p = strchr(p + 1, '/')) {
|
|
*p = '\0';
|
|
if (mkdir(path, mode) == -1) {
|
|
if (errno != EEXIST) {
|
|
*p = '/';
|
|
return -1;
|
|
}
|
|
}
|
|
*p = '/';
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static void dump_coverage(void)
|
|
{
|
|
uint64_t start;
|
|
size_t size;
|
|
const void *coverage;
|
|
struct mapping coverage_mapping;
|
|
unsigned long phys_offset;
|
|
#define phys_to_virt(x) ((void *)(unsigned long)(x) + phys_offset)
|
|
|
|
if (find_cbmem_entry(CBMEM_ID_COVERAGE, &start, &size)) {
|
|
fprintf(stderr, "No coverage information found\n");
|
|
return;
|
|
}
|
|
|
|
/* Map coverage area */
|
|
coverage = map_memory(&coverage_mapping, start, size);
|
|
if (!coverage)
|
|
die("Unable to map coverage area.\n");
|
|
phys_offset = (unsigned long)coverage - (unsigned long)start;
|
|
|
|
printf("Dumping coverage data...\n");
|
|
|
|
struct file *file = (struct file *)coverage;
|
|
while (file && file->magic == COVERAGE_MAGIC) {
|
|
FILE *f;
|
|
char *filename;
|
|
|
|
debug(" -> %s\n", (char *)phys_to_virt(file->filename));
|
|
filename = strdup((char *)phys_to_virt(file->filename));
|
|
if (mkpath(filename, 0755) == -1) {
|
|
perror("Directory for coverage data could "
|
|
"not be created");
|
|
exit(1);
|
|
}
|
|
f = fopen(filename, "wb");
|
|
if (!f) {
|
|
printf("Could not open %s: %s\n",
|
|
filename, strerror(errno));
|
|
exit(1);
|
|
}
|
|
if (fwrite((void *)phys_to_virt(file->data),
|
|
file->len, 1, f) != 1) {
|
|
printf("Could not write to %s: %s\n",
|
|
filename, strerror(errno));
|
|
exit(1);
|
|
}
|
|
fclose(f);
|
|
free(filename);
|
|
|
|
if (file->next)
|
|
file = (struct file *)phys_to_virt(file->next);
|
|
else
|
|
file = NULL;
|
|
}
|
|
unmap_memory(&coverage_mapping);
|
|
}
|
|
|
|
static void print_version(void)
|
|
{
|
|
printf("cbmem v%s -- ", CBMEM_VERSION);
|
|
printf("Copyright (C) 2012 The ChromiumOS Authors. All rights reserved.\n\n");
|
|
printf(
|
|
"This program is free software: you can redistribute it and/or modify\n"
|
|
"it under the terms of the GNU General Public License as published by\n"
|
|
"the Free Software Foundation, version 2 of the License.\n\n"
|
|
"This program is distributed in the hope that it will be useful,\n"
|
|
"but WITHOUT ANY WARRANTY; without even the implied warranty of\n"
|
|
"MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n"
|
|
"GNU General Public License for more details.\n\n");
|
|
}
|
|
|
|
static void print_usage(const char *name, int exit_code)
|
|
{
|
|
printf("usage: %s [-cCltTLxVvh?]\n", name);
|
|
printf("\n"
|
|
" -c | --console: print cbmem console\n"
|
|
" -1 | --oneboot: print cbmem console for last boot only\n"
|
|
" -2 | --2ndtolast: print cbmem console for the boot that came before the last one only\n"
|
|
" -B | --loglevel: maximum loglevel to print; prefix `+` (e.g. -B +INFO) to also print lines that have no level\n"
|
|
" -C | --coverage: dump coverage information\n"
|
|
" -l | --list: print cbmem table of contents\n"
|
|
" -x | --hexdump: print hexdump of cbmem area\n"
|
|
" -r | --rawdump ID: print rawdump of specific ID (in hex) of cbtable\n"
|
|
" -t | --timestamps: print timestamp information\n"
|
|
" -T | --parseable-timestamps: print parseable timestamps\n"
|
|
" -S | --stacked-timestamps: print stacked timestamps (e.g. for flame graph tools)\n"
|
|
" -a | --add-timestamp ID: append timestamp with ID\n"
|
|
" -L | --tcpa-log print TPM log\n"
|
|
" -V | --verbose: verbose (debugging) output\n"
|
|
" -v | --version: print the version\n"
|
|
" -h | --help: print this help\n"
|
|
"\n");
|
|
exit(exit_code);
|
|
}
|
|
|
|
#if defined(__arm__) || defined(__aarch64__)
|
|
static void dt_update_cells(const char *name, int *addr_cells_ptr,
|
|
int *size_cells_ptr)
|
|
{
|
|
if (*addr_cells_ptr >= 0 && *size_cells_ptr >= 0)
|
|
return;
|
|
|
|
int buffer;
|
|
size_t nlen = strlen(name);
|
|
char *prop = alloca(nlen + sizeof("/#address-cells"));
|
|
strcpy(prop, name);
|
|
|
|
if (*addr_cells_ptr < 0) {
|
|
strcpy(prop + nlen, "/#address-cells");
|
|
int fd = open(prop, O_RDONLY);
|
|
if (fd < 0 && errno != ENOENT) {
|
|
perror(prop);
|
|
} else if (fd >= 0) {
|
|
if (read(fd, &buffer, sizeof(int)) < 0)
|
|
perror(prop);
|
|
else
|
|
*addr_cells_ptr = ntohl(buffer);
|
|
close(fd);
|
|
}
|
|
}
|
|
|
|
if (*size_cells_ptr < 0) {
|
|
strcpy(prop + nlen, "/#size-cells");
|
|
int fd = open(prop, O_RDONLY);
|
|
if (fd < 0 && errno != ENOENT) {
|
|
perror(prop);
|
|
} else if (fd >= 0) {
|
|
if (read(fd, &buffer, sizeof(int)) < 0)
|
|
perror(prop);
|
|
else
|
|
*size_cells_ptr = ntohl(buffer);
|
|
close(fd);
|
|
}
|
|
}
|
|
}
|
|
|
|
static char *dt_find_compat(const char *parent, const char *compat,
|
|
int *addr_cells_ptr, int *size_cells_ptr)
|
|
{
|
|
char *ret = NULL;
|
|
struct dirent *entry;
|
|
DIR *dir;
|
|
|
|
if (!(dir = opendir(parent))) {
|
|
perror(parent);
|
|
return NULL;
|
|
}
|
|
|
|
/* Loop through all files in the directory (DT node). */
|
|
while ((entry = readdir(dir))) {
|
|
/* We only care about compatible props or subnodes. */
|
|
if (entry->d_name[0] == '.' || !((entry->d_type & DT_DIR) ||
|
|
!strcmp(entry->d_name, "compatible")))
|
|
continue;
|
|
|
|
/* Assemble the file name (on the stack, for speed). */
|
|
size_t plen = strlen(parent);
|
|
char *name = alloca(plen + strlen(entry->d_name) + 2);
|
|
|
|
strcpy(name, parent);
|
|
name[plen] = '/';
|
|
strcpy(name + plen + 1, entry->d_name);
|
|
|
|
/* If it's a subnode, recurse. */
|
|
if (entry->d_type & DT_DIR) {
|
|
ret = dt_find_compat(name, compat, addr_cells_ptr,
|
|
size_cells_ptr);
|
|
|
|
/* There is only one matching node to find, abort. */
|
|
if (ret) {
|
|
/* Gather cells values on the way up. */
|
|
dt_update_cells(parent, addr_cells_ptr,
|
|
size_cells_ptr);
|
|
break;
|
|
}
|
|
continue;
|
|
}
|
|
|
|
/* If it's a compatible string, see if it's the right one. */
|
|
int fd = open(name, O_RDONLY);
|
|
int clen = strlen(compat);
|
|
char *buffer = alloca(clen + 1);
|
|
|
|
if (fd < 0) {
|
|
perror(name);
|
|
continue;
|
|
}
|
|
|
|
if (read(fd, buffer, clen + 1) < 0) {
|
|
perror(name);
|
|
close(fd);
|
|
continue;
|
|
}
|
|
close(fd);
|
|
|
|
if (!strcmp(compat, buffer)) {
|
|
/* Initialize these to "unset" for the way up. */
|
|
*addr_cells_ptr = *size_cells_ptr = -1;
|
|
|
|
/* Can't leave string on the stack or we'll lose it! */
|
|
ret = strdup(parent);
|
|
break;
|
|
}
|
|
}
|
|
|
|
closedir(dir);
|
|
return ret;
|
|
}
|
|
#endif /* defined(__arm__) || defined(__aarch64__) */
|
|
|
|
int main(int argc, char** argv)
|
|
{
|
|
int print_defaults = 1;
|
|
int print_console = 0;
|
|
int print_coverage = 0;
|
|
int print_list = 0;
|
|
int print_hexdump = 0;
|
|
int print_rawdump = 0;
|
|
int print_tcpa_log = 0;
|
|
enum timestamps_print_type timestamp_type = TIMESTAMPS_PRINT_NONE;
|
|
enum console_print_type console_type = CONSOLE_PRINT_FULL;
|
|
unsigned int rawdump_id = 0;
|
|
int max_loglevel = BIOS_NEVER;
|
|
int print_unknown_logs = 1;
|
|
uint32_t timestamp_id = 0;
|
|
|
|
int opt, option_index = 0;
|
|
static struct option long_options[] = {
|
|
{"console", 0, 0, 'c'},
|
|
{"oneboot", 0, 0, '1'},
|
|
{"2ndtolast", 0, 0, '2'},
|
|
{"loglevel", required_argument, 0, 'B'},
|
|
{"coverage", 0, 0, 'C'},
|
|
{"list", 0, 0, 'l'},
|
|
{"tcpa-log", 0, 0, 'L'},
|
|
{"timestamps", 0, 0, 't'},
|
|
{"parseable-timestamps", 0, 0, 'T'},
|
|
{"stacked-timestamps", 0, 0, 'S'},
|
|
{"add-timestamp", required_argument, 0, 'a'},
|
|
{"hexdump", 0, 0, 'x'},
|
|
{"rawdump", required_argument, 0, 'r'},
|
|
{"verbose", 0, 0, 'V'},
|
|
{"version", 0, 0, 'v'},
|
|
{"help", 0, 0, 'h'},
|
|
{0, 0, 0, 0}
|
|
};
|
|
while ((opt = getopt_long(argc, argv, "c12B:CltTSa:LxVvh?r:",
|
|
long_options, &option_index)) != EOF) {
|
|
switch (opt) {
|
|
case 'c':
|
|
print_console = 1;
|
|
print_defaults = 0;
|
|
break;
|
|
case '1':
|
|
print_console = 1;
|
|
console_type = CONSOLE_PRINT_LAST;
|
|
print_defaults = 0;
|
|
break;
|
|
case '2':
|
|
print_console = 1;
|
|
console_type = CONSOLE_PRINT_PREVIOUS;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'B':
|
|
max_loglevel = parse_loglevel(optarg, &print_unknown_logs);
|
|
break;
|
|
case 'C':
|
|
print_coverage = 1;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'l':
|
|
print_list = 1;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'L':
|
|
print_tcpa_log = 1;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'x':
|
|
print_hexdump = 1;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'r':
|
|
print_rawdump = 1;
|
|
print_defaults = 0;
|
|
rawdump_id = strtoul(optarg, NULL, 16);
|
|
break;
|
|
case 't':
|
|
timestamp_type = TIMESTAMPS_PRINT_NORMAL;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'T':
|
|
timestamp_type = TIMESTAMPS_PRINT_MACHINE_READABLE;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'S':
|
|
timestamp_type = TIMESTAMPS_PRINT_STACKED;
|
|
print_defaults = 0;
|
|
break;
|
|
case 'a':
|
|
print_defaults = 0;
|
|
timestamp_id = timestamp_enum_name_to_id(optarg);
|
|
/* Parse numeric value if name is unknown */
|
|
if (timestamp_id == 0)
|
|
timestamp_id = strtoul(optarg, NULL, 0);
|
|
break;
|
|
case 'V':
|
|
verbose = 1;
|
|
break;
|
|
case 'v':
|
|
print_version();
|
|
exit(0);
|
|
break;
|
|
case 'h':
|
|
print_usage(argv[0], 0);
|
|
break;
|
|
case '?':
|
|
default:
|
|
print_usage(argv[0], 1);
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (optind < argc) {
|
|
fprintf(stderr, "Error: Extra parameter found.\n");
|
|
print_usage(argv[0], 1);
|
|
}
|
|
|
|
mem_fd = open("/dev/mem", timestamp_id ? O_RDWR : O_RDONLY, 0);
|
|
if (mem_fd < 0) {
|
|
fprintf(stderr, "Failed to gain memory access: %s\n",
|
|
strerror(errno));
|
|
return 1;
|
|
}
|
|
|
|
#if defined(__arm__) || defined(__aarch64__)
|
|
int addr_cells, size_cells;
|
|
char *coreboot_node = dt_find_compat("/proc/device-tree", "coreboot",
|
|
&addr_cells, &size_cells);
|
|
|
|
if (!coreboot_node) {
|
|
fprintf(stderr, "Could not find 'coreboot' compatible node!\n");
|
|
return 1;
|
|
}
|
|
|
|
if (addr_cells < 0) {
|
|
fprintf(stderr, "Warning: no #address-cells node in tree!\n");
|
|
addr_cells = 1;
|
|
}
|
|
|
|
int nlen = strlen(coreboot_node);
|
|
char *reg = alloca(nlen + sizeof("/reg"));
|
|
|
|
strcpy(reg, coreboot_node);
|
|
strcpy(reg + nlen, "/reg");
|
|
free(coreboot_node);
|
|
|
|
int fd = open(reg, O_RDONLY);
|
|
if (fd < 0) {
|
|
perror(reg);
|
|
return 1;
|
|
}
|
|
|
|
int i;
|
|
size_t size_to_read = addr_cells * 4 + size_cells * 4;
|
|
u8 *dtbuffer = alloca(size_to_read);
|
|
if (read(fd, dtbuffer, size_to_read) < 0) {
|
|
perror(reg);
|
|
return 1;
|
|
}
|
|
close(fd);
|
|
|
|
/* No variable-length byte swap function anywhere in C... how sad. */
|
|
u64 baseaddr = 0;
|
|
for (i = 0; i < addr_cells * 4; i++) {
|
|
baseaddr <<= 8;
|
|
baseaddr |= *dtbuffer;
|
|
dtbuffer++;
|
|
}
|
|
u64 cb_table_size = 0;
|
|
for (i = 0; i < size_cells * 4; i++) {
|
|
cb_table_size <<= 8;
|
|
cb_table_size |= *dtbuffer;
|
|
dtbuffer++;
|
|
}
|
|
|
|
parse_cbtable(baseaddr, cb_table_size);
|
|
#else
|
|
unsigned long long possible_base_addresses[] = { 0, 0xf0000 };
|
|
|
|
/* Find and parse coreboot table */
|
|
for (size_t j = 0; j < ARRAY_SIZE(possible_base_addresses); j++) {
|
|
if (!parse_cbtable(possible_base_addresses[j], 0))
|
|
break;
|
|
}
|
|
#endif
|
|
|
|
if (mapping_virt(&lbtable_mapping) == NULL)
|
|
die("Table not found.\n");
|
|
|
|
if (print_console)
|
|
dump_console(console_type, max_loglevel, print_unknown_logs);
|
|
|
|
if (print_coverage)
|
|
dump_coverage();
|
|
|
|
if (print_list)
|
|
dump_cbmem_toc();
|
|
|
|
if (print_hexdump)
|
|
dump_cbmem_hex();
|
|
|
|
if (print_rawdump)
|
|
dump_cbmem_raw(rawdump_id);
|
|
|
|
if (timestamp_id)
|
|
timestamp_add_now(timestamp_id);
|
|
|
|
if (print_defaults)
|
|
timestamp_type = TIMESTAMPS_PRINT_NORMAL;
|
|
|
|
if (timestamp_type != TIMESTAMPS_PRINT_NONE)
|
|
dump_timestamps(timestamp_type);
|
|
|
|
if (print_tcpa_log)
|
|
dump_tpm_log();
|
|
|
|
unmap_memory(&lbtable_mapping);
|
|
|
|
close(mem_fd);
|
|
return 0;
|
|
}
|