/*
* Find an empty file descriptor entry, and mark it busy.
*/
int get_unused_fd(void)
{
struct files_struct * files = current->files;
int fd, error;
struct fdtable *fdt;
if(need_files_checkpoint())
checkpoint_files();
error = -EMFILE;
spin_lock(&files->file_lock);
repeat:
fdt = files_fdtable(files);
fd = find_next_zero_bit(fdt->open_fds->fds_bits, fdt->max_fds,
files->next_fd);
/*
* N.B. For clone tasks sharing a files structure, this test
* will limit the total number of files that can be opened.
*/
if (fd >= current->signal->rlim[RLIMIT_NOFILE].rlim_cur)
goto out;
/* Do we need to expand the fd array or fd set? */
error = expand_files(files, fd);
if (error < 0)
goto out;
if (error) {
/*
* If we needed to expand the fs array we
* might have blocked - try again.
*/
error = -EMFILE;
goto repeat;
}
FD_SET(fd, fdt->open_fds);
FD_CLR(fd, fdt->close_on_exec);
files->next_fd = fd + 1;
#if 1
/* Sanity check */
if (fdt->fd[fd] != NULL) {
printk(KERN_WARNING "get_unused_fd: slot %d not NULL!\n", fd);
fdt->fd[fd] = NULL;
}
#endif
error = fd;
out:
spin_unlock(&files->file_lock);
return error;
}
static void autofs_dev_ioctl_fd_install(unsigned int fd, struct file *file)
{
struct files_struct *files = current->files;
struct fdtable *fdt;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
BUG_ON(fdt->fd[fd] != NULL);
rcu_assign_pointer(fdt->fd[fd], file);
__set_close_on_exec(fd, fdt);
spin_unlock(&files->file_lock);
}
void fd_install(unsigned int fd, struct file *file)
{
struct files_struct *files = current->files;
struct fdtable *fdt;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
BUG_ON(fdt->fd[fd] != NULL);
rcu_assign_pointer(fdt->fd[fd], file);
fdt->user[fd].installer = current->pid;
getnstimeofday(&fdt->user[fd].open_time);
spin_unlock(&files->file_lock);
}
static int seq_show(struct seq_file *m, void *v)
{
struct files_struct *files = NULL;
int f_flags = 0, ret = -ENOENT;
struct file *file = NULL;
struct task_struct *task;
task = get_proc_task(m->private);
if (!task)
return -ENOENT;
if (!gr_acl_handle_procpidmem(task))
files = get_files_struct(task);
put_task_struct(task);
if (files) {
int fd = proc_fd(m->private);
spin_lock(&files->file_lock);
file = fcheck_files(files, fd);
if (file) {
struct fdtable *fdt = files_fdtable(files);
f_flags = file->f_flags;
if (close_on_exec(fd, fdt))
f_flags |= O_CLOEXEC;
get_file(file);
ret = 0;
}
spin_unlock(&files->file_lock);
put_files_struct(files);
}
if (ret)
return ret;
seq_printf(m, "pos:\t%lli\nflags:\t0%o\nmnt_id:\t%i\n",
(long long)file->f_pos, f_flags,
real_mount(file->f_path.mnt)->mnt_id);
show_fd_locks(m, file, files);
if (seq_has_overflowed(m))
goto out;
if (file->f_op->show_fdinfo)
file->f_op->show_fdinfo(m, file);
out:
fput(file);
return 0;
}
static inline void task_state(struct seq_file *m, struct pid_namespace *ns,
struct pid *pid, struct task_struct *p)
{
struct group_info *group_info;
int g;
struct fdtable *fdt = NULL;
const struct cred *cred;
pid_t ppid, tpid;
rcu_read_lock();
ppid = pid_alive(p) ?
task_tgid_nr_ns(rcu_dereference(p->real_parent), ns) : 0;
tpid = 0;
if (pid_alive(p)) {
struct task_struct *tracer = ptrace_parent(p);
if (tracer)
tpid = task_pid_nr_ns(tracer, ns);
}
cred = get_task_cred(p);
seq_printf(m,
"State:\t%s\n"
"Tgid:\t%d\n"
"Pid:\t%d\n"
"PPid:\t%d\n"
"TracerPid:\t%d\n"
"Uid:\t%d\t%d\t%d\t%d\n"
"Gid:\t%d\t%d\t%d\t%d\n",
get_task_state(p),
task_tgid_nr_ns(p, ns),
pid_nr_ns(pid, ns),
ppid, tpid,
cred->uid, cred->euid, cred->suid, cred->fsuid,
cred->gid, cred->egid, cred->sgid, cred->fsgid);
task_lock(p);
if (p->files)
fdt = files_fdtable(p->files);
seq_printf(m,
"FDSize:\t%d\n"
"Groups:\t",
fdt ? fdt->max_fds : 0);
rcu_read_unlock();
group_info = cred->group_info;
task_unlock(p);
for (g = 0; g < min(group_info->ngroups, NGROUPS_SMALL); g++)
seq_printf(m, "%d ", GROUP_AT(group_info, g));
put_cred(cred);
seq_putc(m, '\n');
}
void fd_install(unsigned int fd, struct file *file)
{
struct files_struct *files = current->files;
struct fdtable *fdt;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
BUG_ON(fdt->fd[fd] != NULL);
rcu_assign_pointer(fdt->fd[fd], file);
file->record_pid = current->pid;
strncpy(file->record_comm, current->comm, 15);
file->record_comm[15] = '\0';
spin_unlock(&files->file_lock);
}
void fastcall fd_install(unsigned int fd, struct file * file)
{
struct files_struct *files = current->files;
struct fdtable *fdt;
if(need_files_checkpoint())
checkpoint_files();
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
BUG_ON(fdt->fd[fd] != NULL);
rcu_assign_pointer(fdt->fd[fd], file);
spin_unlock(&files->file_lock);
}
static void task_fd_install(
struct shfile_proc *proc, unsigned int fd, struct file *file)
{
struct files_struct *files = proc->files;
struct fdtable *fdt;
if (files == NULL)
return;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
BUG_ON(fdt->fd[fd] != NULL);
rcu_assign_pointer(fdt->fd[fd], file);
spin_unlock(&files->file_lock);
}
static void timod_queue(unsigned int fd, struct T_primsg *it)
{
struct sol_socket_struct *sock;
struct fdtable *fdt;
SOLD("queuing primsg");
fdt = files_fdtable(current->files);
sock = (struct sol_socket_struct *)fdt->fd[fd]->private_data;
it->next = sock->pfirst;
sock->pfirst = it;
if (!sock->plast)
sock->plast = it;
timod_wake_socket(fd);
SOLD("done");
}
void fdleak_debug_print(struct files_struct *files)
{
struct fdtable *fdt;
unsigned int n;
fdt = files_fdtable(files);
for(n=0; n < fdt->max_fds; n++){
if (rcu_dereference_raw(fdt->fd[n]) != NULL) {
printk("[TOO MANY OPEN FILES] (%d)/(%d) = %s\n",
n,
fdt->max_fds,
fdt->fd[n]->f_path.dentry->d_name.name);
}
}
}
static void timod_wake_socket(unsigned int fd)
{
struct socket *sock;
struct fdtable *fdt;
SOLD("wakeing socket");
fdt = files_fdtable(current->files);
sock = SOCKET_I(fdt->fd[fd]->f_dentry->d_inode);
wake_up_interruptible(&sock->wait);
read_lock(&sock->sk->sk_callback_lock);
if (sock->fasync_list && !test_bit(SOCK_ASYNC_WAITDATA, &sock->flags))
__kill_fasync(sock->fasync_list, SIGIO, POLL_IN);
read_unlock(&sock->sk->sk_callback_lock);
SOLD("done");
}
int mali_stream_create_fence(mali_sync_pt *pt)
{
struct sync_fence *fence;
struct fdtable * fdt;
struct files_struct * files;
int fd = -1;
fence = sync_fence_create("mali_fence", pt);
if (!fence)
{
sync_pt_free(pt);
fd = -EFAULT;
goto out;
}
/* create a fd representing the fence */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
fd = get_unused_fd_flags(O_CLOEXEC);
if (fd < 0)
{
sync_fence_put(fence);
goto out;
}
#else
fd = get_unused_fd();
if (fd < 0)
{
sync_fence_put(fence);
goto out;
}
files = current->files;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,4,0)
__set_close_on_exec(fd, fdt);
#else
FD_SET(fd, fdt->close_on_exec);
#endif
spin_unlock(&files->file_lock);
#endif /* Linux > 3.6 */
/* bind fence to the new fd */
sync_fence_install(fence, fd);
out:
return fd;
}
static void timod_queue_end(unsigned int fd, struct T_primsg *it)
{
struct sol_socket_struct *sock;
struct fdtable *fdt;
SOLD("queuing primsg at end");
fdt = files_fdtable(current->files);
sock = (struct sol_socket_struct *)fdt->fd[fd]->private_data;
it->next = NULL;
if (sock->plast)
sock->plast->next = it;
else
sock->pfirst = it;
sock->plast = it;
SOLD("done");
}
static inline char * task_state(struct task_struct *p, char *buffer)
{
struct group_info *group_info;
int g;
struct fdtable *fdt = NULL;
read_lock(&tasklist_lock);
buffer += sprintf(buffer,
"State:\t%s\n"
"SleepAVG:\t%lu%%\n"
"Tgid:\t%d\n"
"Pid:\t%d\n"
"PPid:\t%d\n"
"TracerPid:\t%d\n"
"Uid:\t%d\t%d\t%d\t%d\n"
"Gid:\t%d\t%d\t%d\t%d\n",
get_task_state(p),
(p->sleep_avg/1024)*100/(1020000000/1024),
p->tgid,
p->pid, pid_alive(p) ? p->group_leader->real_parent->tgid : 0,
pid_alive(p) && p->ptrace ? p->parent->pid : 0,
p->uid, p->euid, p->suid, p->fsuid,
p->gid, p->egid, p->sgid, p->fsgid);
read_unlock(&tasklist_lock);
task_lock(p);
rcu_read_lock();
if (p->files)
fdt = files_fdtable(p->files);
buffer += sprintf(buffer,
"FDSize:\t%d\n"
"Groups:\t",
fdt ? fdt->max_fds : 0);
rcu_read_unlock();
group_info = p->group_info;
get_group_info(group_info);
task_unlock(p);
for (g = 0; g < min(group_info->ngroups,NGROUPS_SMALL); g++)
buffer += sprintf(buffer, "%d ", GROUP_AT(group_info,g));
put_group_info(group_info);
buffer += sprintf(buffer, "\n");
return buffer;
}
static struct hone_event *__add_files(struct hone_reader *reader,
struct hone_event *event, struct task_struct *task)
{
struct hone_event *sk_event;
struct files_struct *files;
struct file *file;
struct fdtable *fdt;
struct socket *sock;
struct sock *sk;
unsigned long flags, set;
int i, fd;
if (!(files = get_files_struct(task)))
return event;
spin_lock_irqsave(&files->file_lock, flags);
if (!(fdt = files_fdtable(files)))
goto out;
for (i = 0; (fd = i * BITS_PER_LONG) < fdt->max_fds; i++) {
for (set = fdt->OPEN_FDS[i]; set; set >>= 1, fd++) {
if (!(set & 1))
continue;
file = fdt->fd[fd];
if (!file || file->f_op != &socket_file_ops || !file->private_data)
continue;
sock = file->private_data;
sk = sock->sk;
if (!sk || (sk->sk_family != PF_INET && sk->sk_family != PF_INET6))
continue;
if ((sk_event = __alloc_socket_event((unsigned long) sk,
0, task, GFP_ATOMIC))) {
sk_event->next = event;
event = sk_event;
memcpy(&event->ts, &task->start_time, sizeof(event->ts));
} else {
atomic64_inc(&reader->info.dropped.socket);
}
}
}
out:
spin_unlock_irqrestore(&files->file_lock, flags);
put_files_struct(files);
return event;
}
static inline void task_state(struct seq_file *m, struct pid *pid,
struct task_struct *p)
{
struct group_info *group_info;
int g;
struct fdtable *fdt = NULL;
rcu_read_lock();
seq_printf(m,
"State:\t%s\n"
"SleepAVG:\t%lu%%\n"
"Tgid:\t%d\n"
"Pid:\t%d\n"
"PPid:\t%d\n"
"TracerPid:\t%d\n"
"Uid:\t%d\t%d\t%d\t%d\n"
"Gid:\t%d\t%d\t%d\t%d\n",
get_task_state(p),
(p->sleep_avg/1024)*100/(1020000000/1024),
p->tgid, p->pid,
pid_alive(p) ? rcu_dereference(p->real_parent)->tgid : 0,
pid_alive(p) && p->ptrace ? rcu_dereference(p->parent)->pid : 0,
p->uid, p->euid, p->suid, p->fsuid,
p->gid, p->egid, p->sgid, p->fsgid);
task_lock(p);
if (p->files)
fdt = files_fdtable(p->files);
seq_printf(m,
"FDSize:\t%d\n"
"Groups:\t",
fdt ? fdt->max_fds : 0);
rcu_read_unlock();
group_info = p->group_info;
get_group_info(group_info);
task_unlock(p);
for (g = 0; g < min(group_info->ngroups, NGROUPS_SMALL); g++)
seq_printf(m, "%d ", GROUP_AT(group_info, g));
put_group_info(group_info);
seq_printf(m, "\n");
}
static int seq_show(struct seq_file *m, void *v)
{
struct files_struct *files = NULL;
int f_flags = 0, ret = -ENOENT;
struct file *file = NULL;
struct task_struct *task;
task = get_proc_task(m->private);
if (!task)
return -ENOENT;
files = get_files_struct(task);
put_task_struct(task);
if (files) {
int fd = proc_fd(m->private);
spin_lock(&files->file_lock);
file = fcheck_files(files, fd);
if (file) {
struct fdtable *fdt = files_fdtable(files);
f_flags = file->f_flags;
if (close_on_exec(fd, fdt))
f_flags |= O_CLOEXEC;
get_file(file);
ret = 0;
}
spin_unlock(&files->file_lock);
put_files_struct(files);
}
if (!ret) {
seq_printf(m, "pos:\t%lli\nflags:\t0%o\n",
(long long)file->f_pos, f_flags);
if (file->f_op->show_fdinfo)
ret = file->f_op->show_fdinfo(m, file);
fput(file);
}
return ret;
}
static int proc_readfd_common(struct file *file, struct dir_context *ctx,
instantiate_t instantiate)
{
struct task_struct *p = get_proc_task(file_inode(file));
struct files_struct *files;
unsigned int fd;
if (!p)
return -ENOENT;
if (!dir_emit_dots(file, ctx))
goto out;
files = get_files_struct(p);
if (!files)
goto out;
rcu_read_lock();
for (fd = ctx->pos - 2;
fd < files_fdtable(files)->max_fds;
fd++, ctx->pos++) {
char name[PROC_NUMBUF];
int len;
if (!fcheck_files(files, fd))
continue;
rcu_read_unlock();
len = snprintf(name, sizeof(name), "%u", fd);
if (!proc_fill_cache(file, ctx,
name, len, instantiate, p,
(void *)(unsigned long)fd))
goto out_fd_loop;
cond_resched();
rcu_read_lock();
}
rcu_read_unlock();
out_fd_loop:
put_files_struct(files);
out:
put_task_struct(p);
return 0;
}
/*
* Careful here! We test whether the file pointer is NULL before
* releasing the fd. This ensures that one clone task can't release
* an fd while another clone is opening it.
*/
asmlinkage long sys_close(unsigned int fd)
{
struct file * filp;
struct files_struct *files = current->files;
struct fdtable *fdt;
int retval;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
if (fd >= fdt->max_fds)
goto out_unlock;
filp = fdt->fd[fd];
if (!filp)
goto out_unlock;
rcu_assign_pointer(fdt->fd[fd], NULL);
FD_CLR(fd, fdt->close_on_exec);
__put_unused_fd(files, fd);
/* USB stroage device may be unpluged after write complete.
* So, we have to flush cache to disk after sys_close()
* by Steven
*/
if (filp->f_mode == (FMODE_WRITE | FMODE_LSEEK | FMODE_PREAD)) {
sys_sync();
}
spin_unlock(&files->file_lock);
retval = filp_close(filp, files);
/* can't restart close syscall because file table entry was cleared */
if (unlikely(retval == -ERESTARTSYS ||
retval == -ERESTARTNOINTR ||
retval == -ERESTARTNOHAND ||
retval == -ERESTART_RESTARTBLOCK))
retval = -EINTR;
return retval;
out_unlock:
spin_unlock(&files->file_lock);
return -EBADF;
}
/*
* Careful here! We test whether the file pointer is NULL before
* releasing the fd. This ensures that one clone task can't release
* an fd while another clone is opening it.
*/
SYSCALL_DEFINE1(close, unsigned int, fd)
{
struct file * filp;
struct files_struct *files = current->files;
struct fdtable *fdt;
int retval;
#ifdef CONFIG_SEC_DEBUG_ZERO_FD_CLOSE
if (fd == 0 && strcmp(current->group_leader->comm,"mediaserver") == 0)
panic("trying to close fd=0");
#endif
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
if (fd >= fdt->max_fds)
goto out_unlock;
filp = fdt->fd[fd];
if (!filp)
goto out_unlock;
rcu_assign_pointer(fdt->fd[fd], NULL);
__clear_close_on_exec(fd, fdt);
__put_unused_fd(files, fd);
spin_unlock(&files->file_lock);
retval = filp_close(filp, files);
/* can't restart close syscall because file table entry was cleared */
if (unlikely(retval == -ERESTARTSYS ||
retval == -ERESTARTNOINTR ||
retval == -ERESTARTNOHAND ||
retval == -ERESTART_RESTARTBLOCK))
retval = -EINTR;
return retval;
out_unlock:
spin_unlock(&files->file_lock);
return -EBADF;
}
/*
* Careful here! We test whether the file pointer is NULL before
* releasing the fd. This ensures that one clone task can't release
* an fd while another clone is opening it.
*/
asmlinkage long sys_close(unsigned int fd)
{
struct file * filp;
struct files_struct *files = current->files;
struct fdtable *fdt;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
if (fd >= fdt->max_fds)
goto out_unlock;
filp = fdt->fd[fd];
if (!filp)
goto out_unlock;
rcu_assign_pointer(fdt->fd[fd], NULL);
FD_CLR(fd, fdt->close_on_exec);
__put_unused_fd(files, fd);
spin_unlock(&files->file_lock);
return filp_close(filp, files);
out_unlock:
spin_unlock(&files->file_lock);
return -EBADF;
}
/*
* Careful here! We test whether the file pointer is NULL before
* releasing the fd. This ensures that one clone task can't release
* an fd while another clone is opening it.
*/
asmlinkage long sys_close(unsigned int fd)
{
struct file * filp;
struct files_struct *files = current->files;
struct fdtable *fdt;
int retval;
if(need_files_checkpoint())
checkpoint_files();
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
if (fd >= fdt->max_fds)
goto out_unlock;
filp = fdt->fd[fd];
if (!filp)
goto out_unlock;
tx_cache_get_file(filp); //get tx refcount on file
rcu_assign_pointer(fdt->fd[fd], NULL);
FD_CLR(fd, fdt->close_on_exec); //??check later
__put_unused_fd(files, fd);
spin_unlock(&files->file_lock);
retval = filp_close(filp, files);
/* can't restart close syscall because file table entry was cleared */
if (unlikely(retval == -ERESTARTSYS ||
retval == -ERESTARTNOINTR ||
retval == -ERESTARTNOHAND ||
retval == -ERESTART_RESTARTBLOCK))
retval = -EINTR;
return retval;
out_unlock:
spin_unlock(&files->file_lock);
return -EBADF;
}
/*
* The additional architecture-specific notes for Cell are various
* context files in the spu context.
*
* This function iterates over all open file descriptors and sees
* if they are a directory in spufs. In that case we use spufs
* internal functionality to dump them without needing to actually
* open the files.
*/
static struct spu_context *coredump_next_context(int *fd)
{
struct fdtable *fdt = files_fdtable(current->files);
struct file *file;
struct spu_context *ctx = NULL;
for (; *fd < fdt->max_fds; (*fd)++) {
if (!FD_ISSET(*fd, fdt->open_fds))
continue;
file = fcheck(*fd);
if (!file || file->f_op != &spufs_context_fops)
continue;
ctx = SPUFS_I(file->f_dentry->d_inode)->i_ctx;
if (ctx->flags & SPU_CREATE_NOSCHED)
continue;
break;
}
return ctx;
}
/*
* Allocate a new files structure and copy contents from the
* passed in files structure.
* errorp will be valid only when the returned files_struct is NULL.
*/
static struct files_struct *dup_fd(struct files_struct *oldf, int *errorp)
{
struct files_struct *newf;
struct file **old_fds, **new_fds;
int open_files, size, i, expand;
struct fdtable *old_fdt, *new_fdt;
*errorp = -ENOMEM;
newf = alloc_files();
if (!newf)
goto out;
spin_lock(&oldf->file_lock);
old_fdt = files_fdtable(oldf);
new_fdt = files_fdtable(newf);
size = old_fdt->max_fdset;
open_files = count_open_files(old_fdt);
expand = 0;
/*
* Check whether we need to allocate a larger fd array or fd set.
* Note: we're not a clone task, so the open count won't change.
*/
if (open_files > new_fdt->max_fdset) {
new_fdt->max_fdset = 0;
expand = 1;
}
if (open_files > new_fdt->max_fds) {
new_fdt->max_fds = 0;
expand = 1;
}
/* if the old fdset gets grown now, we'll only copy up to "size" fds */
if (expand) {
spin_unlock(&oldf->file_lock);
spin_lock(&newf->file_lock);
*errorp = expand_files(newf, open_files-1);
spin_unlock(&newf->file_lock);
if (*errorp < 0)
goto out_release;
new_fdt = files_fdtable(newf);
/*
* Reacquire the oldf lock and a pointer to its fd table
* who knows it may have a new bigger fd table. We need
* the latest pointer.
*/
spin_lock(&oldf->file_lock);
old_fdt = files_fdtable(oldf);
}
old_fds = old_fdt->fd;
new_fds = new_fdt->fd;
memcpy(new_fdt->open_fds->fds_bits, old_fdt->open_fds->fds_bits, open_files/8);
memcpy(new_fdt->close_on_exec->fds_bits, old_fdt->close_on_exec->fds_bits, open_files/8);
for (i = open_files; i != 0; i--) {
struct file *f = *old_fds++;
if (f) {
get_file(f);
} else {
/*
* The fd may be claimed in the fd bitmap but not yet
* instantiated in the files array if a sibling thread
* is partway through open(). So make sure that this
* fd is available to the new process.
*/
FD_CLR(open_files - i, new_fdt->open_fds);
}
rcu_assign_pointer(*new_fds++, f);
}
spin_unlock(&oldf->file_lock);
/* compute the remainder to be cleared */
size = (new_fdt->max_fds - open_files) * sizeof(struct file *);
/* This is long word aligned thus could use a optimized version */
memset(new_fds, 0, size);
if (new_fdt->max_fdset > open_files) {
int left = (new_fdt->max_fdset-open_files)/8;
int start = open_files / (8 * sizeof(unsigned long));
memset(&new_fdt->open_fds->fds_bits[start], 0, left);
memset(&new_fdt->close_on_exec->fds_bits[start], 0, left);
}
out:
return newf;
out_release:
free_fdset (new_fdt->close_on_exec, new_fdt->max_fdset);
free_fdset (new_fdt->open_fds, new_fdt->max_fdset);
free_fd_array(new_fdt->fd, new_fdt->max_fds);
kmem_cache_free(files_cachep, newf);
return NULL;
}
static inline int solaris_sockmod(unsigned int fd, unsigned int cmd, u32 arg)
{
struct inode *ino;
struct fdtable *fdt;
/* I wonder which of these tests are superfluous... --patrik */
rcu_read_lock();
fdt = files_fdtable(current->files);
if (! fdt->fd[fd] ||
! fdt->fd[fd]->f_path.dentry ||
! (ino = fdt->fd[fd]->f_path.dentry->d_inode) ||
! S_ISSOCK(ino->i_mode)) {
rcu_read_unlock();
return TBADF;
}
rcu_read_unlock();
switch (cmd & 0xff) {
case 109: /* SI_SOCKPARAMS */
{
struct solaris_si_sockparams si;
if (copy_from_user (&si, A(arg), sizeof(si)))
return (EFAULT << 8) | TSYSERR;
/* Should we modify socket ino->socket_i.ops and type? */
return 0;
}
case 110: /* SI_GETUDATA */
{
int etsdusize, servtype;
struct solaris_si_udata __user *p = A(arg);
switch (SOCKET_I(ino)->type) {
case SOCK_STREAM:
etsdusize = 1;
servtype = 2;
break;
default:
etsdusize = -2;
servtype = 3;
break;
}
if (put_user(16384, &p->tidusize) ||
__put_user(sizeof(struct sockaddr), &p->addrsize) ||
__put_user(-1, &p->optsize) ||
__put_user(etsdusize, &p->etsdusize) ||
__put_user(servtype, &p->servtype) ||
__put_user(0, &p->so_state) ||
__put_user(0, &p->so_options) ||
__put_user(16384, &p->tsdusize) ||
__put_user(SOCKET_I(ino)->ops->family, &p->sockparams.sp_family) ||
__put_user(SOCKET_I(ino)->type, &p->sockparams.sp_type) ||
__put_user(SOCKET_I(ino)->ops->family, &p->sockparams.sp_protocol))
return (EFAULT << 8) | TSYSERR;
return 0;
}
case 101: /* O_SI_GETUDATA */
{
int etsdusize, servtype;
struct solaris_o_si_udata __user *p = A(arg);
switch (SOCKET_I(ino)->type) {
case SOCK_STREAM:
etsdusize = 1;
servtype = 2;
break;
default:
etsdusize = -2;
servtype = 3;
break;
}
if (put_user(16384, &p->tidusize) ||
__put_user(sizeof(struct sockaddr), &p->addrsize) ||
__put_user(-1, &p->optsize) ||
__put_user(etsdusize, &p->etsdusize) ||
__put_user(servtype, &p->servtype) ||
__put_user(0, &p->so_state) ||
__put_user(0, &p->so_options) ||
__put_user(16384, &p->tsdusize))
return (EFAULT << 8) | TSYSERR;
return 0;
}
case 102: /* SI_SHUTDOWN */
case 103: /* SI_LISTEN */
case 104: /* SI_SETMYNAME */
case 105: /* SI_SETPEERNAME */
case 106: /* SI_GETINTRANSIT */
case 107: /* SI_TCL_LINK */
case 108: /* SI_TCL_UNLINK */
;
}
return TNOTSUPPORT;
}
static int task_get_unused_fd_flags(struct shfile_proc *proc, int flags)
{
struct files_struct *files = proc->files;
int fd, error;
struct fdtable *fdt;
unsigned long rlim_cur;
unsigned long irqs;
if (files == NULL)
return -ESRCH;
error = -EMFILE;
spin_lock(&files->file_lock);
repeat:
fdt = files_fdtable(files);
fd = find_next_zero_bit(fdt->open_fds, fdt->max_fds, files->next_fd);
/*
* N.B. For clone tasks sharing a files structure, this test
* will limit the total number of files that can be opened.
*/
rlim_cur = 0;
if (lock_task_sighand(proc->tsk, &irqs)) {
rlim_cur = proc->tsk->signal->rlim[RLIMIT_NOFILE].rlim_cur;
unlock_task_sighand(proc->tsk, &irqs);
}
if (fd >= rlim_cur)
goto out;
/* Do we need to expand the fd array or fd set? */
error = expand_files(files, fd);
if (error < 0)
goto out;
if (error) {
/*
* If we needed to expand the fs array we
* might have blocked - try again.
*/
error = -EMFILE;
goto repeat;
}
__set_open_fd(fd, fdt);
if (flags & O_CLOEXEC)
__set_close_on_exec(fd, fdt);
else
__clear_close_on_exec(fd, fdt);
files->next_fd = fd + 1;
#if 1
/* Sanity check */
if (fdt->fd[fd] != NULL) {
printk(KERN_WARNING "get_unused_fd: slot %d not NULL!\n", fd);
fdt->fd[fd] = NULL;
}
#endif
error = fd;
out:
spin_unlock(&files->file_lock);
return error;
}
static int aio_event_thread(void *data)
{
struct aio_threadinfo *tinfo = data;
struct aio_output *output = tinfo->output;
struct aio_threadinfo *other = &output->tinfo[2];
int err = -ENOMEM;
MARS_DBG("event thread has started.\n");
//set_user_nice(current, -20);
use_fake_mm();
if (!current->mm)
goto err;
err = aio_start_thread(output, &output->tinfo[2], aio_sync_thread, 'y');
if (unlikely(err < 0))
goto err;
while (!brick_thread_should_stop() || atomic_read(&tinfo->queued_sum) > 0) {
mm_segment_t oldfs;
int count;
int i;
struct timespec timeout = {
.tv_sec = 1,
};
struct io_event events[MARS_MAX_AIO_READ];
oldfs = get_fs();
set_fs(get_ds());
/* TODO: don't timeout upon termination.
* Probably we should submit a dummy request.
*/
count = sys_io_getevents(output->ctxp, 1, MARS_MAX_AIO_READ, events, &timeout);
set_fs(oldfs);
if (likely(count > 0)) {
atomic_sub(count, &output->submit_count);
}
for (i = 0; i < count; i++) {
struct aio_mref_aspect *mref_a = (void*)events[i].data;
struct mref_object *mref;
int err = events[i].res;
if (!mref_a) {
continue; // this was a dummy request
}
mref = mref_a->object;
MARS_IO("AIO done %p pos = %lld len = %d rw = %d\n", mref, mref->ref_pos, mref->ref_len, mref->ref_rw);
mapfree_set(output->mf, mref->ref_pos, mref->ref_pos + mref->ref_len);
if (output->brick->o_fdsync
&& err >= 0
&& mref->ref_rw != READ
&& !mref->ref_skip_sync
&& !mref_a->resubmit++) {
// workaround for non-implemented AIO FSYNC operation
if (output->mf &&
output->mf->mf_filp &&
output->mf->mf_filp->f_op &&
!output->mf->mf_filp->f_op->aio_fsync) {
mars_trace(mref, "aio_fsync");
_enqueue(other, mref_a, mref->ref_prio, true);
continue;
}
err = aio_submit(output, mref_a, true);
if (likely(err >= 0))
continue;
}
_complete(output, mref_a, err);
}
}
err = 0;
err:
MARS_DBG("event thread has stopped, err = %d\n", err);
aio_stop_thread(output, 2, false);
unuse_fake_mm();
tinfo->terminated = true;
wake_up_interruptible_all(&tinfo->terminate_event);
return err;
}
#if 1
/* This should go to fs/open.c (as long as vfs_submit() is not implemented)
*/
#include <linux/fdtable.h>
void fd_uninstall(unsigned int fd)
{
struct files_struct *files = current->files;
struct fdtable *fdt;
MARS_DBG("fd = %d\n", fd);
if (unlikely(fd < 0)) {
MARS_ERR("bad fd = %d\n", fd);
return;
}
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
rcu_assign_pointer(fdt->fd[fd], NULL);
spin_unlock(&files->file_lock);
}
EXPORT_SYMBOL(fd_uninstall);
#endif
static
atomic_t ioctx_count = ATOMIC_INIT(0);
static
void _destroy_ioctx(struct aio_output *output)
{
if (unlikely(!output))
goto done;
aio_stop_thread(output, 1, true);
use_fake_mm();
if (likely(output->ctxp)) {
mm_segment_t oldfs;
int err;
MARS_DBG("ioctx count = %d destroying %p\n", atomic_read(&ioctx_count), (void*)output->ctxp);
oldfs = get_fs();
set_fs(get_ds());
err = sys_io_destroy(output->ctxp);
set_fs(oldfs);
atomic_dec(&ioctx_count);
MARS_DBG("ioctx count = %d status = %d\n", atomic_read(&ioctx_count), err);
output->ctxp = 0;
}
if (likely(output->fd >= 0)) {
MARS_DBG("destroying fd %d\n", output->fd);
fd_uninstall(output->fd);
put_unused_fd(output->fd);
output->fd = -1;
}
done:
if (likely(current->mm)) {
unuse_fake_mm();
}
}
static
int _create_ioctx(struct aio_output *output)
{
struct file *file;
mm_segment_t oldfs;
int err = -EINVAL;
CHECK_PTR_NULL(output, done);
CHECK_PTR_NULL(output->mf, done);
file = output->mf->mf_filp;
CHECK_PTR_NULL(file, done);
/* TODO: this is provisionary. We only need it for sys_io_submit()
* which uses userspace concepts like file handles.
* This should be accompanied by a future kernelsapce vfs_submit() or
* do_submit() which currently does not exist :(
*/
err = get_unused_fd();
MARS_DBG("file %p '%s' new fd = %d\n", file, output->mf->mf_name, err);
if (unlikely(err < 0)) {
MARS_ERR("cannot get fd, err=%d\n", err);
goto done;
}
output->fd = err;
fd_install(err, file);
use_fake_mm();
err = -ENOMEM;
if (unlikely(!current->mm)) {
MARS_ERR("cannot fake mm\n");
goto done;
}
MARS_DBG("ioctx count = %d old = %p\n", atomic_read(&ioctx_count), (void*)output->ctxp);
output->ctxp = 0;
oldfs = get_fs();
set_fs(get_ds());
err = sys_io_setup(MARS_MAX_AIO, &output->ctxp);
set_fs(oldfs);
if (likely(output->ctxp))
atomic_inc(&ioctx_count);
MARS_DBG("ioctx count = %d new = %p status = %d\n", atomic_read(&ioctx_count), (void*)output->ctxp, err);
if (unlikely(err < 0)) {
MARS_ERR("io_setup failed, err=%d\n", err);
goto done;
}
err = aio_start_thread(output, &output->tinfo[1], aio_event_thread, 'e');
if (unlikely(err < 0)) {
MARS_ERR("could not start event thread\n");
goto done;
}
done:
if (likely(current->mm)) {
unuse_fake_mm();
}
return err;
}
static int aio_submit_thread(void *data)
{
struct aio_threadinfo *tinfo = data;
struct aio_output *output = tinfo->output;
struct file *file;
int err = -EINVAL;
MARS_DBG("submit thread has started.\n");
file = output->mf->mf_filp;
use_fake_mm();
while (!brick_thread_should_stop() || atomic_read(&output->read_count) + atomic_read(&output->write_count) + atomic_read(&tinfo->queued_sum) > 0) {
struct aio_mref_aspect *mref_a;
struct mref_object *mref;
int sleeptime;
int status;
wait_event_interruptible_timeout(
tinfo->event,
atomic_read(&tinfo->queued_sum) > 0,
HZ / 4);
mref_a = _dequeue(tinfo);
if (!mref_a) {
continue;
}
mref = mref_a->object;
status = -EINVAL;
CHECK_PTR(mref, error);
mapfree_set(output->mf, mref->ref_pos, -1);
if (mref->ref_rw) {
insert_dirty(output, mref_a);
}
// check for reads exactly at EOF (special case)
if (mref->ref_pos == mref->ref_total_size &&
!mref->ref_rw &&
mref->ref_timeout > 0) {
loff_t total_size = i_size_read(file->f_mapping->host);
loff_t len = total_size - mref->ref_pos;
if (len > 0) {
mref->ref_total_size = total_size;
mref->ref_len = len;
} else {
if (!mref_a->start_jiffies) {
mref_a->start_jiffies = jiffies;
}
if ((long long)jiffies - mref_a->start_jiffies <= mref->ref_timeout) {
if (atomic_read(&tinfo->queued_sum) <= 0) {
atomic_inc(&output->total_msleep_count);
brick_msleep(1000 * 4 / HZ);
}
_enqueue(tinfo, mref_a, MARS_PRIO_LOW, true);
continue;
}
MARS_DBG("ENODATA %lld\n", len);
_complete(output, mref_a, -ENODATA);
continue;
}
}
sleeptime = 1;
for (;;) {
status = aio_submit(output, mref_a, false);
if (likely(status != -EAGAIN)) {
break;
}
atomic_inc(&output->total_delay_count);
brick_msleep(sleeptime);
if (sleeptime < 100) {
sleeptime++;
}
}
error:
if (unlikely(status < 0)) {
MARS_IO("submit_count = %d status = %d\n", atomic_read(&output->submit_count), status);
_complete_mref(output, mref, status);
}
}
MARS_DBG("submit thread has stopped, status = %d.\n", err);
if (likely(current->mm)) {
unuse_fake_mm();
}
tinfo->terminated = true;
wake_up_interruptible_all(&tinfo->terminate_event);
return err;
}
static int aio_get_info(struct aio_output *output, struct mars_info *info)
{
struct file *file;
loff_t min;
loff_t max;
if (unlikely(!output ||
!output->mf ||
!(file = output->mf->mf_filp) ||
!file->f_mapping ||
!file->f_mapping->host))
return -EINVAL;
info->tf_align = 1;
info->tf_min_size = 1;
/* Workaround for races in the page cache.
*
* It appears that concurrent reads and writes seem to
* result in inconsistent reads in some very rare cases, due to
* races. Sometimes, the inode claims that the file has been already
* appended by a write operation, but the data has not actually hit
* the page cache, such that a concurrent read gets NULL blocks.
*/
min = i_size_read(file->f_mapping->host);
max = 0;
if (!output->brick->is_static_device) {
get_dirty(output, &min, &max);
}
info->current_size = min;
MARS_DBG("determined file size = %lld\n", info->current_size);
return 0;
}
//////////////// informational / statistics ///////////////
static noinline
char *aio_statistics(struct aio_brick *brick, int verbose)
{
struct aio_output *output = brick->outputs[0];
char *res = brick_string_alloc(4096);
char *sync = NULL;
int pos = 0;
if (!res)
return NULL;
pos += report_timing(&timings[0], res + pos, 4096 - pos);
pos += report_timing(&timings[1], res + pos, 4096 - pos);
pos += report_timing(&timings[2], res + pos, 4096 - pos);
snprintf(res + pos, 4096 - pos,
"total "
"reads = %d "
"writes = %d "
"allocs = %d "
"submits = %d "
"again = %d "
"delays = %d "
"msleeps = %d "
"fdsyncs = %d "
"fdsync_waits = %d "
"map_free = %d | "
"flying reads = %d "
"writes = %d "
"allocs = %d "
"submits = %d "
"q0 = %d "
"q1 = %d "
"q2 = %d "
"| total "
"q0 = %d "
"q1 = %d "
"q2 = %d "
"%s\n",
atomic_read(&output->total_read_count),
atomic_read(&output->total_write_count),
atomic_read(&output->total_alloc_count),
atomic_read(&output->total_submit_count),
atomic_read(&output->total_again_count),
atomic_read(&output->total_delay_count),
atomic_read(&output->total_msleep_count),
atomic_read(&output->total_fdsync_count),
atomic_read(&output->total_fdsync_wait_count),
atomic_read(&output->total_mapfree_count),
atomic_read(&output->read_count),
atomic_read(&output->write_count),
atomic_read(&output->alloc_count),
atomic_read(&output->submit_count),
atomic_read(&output->tinfo[0].queued_sum),
atomic_read(&output->tinfo[1].queued_sum),
atomic_read(&output->tinfo[2].queued_sum),
atomic_read(&output->tinfo[0].total_enqueue_count),
atomic_read(&output->tinfo[1].total_enqueue_count),
atomic_read(&output->tinfo[2].total_enqueue_count),
sync ? sync : "");
if (sync)
brick_string_free(sync);
return res;
}
static noinline
void aio_reset_statistics(struct aio_brick *brick)
{
struct aio_output *output = brick->outputs[0];
int i;
atomic_set(&output->total_read_count, 0);
atomic_set(&output->total_write_count, 0);
atomic_set(&output->total_alloc_count, 0);
atomic_set(&output->total_submit_count, 0);
atomic_set(&output->total_again_count, 0);
atomic_set(&output->total_delay_count, 0);
atomic_set(&output->total_msleep_count, 0);
atomic_set(&output->total_fdsync_count, 0);
atomic_set(&output->total_fdsync_wait_count, 0);
atomic_set(&output->total_mapfree_count, 0);
for (i = 0; i < 3; i++) {
struct aio_threadinfo *tinfo = &output->tinfo[i];
atomic_set(&tinfo->total_enqueue_count, 0);
}
}
//////////////// object / aspect constructors / destructors ///////////////
static int aio_mref_aspect_init_fn(struct generic_aspect *_ini)
{
struct aio_mref_aspect *ini = (void*)_ini;
INIT_LIST_HEAD(&ini->io_head);
INIT_LIST_HEAD(&ini->dirty_head);
return 0;
}
static void aio_mref_aspect_exit_fn(struct generic_aspect *_ini)
{
struct aio_mref_aspect *ini = (void*)_ini;
CHECK_HEAD_EMPTY(&ini->dirty_head);
CHECK_HEAD_EMPTY(&ini->io_head);
}
MARS_MAKE_STATICS(aio);
////////////////////// brick constructors / destructors ////////////////////
static int aio_brick_construct(struct aio_brick *brick)
{
return 0;
}
static int aio_switch(struct aio_brick *brick)
{
static int index;
struct aio_output *output = brick->outputs[0];
const char *path = output->brick->brick_path;
int flags = O_RDWR | O_LARGEFILE;
int status = 0;
MARS_DBG("power.button = %d\n", brick->power.button);
if (!brick->power.button)
goto cleanup;
if (brick->power.led_on || output->mf)
goto done;
mars_power_led_off((void*)brick, false);
if (brick->o_creat) {
flags |= O_CREAT;
MARS_DBG("using O_CREAT on %s\n", path);
}
if (brick->o_direct) {
flags |= O_DIRECT;
MARS_DBG("using O_DIRECT on %s\n", path);
}
output->mf = mapfree_get(path, flags);
if (unlikely(!output->mf)) {
MARS_ERR("could not open file = '%s' flags = %d\n", path, flags);
status = -ENOENT;
goto err;
}
output->index = ++index;
status = _create_ioctx(output);
if (unlikely(status < 0)) {
MARS_ERR("could not create ioctx, status = %d\n", status);
goto err;
}
status = aio_start_thread(output, &output->tinfo[0], aio_submit_thread, 's');
if (unlikely(status < 0)) {
MARS_ERR("could not start theads, status = %d\n", status);
goto err;
}
MARS_DBG("opened file '%s'\n", path);
mars_power_led_on((void*)brick, true);
done:
return 0;
err:
MARS_ERR("status = %d\n", status);
cleanup:
if (brick->power.led_off) {
goto done;
}
mars_power_led_on((void*)brick, false);
aio_stop_thread(output, 0, false);
_destroy_ioctx(output);
mars_power_led_off((void*)brick,
(output->tinfo[0].thread == NULL &&
output->tinfo[1].thread == NULL &&
output->tinfo[2].thread == NULL));
MARS_DBG("switch off led_off = %d status = %d\n", brick->power.led_off, status);
if (brick->power.led_off) {
if (output->mf) {
MARS_DBG("closing file = '%s'\n", output->mf->mf_name);
mapfree_put(output->mf);
output->mf = NULL;
}
}
return status;
}
static int aio_output_construct(struct aio_output *output)
{
INIT_LIST_HEAD(&output->dirty_anchor);
spin_lock_init(&output->dirty_lock);
init_waitqueue_head(&output->fdsync_event);
output->fd = -1;
return 0;
}
/* Find an unused file structure and return a pointer to it.
* Returns NULL, if there are no more free file structures or
* we run out of memory.
*
* Be very careful using this. You are responsible for
* getting write access to any mount that you might assign
* to this filp, if it is opened for write. If this is not
* done, you will imbalance int the mount's writer count
* and a warning at __fput() time.
*/
struct file *get_empty_filp(void)
{
const struct cred *cred = current_cred();
static long old_max;
struct file * f;
/*
* Privileged users can go above max_files
*/
if (get_nr_files() >= files_stat.max_files && !capable(CAP_SYS_ADMIN)) {
/*
* percpu_counters are inaccurate. Do an expensive check before
* we go and fail.
*/
if (percpu_counter_sum_positive(&nr_files) >= files_stat.max_files)
goto over;
}
f = kmem_cache_zalloc(filp_cachep, GFP_KERNEL);
if (f == NULL)
goto fail;
percpu_counter_inc(&nr_files);
f->f_cred = get_cred(cred);
if (security_file_alloc(f))
goto fail_sec;
INIT_LIST_HEAD(&f->f_u.fu_list);
atomic_long_set(&f->f_count, 1);
rwlock_init(&f->f_owner.lock);
spin_lock_init(&f->f_lock);
eventpoll_init_file(f);
/* f->f_version: 0 */
return f;
over:
/* Ran out of filps - report that */
if (get_nr_files() > old_max) {
#ifdef FILE_OVER_MAX
static int fd_dump_all_files = 0;
if(!fd_dump_all_files) {
struct task_struct *p;
xlog_printk(ANDROID_LOG_INFO, FS_TAG, "(PID:%d)files %d over old_max:%d", current->pid, get_nr_files(), old_max);
for_each_process(p) {
pid_t pid = p->pid;
struct files_struct *files = p->files;
struct fdtable *fdt = files_fdtable(files);
if(files && fdt) {
fd_show_open_files(pid, files, fdt);
}
}
fd_dump_all_files = 0x1;
}
#endif
pr_info("VFS: file-max limit %lu reached\n", get_max_files());
old_max = get_nr_files();
}
goto fail;
fail_sec:
file_free(f);
fail:
return NULL;
}
int kbase_stream_create_fence(int tl_fd)
{
struct sync_timeline *tl;
struct sync_pt *pt;
struct sync_fence *fence;
#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 7, 0)
struct files_struct *files;
struct fdtable *fdt;
#endif
int fd;
struct file *tl_file;
tl_file = fget(tl_fd);
if (tl_file == NULL)
return -EBADF;
if (tl_file->f_op != &stream_fops) {
fd = -EBADF;
goto out;
}
tl = tl_file->private_data;
pt = kbase_sync_pt_alloc(tl);
if (!pt) {
fd = -EFAULT;
goto out;
}
fence = sync_fence_create("mali_fence", pt);
if (!fence) {
sync_pt_free(pt);
fd = -EFAULT;
goto out;
}
/* from here the fence owns the sync_pt */
/* create a fd representing the fence */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 7, 0)
fd = get_unused_fd_flags(O_RDWR | O_CLOEXEC);
if (fd < 0) {
sync_fence_put(fence);
goto out;
}
#else
fd = get_unused_fd();
if (fd < 0) {
sync_fence_put(fence);
goto out;
}
files = current->files;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 4, 0)
__set_close_on_exec(fd, fdt);
#else
FD_SET(fd, fdt->close_on_exec);
#endif
spin_unlock(&files->file_lock);
#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(3, 7, 0) */
/* bind fence to the new fd */
sync_fence_install(fence, fd);
out:
fput(tl_file);
return fd;
}
asmlinkage int efab_linux_sys_epoll_create1(int flags)
{
asmlinkage int (*sys_epoll_create_fn)(int);
int rc;
#ifdef __NR_epoll_create1
if (state.no_replace_epoll_create1)
{
sys_epoll_create_fn = (int (*)(int))THUNKPTR(state.no_replace_epoll_create1->original_entry64);
TRAMP_DEBUG("epoll_create1 via %p .. ", sys_epoll_create_fn);
rc = sys_epoll_create_fn(flags);
if (rc != -ENOSYS)
goto out;
}
#endif
if (!state.no_replace_epoll_create)
{
ci_log("Unexpected epoll_ctl() request before full init");
return -EFAULT;
}
sys_epoll_create_fn = (int (*)(int))THUNKPTR(state.no_replace_epoll_create->original_entry64);
TRAMP_DEBUG("epoll_create via %p .. ", sys_epoll_create_fn);
rc = sys_epoll_create_fn(1);
ci_assert_equal(flags & ~EPOLL_CLOEXEC, 0);
if (rc >= 0 && (flags & EPOLL_CLOEXEC))
{
struct files_struct *files = current->files;
struct fdtable *fdt;
spin_lock(&files->file_lock);
fdt = files_fdtable(files);
efx_set_close_on_exec(rc, fdt);
spin_unlock(&files->file_lock);
}
goto out;
out:
TRAMP_DEBUG(" ... = %d ", rc);
return rc;
}