Jetpack/kernel/nvidia/drivers/firmware/tegra/fwdebug.c

1087 lines
23 KiB
C
Raw Normal View History

/*
* Copyright (c) 2015-2018, NVIDIA CORPORATION. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <linux/debugfs.h>
#include <linux/firmware.h>
#include <linux/module.h>
#include <linux/of.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
#include <soc/tegra/bpmp_abi.h>
#include <soc/tegra/chip-id.h>
#include <soc/tegra/tegra_bpmp.h>
#include "bpmp.h"
#define BPMP_MODULE_MAGIC 0x646f6d
static struct device *device;
extern const struct of_device_id bpmp_of_matches[];
struct seqbuf {
char *buf;
size_t pos;
size_t size;
};
static void seqbuf_init(struct seqbuf *seqbuf, void *buf, size_t size)
{
seqbuf->buf = buf;
seqbuf->size = size;
seqbuf->pos = 0;
}
static size_t seqbuf_avail(struct seqbuf *seqbuf)
{
return seqbuf->pos < seqbuf->size ? seqbuf->size - seqbuf->pos : 0;
}
static size_t seqbuf_status(struct seqbuf *seqbuf)
{
return seqbuf->pos <= seqbuf->size ? 0 : -EOVERFLOW;
}
static int seqbuf_eof(struct seqbuf *seqbuf)
{
return seqbuf->pos >= seqbuf->size;
}
static int seqbuf_read(struct seqbuf *seqbuf, void *buf, size_t nbyte)
{
nbyte = min(nbyte, seqbuf_avail(seqbuf));
memcpy(buf, seqbuf->buf + seqbuf->pos,
min(nbyte, seqbuf_avail(seqbuf)));
seqbuf->pos += nbyte;
return seqbuf_status(seqbuf);
}
static int seqbuf_read_u32(struct seqbuf *seqbuf, u32 *v)
{
int err;
err = seqbuf_read(seqbuf, v, 4);
*v = le32_to_cpu(*v);
return err;
}
static const char *seqbuf_strget(struct seqbuf *seqbuf)
{
const char *ptr = seqbuf->buf + seqbuf->pos;
seqbuf->pos += strnlen(seqbuf->buf + seqbuf->pos, seqbuf_avail(seqbuf));
seqbuf->pos++;
if (seqbuf_status(seqbuf))
return NULL;
else
return ptr;
}
static int seqbuf_seek(struct seqbuf *seqbuf, ssize_t offset)
{
seqbuf->pos += offset;
return seqbuf_status(seqbuf);
}
static const char *root_path;
static const char *get_filename(const struct file *file, char *buf, int size)
{
const char *filename;
size_t root_len;
if (!root_path)
return NULL;
root_len = strlen(root_path);
filename = dentry_path(file->f_path.dentry, buf, size);
if (IS_ERR_OR_NULL(filename))
return NULL;
if (strlen(filename) < root_len ||
strncmp(filename, root_path, root_len))
return NULL;
filename += root_len;
return filename;
}
static int bpmp_debugfs_read(uint32_t name, uint32_t sz_name,
dma_addr_t data, size_t sz_data, uint32_t *nbytes)
{
struct mrq_debugfs_request rq;
struct mrq_debugfs_response re;
int r;
rq.cmd = (uint32_t)cpu_to_le32(CMD_DEBUGFS_READ);
rq.fop.fnameaddr = (uint32_t)cpu_to_le32(name);
rq.fop.fnamelen = (uint32_t)cpu_to_le32(sz_name);
rq.fop.dataaddr = (uint32_t)cpu_to_le32(data);
rq.fop.datalen = (uint32_t)cpu_to_le32(sz_data);
r = tegra_bpmp_send_receive(MRQ_DEBUGFS, &rq, sizeof(rq),
&re, sizeof(re));
if (r)
return r;
*nbytes = re.fop.nbytes;
return 0;
}
static int debugfs_show(struct seq_file *m, void *p)
{
struct file *file = m->private;
const size_t namesize = SZ_256;
char *databuf = NULL;
char *namebuf = NULL;
dma_addr_t dataphys;
dma_addr_t namephys;
const char *filename;
size_t off;
uint32_t nbytes;
int len;
int ret;
namebuf = tegra_bpmp_alloc_coherent(namesize, &namephys, GFP_KERNEL);
if (!namebuf)
return -ENOMEM;
filename = get_filename(file, namebuf, namesize);
if (!filename) {
ret = -ENOENT;
goto out;
}
databuf = tegra_bpmp_alloc_coherent(m->size, &dataphys, GFP_KERNEL);
if (!databuf) {
ret = -ENOMEM;
goto out;
}
off = filename - namebuf;
len = strlen(filename);
ret = bpmp_debugfs_read(namephys + off, len, dataphys,
m->size, &nbytes);
if (!ret)
seq_write(m, databuf, nbytes);
out:
tegra_bpmp_free_coherent(namesize, namebuf, namephys);
if (databuf)
tegra_bpmp_free_coherent(m->size, databuf, dataphys);
return ret;
}
static int debugfs_open(struct inode *inode, struct file *file)
{
return single_open_size(file, debugfs_show, file, SZ_256K);
}
static int bpmp_debugfs_write(uint32_t name, size_t sz_name,
uint32_t data, size_t sz_data)
{
struct mrq_debugfs_request rq;
rq.cmd = (uint32_t)cpu_to_le32(CMD_DEBUGFS_WRITE);
rq.fop.fnameaddr = (uint32_t)cpu_to_le32(name);
rq.fop.fnamelen = (uint32_t)cpu_to_le32(sz_name);
rq.fop.dataaddr = (uint32_t)cpu_to_le32(data);
rq.fop.datalen = (uint32_t)cpu_to_le32(sz_data);
return tegra_bpmp_send_receive(MRQ_DEBUGFS, &rq, sizeof(rq),
NULL, 0);
}
static ssize_t debugfs_store(struct file *file, const char __user *buf,
size_t count, loff_t *f_pos)
{
const size_t namesize = SZ_256;
char *databuf = NULL;
char *namebuf = NULL;
const char *filename;
ssize_t ret;
dma_addr_t phys_data;
dma_addr_t phys_name;
size_t off;
int len;
databuf = tegra_bpmp_alloc_coherent(count, &phys_data, GFP_KERNEL);
if (!databuf)
return -ENOMEM;
namebuf = tegra_bpmp_alloc_coherent(namesize, &phys_name, GFP_KERNEL);
if (!namebuf) {
ret = -ENOMEM;
goto out;
}
if (copy_from_user(databuf, buf, count)) {
ret = -EFAULT;
goto out;
}
filename = get_filename(file, namebuf, namesize);
if (!filename) {
ret = -EFAULT;
goto out;
}
off = filename - namebuf;
len = strlen(filename);
ret = bpmp_debugfs_write(phys_name + off, len, phys_data, count);
out:
tegra_bpmp_free_coherent(namesize, namebuf, phys_name);
if (databuf)
tegra_bpmp_free_coherent(count, databuf, phys_data);
return ret ?: count;
}
static const struct file_operations debugfs_fops = {
.open = debugfs_open,
.read = seq_read,
.llseek = seq_lseek,
.write = debugfs_store,
.release = single_release,
};
static int bpmp_populate_dir(struct seqbuf *seqbuf, struct dentry *parent,
u32 depth)
{
int err;
while (!seqbuf_eof(seqbuf)) {
u32 d, t;
const char *name;
struct dentry *dentry;
seqbuf_read_u32(seqbuf, &d);
if (d < depth) {
seqbuf_seek(seqbuf, -4);
/* go up a level */
return 0;
}
seqbuf_read_u32(seqbuf, &t);
name = seqbuf_strget(seqbuf);
if (seqbuf_status(seqbuf))
return seqbuf_status(seqbuf);
if (d != depth) {
/* malformed data received from BPMP */
return -EIO;
}
if (t & DEBUGFS_S_ISDIR) {
dentry = debugfs_create_dir(name, parent);
if (IS_ERR_OR_NULL(dentry))
return dentry ? PTR_ERR(dentry) : -ENOMEM;
err = bpmp_populate_dir(seqbuf, dentry, depth+1);
if (err)
return err;
} else {
umode_t mode;
mode = t & DEBUGFS_S_IRUSR ? S_IRUSR : 0;
mode |= t & DEBUGFS_S_IWUSR ? S_IWUSR : 0;
dentry = debugfs_create_file(name, mode,
parent, NULL,
&debugfs_fops);
if (IS_ERR_OR_NULL(dentry))
return -ENOMEM;
}
}
return 0;
}
static DEFINE_MUTEX(lock);
static struct dentry *bpmp_debugfs_root;
static char root_path_buf[256];
static int bpmp_fwdebug_create(void *buf, size_t bufsize, struct dentry *root)
{
struct seqbuf seqbuf;
int err;
bpmp_debugfs_root = debugfs_create_dir("debug", root);
if (IS_ERR_OR_NULL(bpmp_debugfs_root)) {
pr_err("failed to create bpmp debugfs directory\n");
bpmp_debugfs_root = NULL;
return -ENOMEM;
}
root_path = dentry_path_raw(bpmp_debugfs_root, root_path_buf,
sizeof(root_path_buf));
if (IS_ERR_OR_NULL(root_path)) {
/* if this happens, then to recover bpmp debugfs needs
* to be unmounted from userspace */
pr_err("failed to figure out bpmp root path\n");
err = root_path ? PTR_ERR(root_path) : -ENOENT;
root_path = NULL;
return err;
}
seqbuf_init(&seqbuf, buf, bufsize);
err = bpmp_populate_dir(&seqbuf, bpmp_debugfs_root, 0);
return err;
}
static int bpmp_debugfs_dumpdir(uint32_t addr, size_t size, uint32_t *nbytes)
{
struct mrq_debugfs_request rq;
struct mrq_debugfs_response re;
int r;
rq.cmd = (uint32_t)cpu_to_le32(CMD_DEBUGFS_DUMPDIR);
rq.dumpdir.dataaddr = (uint32_t)cpu_to_le32(addr);
rq.dumpdir.datalen = (uint32_t)cpu_to_le32(size);
r = tegra_bpmp_send_receive(MRQ_DEBUGFS, &rq, sizeof(rq),
&re, sizeof(re));
if (r)
return r;
*nbytes = re.dumpdir.nbytes;
return 0;
}
static void do_debugfs_unmount(struct work_struct *work)
{
debugfs_remove_recursive(bpmp_debugfs_root);
bpmp_debugfs_root = NULL;
}
static DECLARE_WORK(debugfs_unmount_work, do_debugfs_unmount);
static int bpmp_mrq_is_supported(int mrq)
{
struct mrq_query_abi_request rq;
struct mrq_query_abi_response re;
int r;
rq.mrq = mrq;
r = tegra_bpmp_send_receive(MRQ_QUERY_ABI, &rq, sizeof(rq),
&re, sizeof(re));
/* something went wrong; assume not supported */
if (WARN_ON(r))
return 0;
return re.status ? 0 : 1;
}
static int bpmp_fwdebug_init(struct dentry *root)
{
dma_addr_t phys;
void *virt;
const int sz = SZ_256K;
uint32_t nbytes;
int ret;
if (!root)
return -EINVAL;
if (!bpmp_mrq_is_supported(MRQ_DEBUGFS))
return 0;
mutex_lock(&lock);
if (bpmp_debugfs_root) {
mutex_unlock(&lock);
return -EINVAL;
}
virt = tegra_bpmp_alloc_coherent(sz, &phys, GFP_KERNEL);
if (!virt) {
pr_err("%s: dma_alloc_coherent() failed\n", __func__);
mutex_unlock(&lock);
return -ENOMEM;
}
ret = bpmp_debugfs_dumpdir(phys, sz, &nbytes);
if (ret) {
pr_err("bpmp_debugfs_dumpdir() failed (%d)\n", ret);
goto out;
}
cancel_work_sync(&debugfs_unmount_work);
ret = bpmp_fwdebug_create(virt, nbytes, root);
if (ret) {
pr_err("create_bpmp_debugfs() failed (%d)\n", ret);
goto out;
}
pr_info("bpmp: mounted debugfs mirror\n");
out:
tegra_bpmp_free_coherent(sz, virt, phys);
mutex_unlock(&lock);
return ret;
}
static int bpmp_fwdebug_uninit(struct dentry *root)
{
mutex_lock(&lock);
if (!bpmp_debugfs_root) {
mutex_unlock(&lock);
return -EINVAL;
}
schedule_work(&debugfs_unmount_work);
mutex_unlock(&lock);
return 0;
}
static struct dentry *bpmp_root;
static struct dentry *module_root;
static LIST_HEAD(modules);
static DEFINE_MUTEX(bpmp_lock);
struct bpmp_module {
struct list_head entry;
struct work_struct unload_work;
char name[MODULE_NAME_LEN];
struct dentry *root;
u32 handle;
u32 size;
};
struct module_hdr {
u32 magic;
u32 size;
u32 reloc_size;
u32 bss_size;
u32 init_offset;
u32 cleanup_offset;
u8 reserved[72];
u8 parent_tag[32];
};
int bpmp_create_attrs(const struct fops_entry *fent,
struct dentry *parent, void *data)
{
struct dentry *d;
while (fent->name) {
d = debugfs_create_file(fent->name, fent->mode, parent, data,
fent->fops);
if (IS_ERR_OR_NULL(d))
return -EFAULT;
fent++;
}
return 0;
}
static struct bpmp_module *bpmp_find_module(const char *name)
{
struct bpmp_module *m;
list_for_each_entry(m, &modules, entry) {
if (!strncmp(m->name, name, MODULE_NAME_LEN))
return m;
}
return NULL;
}
static int bpmp_module_load(struct device *dev, const void *base, u32 size,
u32 *handle)
{
void *virt;
dma_addr_t phys;
struct { u32 phys; u32 size; } __packed msg;
int r;
virt = tegra_bpmp_alloc_coherent(size, &phys, GFP_KERNEL);
if (virt == NULL)
return -ENOMEM;
memcpy(virt, base, size);
msg.phys = phys;
msg.size = size;
r = tegra_bpmp_send_receive(MRQ_MODULE_LOAD, &msg, sizeof(msg),
handle, sizeof(*handle));
tegra_bpmp_free_coherent(size, virt, phys);
return r;
}
static int bpmp_module_unload(struct device *dev, u32 handle)
{
return tegra_bpmp_send_receive(MRQ_MODULE_UNLOAD,
&handle, sizeof(handle), NULL, 0);
}
static void do_unload_module(struct work_struct *w)
{
struct bpmp_module *m;
int err;
m = container_of(w, struct bpmp_module, unload_work);
if (m->handle) {
err = bpmp_module_unload(device, m->handle);
if (err) {
dev_err(device, "%s: failed to unload module (%d)\n",
m->name, err);
return;
}
}
debugfs_remove_recursive(m->root);
kfree(m);
}
static ssize_t bpmp_module_unload_store(struct file *file,
const char __user *user_buf, size_t count, loff_t *ppos)
{
struct bpmp_module *m;
char buf[MODULE_NAME_LEN];
char *name;
if (count >= sizeof(buf))
return -EINVAL;
if (strncpy_from_user(buf, user_buf, count) <= 0)
return -EFAULT;
buf[count] = 0;
name = strim(buf);
mutex_lock(&bpmp_lock);
m = bpmp_find_module(name);
if (!m) {
mutex_unlock(&bpmp_lock);
return -ENODEV;
}
list_del(&m->entry);
schedule_work(&m->unload_work);
mutex_unlock(&bpmp_lock);
return count;
}
static const struct file_operations bpmp_module_unload_fops = {
.write = bpmp_module_unload_store
};
static int bpmp_module_ready(const char *name, const struct firmware *fw,
struct bpmp_module *m)
{
struct module_hdr *hdr;
const int sz = sizeof(firmware_tag);
char fmt[sz + 1];
int err;
hdr = (struct module_hdr *)fw->data;
if (fw->size < sizeof(struct module_hdr) ||
hdr->magic != BPMP_MODULE_MAGIC ||
hdr->size + hdr->reloc_size != fw->size) {
dev_err(device, "%s: invalid module format\n", name);
return -EINVAL;
}
if (memcmp(hdr->parent_tag, firmware_tag, sz)) {
dev_err(device, "%s: bad module - tag mismatch\n", name);
memcpy(fmt, firmware_tag, sz);
fmt[sz] = 0;
dev_err(device, "firmware: %s\n", fmt);
memcpy(fmt, hdr->parent_tag, sz);
fmt[sz] = 0;
dev_err(device, "%s : %s\n", name, fmt);
return -EINVAL;
}
m->size = hdr->size + hdr->bss_size;
err = bpmp_module_load(device, fw->data, fw->size, &m->handle);
if (err) {
dev_err(device, "failed to load module, code=%d\n", err);
return err;
}
if (!debugfs_create_x32("handle", S_IRUGO, m->root, &m->handle))
return -ENOMEM;
if (!debugfs_create_x32("size", S_IRUGO, m->root, &m->size))
return -ENOMEM;
list_add_tail(&m->entry, &modules);
return 0;
}
static ssize_t bpmp_module_load_store(struct file *file,
const char __user *user_buf, size_t count, loff_t *ppos)
{
const struct firmware *fw;
struct bpmp_module *m;
char buf[MODULE_NAME_LEN];
int r;
if (count >= sizeof(buf))
return -EINVAL;
if (strncpy_from_user(buf, user_buf, count) <= 0)
return -EFAULT;
buf[count] = 0;
m = kzalloc(sizeof(*m), GFP_KERNEL);
if (m == NULL)
return -ENOMEM;
mutex_lock(&bpmp_lock);
strncpy(m->name, strim(buf), sizeof(m->name) - 1);
m->name[sizeof(m->name) - 1] = 0;
INIT_WORK(&m->unload_work, do_unload_module);
if (bpmp_find_module(m->name)) {
dev_err(device, "module %s exist\n", m->name);
r = -EEXIST;
goto clean;
}
m->root = debugfs_create_dir(m->name, module_root);
if (!m->root) {
r = -ENOMEM;
goto clean;
}
r = request_firmware(&fw, m->name, device);
if (r) {
dev_err(device, "request_firmware() failed: %d\n", r);
goto clean;
}
if (!fw) {
r = -EFAULT;
WARN_ON(0);
goto clean;
}
dev_info(device, "%s: module ready %zu@%p\n",
m->name, fw->size, fw->data);
r = bpmp_module_ready(m->name, fw, m);
release_firmware(fw);
clean:
mutex_unlock(&bpmp_lock);
if (r) {
schedule_work(&m->unload_work);
return r;
}
return count;
}
static const struct file_operations bpmp_module_load_fops = {
.write = bpmp_module_load_store
};
static int bpmp_init_modules(struct platform_device *pdev,
struct dentry *parent)
{
const struct fops_entry mod_attrs[] = {
{ "load", &bpmp_module_load_fops, S_IWUSR },
{ "unload", &bpmp_module_unload_fops, S_IWUSR },
{ NULL, NULL, 0 }
};
module_root = debugfs_create_dir("module", parent);
if (IS_ERR_OR_NULL(module_root))
goto clean;
if (bpmp_create_attrs(mod_attrs, module_root, pdev))
goto clean;
return 0;
clean:
WARN_ON(1);
debugfs_remove_recursive(module_root);
module_root = NULL;
return -EFAULT;
}
static int bpmp_ping_show(void *data, u64 *val)
{
unsigned long flags;
ktime_t tm;
int ret;
local_irq_save(flags);
tm = ktime_get();
ret = __bpmp_do_ping();
tm = ktime_sub(ktime_get(), tm);
local_irq_restore(flags);
*val = ret ?: ktime_to_us(tm);
return 0;
}
static int bpmp_modify_trace_mask(uint32_t clr, uint32_t set)
{
uint32_t mb[] = { clr, set };
uint32_t new;
return tegra_bpmp_send_receive(MRQ_TRACE_MODIFY, mb, sizeof(mb),
&new, sizeof(new)) ?: new;
}
static int bpmp_trace_enable_show(void *data, u64 *val)
{
*val = bpmp_modify_trace_mask(0, 0);
return 0;
}
static int bpmp_trace_enable_store(void *data, u64 val)
{
int r;
r = bpmp_modify_trace_mask(0, val);
if (r < 0)
return r;
return 0;
}
static int bpmp_trace_disable_store(void *data, u64 val)
{
int r;
r = bpmp_modify_trace_mask(val, 0);
if (r < 0)
return r;
return 0;
}
static int bpmp_mount_show(void *data, u64 *val)
{
*val = bpmp_fwdebug_init(bpmp_root);
return 0;
}
static int bpmp_unmount_show(void *data, u64 *val)
{
*val = bpmp_fwdebug_uninit(bpmp_root);
return 0;
}
DEFINE_SIMPLE_ATTRIBUTE(bpmp_ping_fops, bpmp_ping_show, NULL, "%lld\n");
DEFINE_SIMPLE_ATTRIBUTE(trace_enable_fops, bpmp_trace_enable_show,
bpmp_trace_enable_store, "0x%llx\n");
DEFINE_SIMPLE_ATTRIBUTE(trace_disable_fops, NULL,
bpmp_trace_disable_store, "0x%llx\n");
DEFINE_SIMPLE_ATTRIBUTE(bpmp_mount_fops, bpmp_mount_show, NULL, "%lld\n");
DEFINE_SIMPLE_ATTRIBUTE(bpmp_unmount_fops, bpmp_unmount_show, NULL, "%lld\n");
#ifdef CONFIG_BPMP_DEBUGFS_MOUNT_ON_BOOT
static __init int bpmp_init_mount(void)
{
struct device_node *np = NULL;
/* mirroring takes a while */
if (!tegra_platform_is_silicon())
return 0;
/* continue with the init only if the bpmp node is active in the DTB */
np = of_find_matching_node(NULL, bpmp_of_matches);
if (!np || !of_device_is_available(np))
return 0;
return bpmp_fwdebug_init(bpmp_root);
}
late_initcall(bpmp_init_mount);
#endif
struct bpmp_trace_iter {
dma_addr_t phys;
void *virt;
int eof;
};
static int bpmp_trace_show(struct seq_file *file, void *v)
{
struct bpmp_trace_iter *i = file->private;
uint32_t mb[] = { i->phys, PAGE_SIZE };
int ret;
i->eof = 0;
ret = tegra_bpmp_send_receive(MRQ_WRITE_TRACE, mb, sizeof(mb),
&i->eof, sizeof(i->eof));
pr_debug("%s: ret %d eof %d\n", __func__, ret, i->eof);
if (ret < 0)
return ret;
ret = seq_write(file, i->virt, ret);
if (ret < 0)
return ret;
return 0;
}
static void *bpmp_trace_start(struct seq_file *file, loff_t *pos)
{
struct bpmp_trace_iter *i = file->private;
uint32_t cmd;
int first;
int ret;
first = (*pos == 0);
if (first && bpmp_mrq_is_supported(MRQ_TRACE_ITER)) {
cmd = 0;
ret = tegra_bpmp_send_receive(MRQ_TRACE_ITER,
&cmd, sizeof(cmd), NULL, 0);
if (WARN_ON(ret))
return NULL;
}
pr_debug("%s: first %d eof %d pos %llu\n",
__func__, first, i->eof, *pos);
if (!first && i->eof == 1)
return NULL;
i->virt = tegra_bpmp_alloc_coherent(PAGE_SIZE, &i->phys, GFP_KERNEL);
if (!i->virt)
return NULL;
return i;
}
static void *bpmp_trace_next(struct seq_file *file, void *v, loff_t *pos)
{
struct bpmp_trace_iter *i = file->private;
pr_debug("%s: eof %d pos %llu\n", __func__, i->eof, *pos);
return NULL;
}
static void bpmp_trace_stop(struct seq_file *file, void *v)
{
struct bpmp_trace_iter *i = file->private;
pr_debug("%s: eof %d\n", __func__, i->eof);
if (i->virt) {
tegra_bpmp_free_coherent(PAGE_SIZE, i->virt, i->phys);
i->virt = NULL;
}
}
static const struct seq_operations bpmp_trace_seq_ops = {
.start = bpmp_trace_start,
.show = bpmp_trace_show,
.next = bpmp_trace_next,
.stop = bpmp_trace_stop
};
static ssize_t bpmp_trace_store(struct file *file, const char __user *user_buf,
size_t count, loff_t *ppos)
{
uint32_t cmd = 1;
int ret;
if (!bpmp_mrq_is_supported(MRQ_TRACE_ITER))
return count;
ret = tegra_bpmp_send_receive(MRQ_TRACE_ITER,
&cmd, sizeof(cmd), NULL, 0);
return ret ?: count;
}
static int bpmp_trace_open(struct inode *inode, struct file *file)
{
return seq_open_private(file, &bpmp_trace_seq_ops,
sizeof(struct bpmp_trace_iter));
}
static const struct file_operations trace_fops = {
.open = bpmp_trace_open,
.read = seq_read,
.llseek = seq_lseek,
.write = bpmp_trace_store,
.release = seq_release_private,
};
static int bpmp_tag_show(struct seq_file *file, void *data)
{
seq_write(file, firmware_tag, sizeof(firmware_tag));
seq_putc(file, '\n');
return 0;
}
static int bpmp_tag_open(struct inode *inode, struct file *file)
{
return single_open(file, bpmp_tag_show, inode->i_private);
}
static const struct file_operations bpmp_tag_fops = {
.open = bpmp_tag_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release
};
#define MSG_NR_FIELDS ((MSG_DATA_MIN_SZ + 3) / 4)
#define MSG_DATA_COUNT (MSG_NR_FIELDS + 1)
static uint32_t inbox_data[MSG_DATA_COUNT];
static ssize_t bpmp_mrq_write(struct file *file, const char __user *user_buf,
size_t count, loff_t *ppos)
{
/* size in dec, space, new line, terminator */
char buf[MSG_DATA_COUNT * 11 + 1 + 1];
uint32_t outbox_data[MSG_DATA_COUNT];
char *line;
char *p;
int i;
int ret;
memset(outbox_data, 0, sizeof(outbox_data));
memset(inbox_data, 0, sizeof(inbox_data));
count = min(count, sizeof(buf) - 1);
if (copy_from_user(buf, user_buf, count)) {
ret = -EFAULT;
goto complete;
}
buf[count] = 0;
line = strim(buf);
for (i = 0; i < MSG_DATA_COUNT && line; i++) {
p = strsep(&line, " ");
ret = kstrtouint(p, 0, outbox_data + i);
if (ret)
break;
}
if (!i) {
ret = -EINVAL;
goto complete;
}
ret = tegra_bpmp_send_receive(outbox_data[0], outbox_data + 1,
MSG_DATA_MIN_SZ, inbox_data + 1, MSG_DATA_MIN_SZ);
complete:
inbox_data[0] = ret;
return ret ?: count;
}
static int bpmp_mrq_show(struct seq_file *file, void *data)
{
int i;
for (i = 0; i < MSG_DATA_COUNT; i++) {
seq_printf(file, "0x%x%s", inbox_data[i],
i == MSG_DATA_COUNT - 1 ? "\n" : " ");
}
return 0;
}
static int bpmp_mrq_open(struct inode *inode, struct file *file)
{
return single_open(file, bpmp_mrq_show, inode->i_private);
}
static const struct file_operations bpmp_mrq_fops = {
.open = bpmp_mrq_open,
.llseek = seq_lseek,
.read = seq_read,
.write = bpmp_mrq_write,
.release = single_release
};
static const struct fops_entry root_attrs[] = {
{ "ping", &bpmp_ping_fops, S_IRUGO },
{ "trace_enable", &trace_enable_fops, S_IRUGO | S_IWUSR },
{ "trace_disable", &trace_disable_fops, S_IWUSR },
{ "trace", &trace_fops, S_IRUGO | S_IWUSR },
{ "tag", &bpmp_tag_fops, S_IRUGO },
{ "mrq", &bpmp_mrq_fops, S_IRUGO | S_IWUSR },
{ "mount", &bpmp_mount_fops, S_IRUGO },
{ "unmount", &bpmp_unmount_fops, S_IRUGO },
{ NULL, NULL, 0 }
};
struct dentry *bpmp_init_debug(struct platform_device *pdev)
{
struct dentry *root;
root = debugfs_create_dir("bpmp", NULL);
if (IS_ERR_OR_NULL(root))
goto clean;
if (bpmp_create_attrs(root_attrs, root, pdev))
goto clean;
if (bpmp_init_modules(pdev, root))
goto clean;
device = &pdev->dev;
bpmp_root = root;
return root;
clean:
WARN_ON(1);
debugfs_remove_recursive(root);
return NULL;
}
struct dentry *tegra_bpmp_debugfs_add_file(char *name,
umode_t mode, void *data, const struct file_operations *fops)
{
if (!bpmp_root)
return NULL;
return debugfs_create_file(name, mode, bpmp_root, data, fops);
}