summaryrefslogtreecommitdiffstats
path: root/liblog/log_read_kern.c
diff options
context:
space:
mode:
authorMark Salyzyn <salyzyn@google.com>2014-02-20 14:59:07 -0800
committerMark Salyzyn <salyzyn@google.com>2014-02-28 13:49:11 -0800
commit154f4608aac6218af0e25c98b71d0803278c047e (patch)
treed324e05b44b362cbc0a95033ad1c06a0438bd523 /liblog/log_read_kern.c
parent1114f1806521b2a6447b7c68934e4f3c29b60cb5 (diff)
downloadsystem_core-154f4608aac6218af0e25c98b71d0803278c047e.zip
system_core-154f4608aac6218af0e25c98b71d0803278c047e.tar.gz
system_core-154f4608aac6218af0e25c98b71d0803278c047e.tar.bz2
liblog: enable logging to logd.
* Modify liblog to send all messages to the new syslog user space daemon. Original-Change-Id: I0ce439738cd921efb2db4c1d6a289a96bdbc8bc2 Original-Change-Id: If4eb0d09409f7e9be3eb4bb7017073dc7e931ab4 Signed-off-by: Nick Kralevich <nnk@google.com> * Add a TARGET_USES_LOGD make flag for BoardConfig.mk to manage whether logd is enabled for use or not. * rename syslog to logd to avert confusion with bionic syslog * Add fake log support back in * prefilter for logging messages from logd * Fill in timestamps at logging source * update abstract log reader * switch from using suffix for id to v3 format * log a message when creating devices that a deprecated interface is being utilized. Signed-off-by: Mark Salyzyn <salyzyn@google.com> (cherry pick from commit 099e2c1f6f706a8600c1cef74cce9066fc315480) Change-Id: I47929a5432977a1d7235267a435cec0a7d6bd440
Diffstat (limited to 'liblog/log_read_kern.c')
-rw-r--r--liblog/log_read_kern.c695
1 files changed, 695 insertions, 0 deletions
diff --git a/liblog/log_read_kern.c b/liblog/log_read_kern.c
new file mode 100644
index 0000000..b454729
--- /dev/null
+++ b/liblog/log_read_kern.c
@@ -0,0 +1,695 @@
+/*
+** Copyright 2013-2014, The Android Open Source Project
+**
+** Licensed under the Apache License, Version 2.0 (the "License");
+** you may not use this file except in compliance with the License.
+** You may obtain a copy of the License at
+**
+** http://www.apache.org/licenses/LICENSE-2.0
+**
+** Unless required by applicable law or agreed to in writing, software
+** distributed under the License is distributed on an "AS IS" BASIS,
+** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+** See the License for the specific language governing permissions and
+** limitations under the License.
+*/
+
+#define _GNU_SOURCE /* asprintf for x86 host */
+#include <errno.h>
+#include <fcntl.h>
+#include <poll.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <cutils/list.h>
+#include <log/log.h>
+#include <log/logger.h>
+
+#include <sys/ioctl.h>
+
+#define __LOGGERIO 0xAE
+
+#define LOGGER_GET_LOG_BUF_SIZE _IO(__LOGGERIO, 1) /* size of log */
+#define LOGGER_GET_LOG_LEN _IO(__LOGGERIO, 2) /* used log len */
+#define LOGGER_GET_NEXT_ENTRY_LEN _IO(__LOGGERIO, 3) /* next entry len */
+#define LOGGER_FLUSH_LOG _IO(__LOGGERIO, 4) /* flush log */
+#define LOGGER_GET_VERSION _IO(__LOGGERIO, 5) /* abi version */
+#define LOGGER_SET_VERSION _IO(__LOGGERIO, 6) /* abi version */
+
+typedef char bool;
+#define false (const bool)0
+#define true (const bool)1
+
+#define LOG_FILE_DIR "/dev/log/"
+
+/* timeout in milliseconds */
+#define LOG_TIMEOUT_FLUSH 5
+#define LOG_TIMEOUT_NEVER -1
+
+#define logger_for_each(logger, logger_list) \
+ for (logger = node_to_item((logger_list)->node.next, struct logger, node); \
+ logger != node_to_item(&(logger_list)->node, struct logger, node); \
+ logger = node_to_item((logger)->node.next, struct logger, node))
+
+/* In the future, we would like to make this list extensible */
+static const char *LOG_NAME[LOG_ID_MAX] = {
+ [LOG_ID_MAIN] = "main",
+ [LOG_ID_RADIO] = "radio",
+ [LOG_ID_EVENTS] = "events",
+ [LOG_ID_SYSTEM] = "system"
+};
+
+const char *android_log_id_to_name(log_id_t log_id)
+{
+ if (log_id >= LOG_ID_MAX) {
+ log_id = LOG_ID_MAIN;
+ }
+ return LOG_NAME[log_id];
+}
+
+static int accessmode(int mode)
+{
+ if ((mode & O_ACCMODE) == O_WRONLY) {
+ return W_OK;
+ }
+ if ((mode & O_ACCMODE) == O_RDWR) {
+ return R_OK | W_OK;
+ }
+ return R_OK;
+}
+
+/* repeated fragment */
+static int check_allocate_accessible(char **n, const char *b, int mode)
+{
+ *n = NULL;
+
+ if (!b) {
+ return -EINVAL;
+ }
+
+ asprintf(n, LOG_FILE_DIR "%s", b);
+ if (!*n) {
+ return -1;
+ }
+
+ return access(*n, accessmode(mode));
+}
+
+log_id_t android_name_to_log_id(const char *logName)
+{
+ const char *b;
+ char *n;
+ int ret;
+
+ if (!logName) {
+ return -1; /* NB: log_id_t is unsigned */
+ }
+ b = strrchr(logName, '/');
+ if (!b) {
+ b = logName;
+ } else {
+ ++b;
+ }
+
+ ret = check_allocate_accessible(&n, b, O_RDONLY);
+ free(n);
+ if (ret) {
+ return ret;
+ }
+
+ for(ret = LOG_ID_MIN; ret < LOG_ID_MAX; ++ret) {
+ const char *l = LOG_NAME[ret];
+ if (l && !strcmp(b, l)) {
+ return ret;
+ }
+ }
+ return -1; /* should never happen */
+}
+
+struct logger_list {
+ struct listnode node;
+ int mode;
+ unsigned int tail;
+ pid_t pid;
+ unsigned int queued_lines;
+ int timeout_ms;
+ int error;
+ bool flush;
+ bool valid_entry; /* valiant(?) effort to deal with memory starvation */
+ struct log_msg entry;
+};
+
+struct log_list {
+ struct listnode node;
+ struct log_msg entry; /* Truncated to event->len() + 1 to save space */
+};
+
+struct logger {
+ struct listnode node;
+ struct logger_list *top;
+ int fd;
+ log_id_t id;
+ short *revents;
+ struct listnode log_list;
+};
+
+/* android_logger_alloc unimplemented, no use case */
+/* android_logger_free not exported */
+static void android_logger_free(struct logger *logger)
+{
+ if (!logger) {
+ return;
+ }
+
+ while (!list_empty(&logger->log_list)) {
+ struct log_list *entry = node_to_item(
+ list_head(&logger->log_list), struct log_list, node);
+ list_remove(&entry->node);
+ free(entry);
+ if (logger->top->queued_lines) {
+ logger->top->queued_lines--;
+ }
+ }
+
+ if (logger->fd >= 0) {
+ close(logger->fd);
+ }
+
+ list_remove(&logger->node);
+
+ free(logger);
+}
+
+log_id_t android_logger_get_id(struct logger *logger)
+{
+ return logger->id;
+}
+
+/* worker for sending the command to the logger */
+static int logger_ioctl(struct logger *logger, int cmd, int mode)
+{
+ char *n;
+ int f, ret;
+
+ if (!logger || !logger->top) {
+ return -EFAULT;
+ }
+
+ if (((mode & O_ACCMODE) == O_RDWR)
+ || (((mode ^ logger->top->mode) & O_ACCMODE) == 0)) {
+ return ioctl(logger->fd, cmd);
+ }
+
+ /* We go here if android_logger_list_open got mode wrong for this ioctl */
+ ret = check_allocate_accessible(&n, android_log_id_to_name(logger->id), mode);
+ if (ret) {
+ free(n);
+ return ret;
+ }
+
+ f = open(n, mode);
+ free(n);
+ if (f < 0) {
+ return f;
+ }
+
+ ret = ioctl(f, cmd);
+ close (f);
+
+ return ret;
+}
+
+int android_logger_clear(struct logger *logger)
+{
+ return logger_ioctl(logger, LOGGER_FLUSH_LOG, O_WRONLY);
+}
+
+/* returns the total size of the log's ring buffer */
+int android_logger_get_log_size(struct logger *logger)
+{
+ return logger_ioctl(logger, LOGGER_GET_LOG_BUF_SIZE, O_RDWR);
+}
+
+/*
+ * returns the readable size of the log's ring buffer (that is, amount of the
+ * log consumed)
+ */
+int android_logger_get_log_readable_size(struct logger *logger)
+{
+ return logger_ioctl(logger, LOGGER_GET_LOG_LEN, O_RDONLY);
+}
+
+/*
+ * returns the logger version
+ */
+int android_logger_get_log_version(struct logger *logger)
+{
+ int ret = logger_ioctl(logger, LOGGER_GET_VERSION, O_RDWR);
+ return (ret < 0) ? 1 : ret;
+}
+
+struct logger_list *android_logger_list_alloc(int mode,
+ unsigned int tail,
+ pid_t pid)
+{
+ struct logger_list *logger_list;
+
+ logger_list = calloc(1, sizeof(*logger_list));
+ if (!logger_list) {
+ return NULL;
+ }
+ list_init(&logger_list->node);
+ logger_list->mode = mode;
+ logger_list->tail = tail;
+ logger_list->pid = pid;
+ return logger_list;
+}
+
+/* android_logger_list_register unimplemented, no use case */
+/* android_logger_list_unregister unimplemented, no use case */
+
+/* Open the named log and add it to the logger list */
+struct logger *android_logger_open(struct logger_list *logger_list,
+ log_id_t id)
+{
+ struct listnode *node;
+ struct logger *logger;
+ char *n;
+
+ if (!logger_list || (id >= LOG_ID_MAX)) {
+ goto err;
+ }
+
+ logger_for_each(logger, logger_list) {
+ if (logger->id == id) {
+ goto ok;
+ }
+ }
+
+ logger = calloc(1, sizeof(*logger));
+ if (!logger) {
+ goto err;
+ }
+
+ if (check_allocate_accessible(&n, android_log_id_to_name(id),
+ logger_list->mode)) {
+ goto err_name;
+ }
+
+ logger->fd = open(n, logger_list->mode);
+ if (logger->fd < 0) {
+ goto err_name;
+ }
+
+ free(n);
+ logger->id = id;
+ list_init(&logger->log_list);
+ list_add_tail(&logger_list->node, &logger->node);
+ logger->top = logger_list;
+ logger_list->timeout_ms = LOG_TIMEOUT_FLUSH;
+ goto ok;
+
+err_name:
+ free(n);
+err_logger:
+ free(logger);
+err:
+ logger = NULL;
+ok:
+ return logger;
+}
+
+/* Open the single named log and make it part of a new logger list */
+struct logger_list *android_logger_list_open(log_id_t id,
+ int mode,
+ unsigned int tail,
+ pid_t pid)
+{
+ struct logger_list *logger_list = android_logger_list_alloc(mode, tail, pid);
+ if (!logger_list) {
+ return NULL;
+ }
+
+ if (!android_logger_open(logger_list, id)) {
+ android_logger_list_free(logger_list);
+ return NULL;
+ }
+
+ return logger_list;
+}
+
+/* prevent memory starvation when backfilling */
+static unsigned int queue_threshold(struct logger_list *logger_list)
+{
+ return (logger_list->tail < 64) ? 64 : logger_list->tail;
+}
+
+static bool low_queue(struct listnode *node)
+{
+ /* low is considered less than 2 */
+ return list_head(node) == list_tail(node);
+}
+
+/* Flush queues in sequential order, one at a time */
+static int android_logger_list_flush(struct logger_list *logger_list,
+ struct log_msg *log_msg)
+{
+ int ret = 0;
+ struct log_list *firstentry = NULL;
+
+ while ((ret == 0)
+ && (logger_list->flush
+ || (logger_list->queued_lines > logger_list->tail))) {
+ struct logger *logger;
+
+ /* Merge sort */
+ bool at_least_one_is_low = false;
+ struct logger *firstlogger = NULL;
+ firstentry = NULL;
+
+ logger_for_each(logger, logger_list) {
+ struct listnode *node;
+ struct log_list *oldest = NULL;
+
+ /* kernel logger channels not necessarily time-sort order */
+ list_for_each(node, &logger->log_list) {
+ struct log_list *entry = node_to_item(node,
+ struct log_list, node);
+ if (!oldest
+ || (entry->entry.entry.sec < oldest->entry.entry.sec)
+ || ((entry->entry.entry.sec == oldest->entry.entry.sec)
+ && (entry->entry.entry.nsec < oldest->entry.entry.nsec))) {
+ oldest = entry;
+ }
+ }
+
+ if (!oldest) {
+ at_least_one_is_low = true;
+ continue;
+ } else if (low_queue(&logger->log_list)) {
+ at_least_one_is_low = true;
+ }
+
+ if (!firstentry
+ || (oldest->entry.entry.sec < firstentry->entry.entry.sec)
+ || ((oldest->entry.entry.sec == firstentry->entry.entry.sec)
+ && (oldest->entry.entry.nsec < firstentry->entry.entry.nsec))) {
+ firstentry = oldest;
+ firstlogger = logger;
+ }
+ }
+
+ if (!firstentry) {
+ break;
+ }
+
+ /* when trimming list, tries to keep one entry behind in each bucket */
+ if (!logger_list->flush
+ && at_least_one_is_low
+ && (logger_list->queued_lines < queue_threshold(logger_list))) {
+ break;
+ }
+
+ /* within tail?, send! */
+ if ((logger_list->tail == 0)
+ || (logger_list->queued_lines <= logger_list->tail)) {
+ int diff;
+ ret = firstentry->entry.entry.hdr_size;
+ if (!ret) {
+ ret = sizeof(firstentry->entry.entry_v1);
+ }
+
+ /* Promote entry to v3 format */
+ memcpy(log_msg->buf, firstentry->entry.buf, ret);
+ diff = sizeof(firstentry->entry.entry_v3) - ret;
+ if (diff < 0) {
+ diff = 0;
+ } else if (diff > 0) {
+ memset(log_msg->buf + ret, 0, diff);
+ }
+ memcpy(log_msg->buf + ret + diff, firstentry->entry.buf + ret,
+ firstentry->entry.entry.len + 1);
+ ret += diff;
+ log_msg->entry.hdr_size = ret;
+ log_msg->entry.lid = firstlogger->id;
+
+ ret += firstentry->entry.entry.len;
+ }
+
+ /* next entry */
+ list_remove(&firstentry->node);
+ free(firstentry);
+ if (logger_list->queued_lines) {
+ logger_list->queued_lines--;
+ }
+ }
+
+ /* Flushed the list, no longer in tail mode for continuing content */
+ if (logger_list->flush && !firstentry) {
+ logger_list->tail = 0;
+ }
+ return ret;
+}
+
+/* Read from the selected logs */
+int android_logger_list_read(struct logger_list *logger_list,
+ struct log_msg *log_msg)
+{
+ struct logger *logger;
+ nfds_t nfds;
+ struct pollfd *p, *pollfds = NULL;
+ int error = 0, ret = 0;
+
+ memset(log_msg, 0, sizeof(struct log_msg));
+
+ if (!logger_list) {
+ return -ENODEV;
+ }
+
+ if (!(accessmode(logger_list->mode) & R_OK)) {
+ logger_list->error = EPERM;
+ goto done;
+ }
+
+ nfds = 0;
+ logger_for_each(logger, logger_list) {
+ ++nfds;
+ }
+ if (nfds <= 0) {
+ error = ENODEV;
+ goto done;
+ }
+
+ /* Do we have anything to offer from the buffer or state? */
+ if (logger_list->valid_entry) { /* implies we are also in a flush state */
+ goto flush;
+ }
+
+ ret = android_logger_list_flush(logger_list, log_msg);
+ if (ret) {
+ goto done;
+ }
+
+ if (logger_list->error) { /* implies we are also in a flush state */
+ goto done;
+ }
+
+ /* Lets start grinding on metal */
+ pollfds = calloc(nfds, sizeof(struct pollfd));
+ if (!pollfds) {
+ error = ENOMEM;
+ goto flush;
+ }
+
+ p = pollfds;
+ logger_for_each(logger, logger_list) {
+ p->fd = logger->fd;
+ p->events = POLLIN;
+ logger->revents = &p->revents;
+ ++p;
+ }
+
+ while (!ret && !error) {
+ int result;
+
+ /* If we oversleep it's ok, i.e. ignore EINTR. */
+ result = TEMP_FAILURE_RETRY(
+ poll(pollfds, nfds, logger_list->timeout_ms));
+
+ if (result <= 0) {
+ if (result) {
+ error = errno;
+ } else if (logger_list->mode & O_NDELAY) {
+ error = EAGAIN;
+ } else {
+ logger_list->timeout_ms = LOG_TIMEOUT_NEVER;
+ }
+
+ logger_list->flush = true;
+ goto try_flush;
+ }
+
+ logger_list->timeout_ms = LOG_TIMEOUT_FLUSH;
+
+ /* Anti starvation */
+ if (!logger_list->flush
+ && (logger_list->queued_lines > (queue_threshold(logger_list) / 2))) {
+ /* Any queues with input pending that is low? */
+ bool starving = false;
+ logger_for_each(logger, logger_list) {
+ if ((*(logger->revents) & POLLIN)
+ && low_queue(&logger->log_list)) {
+ starving = true;
+ break;
+ }
+ }
+
+ /* pushback on any queues that are not low */
+ if (starving) {
+ logger_for_each(logger, logger_list) {
+ if ((*(logger->revents) & POLLIN)
+ && !low_queue(&logger->log_list)) {
+ *(logger->revents) &= ~POLLIN;
+ }
+ }
+ }
+ }
+
+ logger_for_each(logger, logger_list) {
+ unsigned int hdr_size;
+ struct log_list *entry;
+ int diff;
+
+ if (!(*(logger->revents) & POLLIN)) {
+ continue;
+ }
+
+ memset(logger_list->entry.buf, 0, sizeof(struct log_msg));
+ /* NOTE: driver guarantees we read exactly one full entry */
+ result = read(logger->fd, logger_list->entry.buf,
+ LOGGER_ENTRY_MAX_LEN);
+ if (result <= 0) {
+ if (!result) {
+ error = EIO;
+ } else if (errno != EINTR) {
+ error = errno;
+ }
+ continue;
+ }
+
+ if (logger_list->pid
+ && (logger_list->pid != logger_list->entry.entry.pid)) {
+ continue;
+ }
+
+ hdr_size = logger_list->entry.entry.hdr_size;
+ if (!hdr_size) {
+ hdr_size = sizeof(logger_list->entry.entry_v1);
+ }
+
+ if ((hdr_size > sizeof(struct log_msg))
+ || (logger_list->entry.entry.len
+ > sizeof(logger_list->entry.buf) - hdr_size)
+ || (logger_list->entry.entry.len != result - hdr_size)) {
+ error = EINVAL;
+ continue;
+ }
+
+ /* Promote entry to v3 format */
+ diff = sizeof(logger_list->entry.entry_v3) - hdr_size;
+ if (diff > 0) {
+ if (logger_list->entry.entry.len
+ > sizeof(logger_list->entry.buf) - hdr_size - diff) {
+ error = EINVAL;
+ continue;
+ }
+ result += diff;
+ memmove(logger_list->entry.buf + hdr_size + diff,
+ logger_list->entry.buf + hdr_size,
+ logger_list->entry.entry.len + 1);
+ memset(logger_list->entry.buf + hdr_size, 0, diff);
+ logger_list->entry.entry.hdr_size = hdr_size + diff;
+ }
+ logger_list->entry.entry.lid = logger->id;
+
+ /* speedup: If not tail, and only one list, send directly */
+ if (!logger_list->tail
+ && (list_head(&logger_list->node)
+ == list_tail(&logger_list->node))) {
+ ret = result;
+ memcpy(log_msg->buf, logger_list->entry.buf, result + 1);
+ break;
+ }
+
+ entry = malloc(sizeof(*entry) - sizeof(entry->entry) + result + 1);
+
+ if (!entry) {
+ logger_list->valid_entry = true;
+ error = ENOMEM;
+ break;
+ }
+
+ logger_list->queued_lines++;
+
+ memcpy(entry->entry.buf, logger_list->entry.buf, result);
+ entry->entry.buf[result] = '\0';
+ list_add_tail(&logger->log_list, &entry->node);
+ }
+
+ if (ret <= 0) {
+try_flush:
+ ret = android_logger_list_flush(logger_list, log_msg);
+ }
+ }
+
+ free(pollfds);
+
+flush:
+ if (error) {
+ logger_list->flush = true;
+ }
+
+ if (ret <= 0) {
+ ret = android_logger_list_flush(logger_list, log_msg);
+
+ if (!ret && logger_list->valid_entry) {
+ ret = logger_list->entry.entry.hdr_size;
+ if (!ret) {
+ ret = sizeof(logger_list->entry.entry_v1);
+ }
+ ret += logger_list->entry.entry.len;
+
+ memcpy(log_msg->buf, logger_list->entry.buf,
+ sizeof(struct log_msg));
+ logger_list->valid_entry = false;
+ }
+ }
+
+done:
+ if (logger_list->error) {
+ error = logger_list->error;
+ }
+ if (error) {
+ logger_list->error = error;
+ if (!ret) {
+ ret = -error;
+ }
+ }
+ return ret;
+}
+
+/* Close all the logs */
+void android_logger_list_free(struct logger_list *logger_list)
+{
+ if (logger_list == NULL) {
+ return;
+ }
+
+ while (!list_empty(&logger_list->node)) {
+ struct listnode *node = list_head(&logger_list->node);
+ struct logger *logger = node_to_item(node, struct logger, node);
+ android_logger_free(logger);
+ }
+
+ free(logger_list);
+}