struct fdtable_defer {
spinlock_t lock;
struct work_struct wq;
- struct timer_list timer;
struct fdtable *next;
};
*/
static DEFINE_PER_CPU(struct fdtable_defer, fdtable_defer_list);
-
-/*
- * Allocate an fd array, using kmalloc or vmalloc.
- * Note: the array isn't cleared at allocation time.
- */
-struct file ** alloc_fd_array(int num)
+static inline void * alloc_fdmem(unsigned int size)
{
- struct file **new_fds;
- int size = num * sizeof(struct file *);
-
if (size <= PAGE_SIZE)
- new_fds = (struct file **) kmalloc(size, GFP_KERNEL);
- else
- new_fds = (struct file **) vmalloc(size);
- return new_fds;
-}
-
-void free_fd_array(struct file **array, int num)
-{
- int size = num * sizeof(struct file *);
-
- if (!array) {
- printk (KERN_ERR "free_fd_array: array = 0 (num = %d)\n", num);
- return;
- }
-
- if (num <= NR_OPEN_DEFAULT) /* Don't free the embedded fd array! */
- return;
- else if (size <= PAGE_SIZE)
- kfree(array);
+ return kmalloc(size, GFP_KERNEL);
else
- vfree(array);
+ return vmalloc(size);
}
-static void __free_fdtable(struct fdtable *fdt)
+static inline void free_fdarr(struct fdtable *fdt)
{
- free_fdset(fdt->open_fds, fdt->max_fdset);
- free_fdset(fdt->close_on_exec, fdt->max_fdset);
- free_fd_array(fdt->fd, fdt->max_fds);
- kfree(fdt);
+ if (fdt->max_fds <= (PAGE_SIZE / sizeof(struct file *)))
+ kfree(fdt->fd);
+ else
+ vfree(fdt->fd);
}
-static void fdtable_timer(unsigned long data)
+static inline void free_fdset(struct fdtable *fdt)
{
- struct fdtable_defer *fddef = (struct fdtable_defer *)data;
-
- spin_lock(&fddef->lock);
- /*
- * If someone already emptied the queue return.
- */
- if (!fddef->next)
- goto out;
- if (!schedule_work(&fddef->wq))
- mod_timer(&fddef->timer, 5);
-out:
- spin_unlock(&fddef->lock);
+ if (fdt->max_fds <= (PAGE_SIZE * BITS_PER_BYTE / 2))
+ kfree(fdt->open_fds);
+ else
+ vfree(fdt->open_fds);
}
-static void free_fdtable_work(struct fdtable_defer *f)
+static void free_fdtable_work(struct work_struct *work)
{
+ struct fdtable_defer *f =
+ container_of(work, struct fdtable_defer, wq);
struct fdtable *fdt;
spin_lock_bh(&f->lock);
spin_unlock_bh(&f->lock);
while(fdt) {
struct fdtable *next = fdt->next;
- __free_fdtable(fdt);
+ vfree(fdt->fd);
+ free_fdset(fdt);
+ kfree(fdt);
fdt = next;
}
}
-static void free_fdtable_rcu(struct rcu_head *rcu)
+void free_fdtable_rcu(struct rcu_head *rcu)
{
struct fdtable *fdt = container_of(rcu, struct fdtable, rcu);
- int fdset_size, fdarray_size;
struct fdtable_defer *fddef;
BUG_ON(!fdt);
- fdset_size = fdt->max_fdset / 8;
- fdarray_size = fdt->max_fds * sizeof(struct file *);
- if (fdt->free_files) {
+ if (fdt->max_fds <= NR_OPEN_DEFAULT) {
/*
- * The this fdtable was embedded in the files structure
- * and the files structure itself was getting destroyed.
- * It is now safe to free the files structure.
+ * This fdtable is embedded in the files structure and that
+ * structure itself is getting destroyed.
*/
- kmem_cache_free(files_cachep, fdt->free_files);
+ kmem_cache_free(files_cachep,
+ container_of(fdt, struct files_struct, fdtab));
return;
}
- if (fdt->max_fdset <= EMBEDDED_FD_SET_SIZE &&
- fdt->max_fds <= NR_OPEN_DEFAULT) {
- /*
- * The fdtable was embedded
- */
- return;
- }
- if (fdset_size <= PAGE_SIZE && fdarray_size <= PAGE_SIZE) {
- kfree(fdt->open_fds);
- kfree(fdt->close_on_exec);
+ if (fdt->max_fds <= (PAGE_SIZE / sizeof(struct file *))) {
kfree(fdt->fd);
+ kfree(fdt->open_fds);
kfree(fdt);
} else {
fddef = &get_cpu_var(fdtable_defer_list);
spin_lock(&fddef->lock);
fdt->next = fddef->next;
fddef->next = fdt;
- /*
- * vmallocs are handled from the workqueue context.
- * If the per-cpu workqueue is running, then we
- * defer work scheduling through a timer.
- */
- if (!schedule_work(&fddef->wq))
- mod_timer(&fddef->timer, 5);
+ /* vmallocs are handled from the workqueue context */
+ schedule_work(&fddef->wq);
spin_unlock(&fddef->lock);
put_cpu_var(fdtable_defer_list);
}
}
-void free_fdtable(struct fdtable *fdt)
-{
- if (fdt->free_files ||
- fdt->max_fdset > EMBEDDED_FD_SET_SIZE ||
- fdt->max_fds > NR_OPEN_DEFAULT)
- call_rcu(&fdt->rcu, free_fdtable_rcu);
-}
-
/*
* Expand the fdset in the files_struct. Called with the files spinlock
* held for write.
*/
-static void copy_fdtable(struct fdtable *nfdt, struct fdtable *fdt)
+static void copy_fdtable(struct fdtable *nfdt, struct fdtable *ofdt)
{
- int i;
- int count;
-
- BUG_ON(nfdt->max_fdset < fdt->max_fdset);
- BUG_ON(nfdt->max_fds < fdt->max_fds);
- /* Copy the existing tables and install the new pointers */
-
- i = fdt->max_fdset / (sizeof(unsigned long) * 8);
- count = (nfdt->max_fdset - fdt->max_fdset) / 8;
+ unsigned int cpy, set;
- /*
- * Don't copy the entire array if the current fdset is
- * not yet initialised.
- */
- if (i) {
- memcpy (nfdt->open_fds, fdt->open_fds,
- fdt->max_fdset/8);
- memcpy (nfdt->close_on_exec, fdt->close_on_exec,
- fdt->max_fdset/8);
- memset (&nfdt->open_fds->fds_bits[i], 0, count);
- memset (&nfdt->close_on_exec->fds_bits[i], 0, count);
- }
-
- /* Don't copy/clear the array if we are creating a new
- fd array for fork() */
- if (fdt->max_fds) {
- memcpy(nfdt->fd, fdt->fd,
- fdt->max_fds * sizeof(struct file *));
- /* clear the remainder of the array */
- memset(&nfdt->fd[fdt->max_fds], 0,
- (nfdt->max_fds - fdt->max_fds) *
- sizeof(struct file *));
- }
-}
-
-/*
- * Allocate an fdset array, using kmalloc or vmalloc.
- * Note: the array isn't cleared at allocation time.
- */
-fd_set * alloc_fdset(int num)
-{
- fd_set *new_fdset;
- int size = num / 8;
+ BUG_ON(nfdt->max_fds < ofdt->max_fds);
+ if (ofdt->max_fds == 0)
+ return;
- if (size <= PAGE_SIZE)
- new_fdset = (fd_set *) kmalloc(size, GFP_KERNEL);
- else
- new_fdset = (fd_set *) vmalloc(size);
- return new_fdset;
+ cpy = ofdt->max_fds * sizeof(struct file *);
+ set = (nfdt->max_fds - ofdt->max_fds) * sizeof(struct file *);
+ memcpy(nfdt->fd, ofdt->fd, cpy);
+ memset((char *)(nfdt->fd) + cpy, 0, set);
+
+ cpy = ofdt->max_fds / BITS_PER_BYTE;
+ set = (nfdt->max_fds - ofdt->max_fds) / BITS_PER_BYTE;
+ memcpy(nfdt->open_fds, ofdt->open_fds, cpy);
+ memset((char *)(nfdt->open_fds) + cpy, 0, set);
+ memcpy(nfdt->close_on_exec, ofdt->close_on_exec, cpy);
+ memset((char *)(nfdt->close_on_exec) + cpy, 0, set);
}
-void free_fdset(fd_set *array, int num)
+static struct fdtable * alloc_fdtable(unsigned int nr)
{
- if (num <= EMBEDDED_FD_SET_SIZE) /* Don't free an embedded fdset */
- return;
- else if (num <= 8 * PAGE_SIZE)
- kfree(array);
- else
- vfree(array);
-}
+ struct fdtable *fdt;
+ char *data;
-static struct fdtable *alloc_fdtable(int nr)
-{
- struct fdtable *fdt = NULL;
- int nfds = 0;
- fd_set *new_openset = NULL, *new_execset = NULL;
- struct file **new_fds;
+ /*
+ * Figure out how many fds we actually want to support in this fdtable.
+ * Allocation steps are keyed to the size of the fdarray, since it
+ * grows far faster than any of the other dynamic data. We try to fit
+ * the fdarray into comfortable page-tuned chunks: starting at 1024B
+ * and growing in powers of two from there on.
+ */
+ nr /= (1024 / sizeof(struct file *));
+ nr = roundup_pow_of_two(nr + 1);
+ nr *= (1024 / sizeof(struct file *));
+ if (nr > NR_OPEN)
+ nr = NR_OPEN;
- fdt = kzalloc(sizeof(*fdt), GFP_KERNEL);
+ fdt = kmalloc(sizeof(struct fdtable), GFP_KERNEL);
if (!fdt)
- goto out;
-
- nfds = max_t(int, 8 * L1_CACHE_BYTES, roundup_pow_of_two(nr + 1));
- if (nfds > NR_OPEN)
- nfds = NR_OPEN;
-
- new_openset = alloc_fdset(nfds);
- new_execset = alloc_fdset(nfds);
- if (!new_openset || !new_execset)
- goto out;
- fdt->open_fds = new_openset;
- fdt->close_on_exec = new_execset;
- fdt->max_fdset = nfds;
+ goto out;
+ fdt->max_fds = nr;
+ data = alloc_fdmem(nr * sizeof(struct file *));
+ if (!data)
+ goto out_fdt;
+ fdt->fd = (struct file **)data;
+ data = alloc_fdmem(max_t(unsigned int,
+ 2 * nr / BITS_PER_BYTE, L1_CACHE_BYTES));
+ if (!data)
+ goto out_arr;
+ fdt->open_fds = (fd_set *)data;
+ data += nr / BITS_PER_BYTE;
+ fdt->close_on_exec = (fd_set *)data;
+ INIT_RCU_HEAD(&fdt->rcu);
+ fdt->next = NULL;
- nfds = NR_OPEN_DEFAULT;
- /*
- * Expand to the max in easy steps, and keep expanding it until
- * we have enough for the requested fd array size.
- */
- do {
-#if NR_OPEN_DEFAULT < 256
- if (nfds < 256)
- nfds = 256;
- else
-#endif
- if (nfds < (PAGE_SIZE / sizeof(struct file *)))
- nfds = PAGE_SIZE / sizeof(struct file *);
- else {
- nfds = nfds * 2;
- if (nfds > NR_OPEN)
- nfds = NR_OPEN;
- }
- } while (nfds <= nr);
- new_fds = alloc_fd_array(nfds);
- if (!new_fds)
- goto out2;
- fdt->fd = new_fds;
- fdt->max_fds = nfds;
- fdt->free_files = NULL;
return fdt;
-out2:
- nfds = fdt->max_fdset;
-out:
- free_fdset(new_openset, nfds);
- free_fdset(new_execset, nfds);
+
+out_arr:
+ free_fdarr(fdt);
+out_fdt:
kfree(fdt);
+out:
return NULL;
}
/*
- * Expands the file descriptor table - it will allocate a new fdtable and
- * both fd array and fdset. It is expected to be called with the
- * files_lock held.
+ * Expand the file descriptor table.
+ * This function will allocate a new fdtable and both fd array and fdset, of
+ * the given size.
+ * Return <0 error code on error; 1 on successful completion.
+ * The files->file_lock should be held on entry, and will be held on exit.
*/
static int expand_fdtable(struct files_struct *files, int nr)
__releases(files->file_lock)
__acquires(files->file_lock)
{
- int error = 0;
- struct fdtable *fdt;
- struct fdtable *nfdt = NULL;
+ struct fdtable *new_fdt, *cur_fdt;
spin_unlock(&files->file_lock);
- nfdt = alloc_fdtable(nr);
- if (!nfdt) {
- error = -ENOMEM;
- spin_lock(&files->file_lock);
- goto out;
- }
-
+ new_fdt = alloc_fdtable(nr);
spin_lock(&files->file_lock);
- fdt = files_fdtable(files);
+ if (!new_fdt)
+ return -ENOMEM;
/*
- * Check again since another task may have expanded the
- * fd table while we dropped the lock
+ * Check again since another task may have expanded the fd table while
+ * we dropped the lock
*/
- if (nr >= fdt->max_fds || nr >= fdt->max_fdset) {
- copy_fdtable(nfdt, fdt);
+ cur_fdt = files_fdtable(files);
+ if (nr >= cur_fdt->max_fds) {
+ /* Continue as planned */
+ copy_fdtable(new_fdt, cur_fdt);
+ rcu_assign_pointer(files->fdt, new_fdt);
+ if (cur_fdt->max_fds > NR_OPEN_DEFAULT)
+ free_fdtable(cur_fdt);
} else {
- /* Somebody expanded while we dropped file_lock */
- spin_unlock(&files->file_lock);
- __free_fdtable(nfdt);
- spin_lock(&files->file_lock);
- goto out;
+ /* Somebody else expanded, so undo our attempt */
+ free_fdarr(new_fdt);
+ free_fdset(new_fdt);
+ kfree(new_fdt);
}
- rcu_assign_pointer(files->fdt, nfdt);
- free_fdtable(fdt);
-out:
- return error;
+ return 1;
}
/*
* Expand files.
- * Return <0 on error; 0 nothing done; 1 files expanded, we may have blocked.
- * Should be called with the files->file_lock spinlock held for write.
+ * This function will expand the file structures, if the requested size exceeds
+ * the current capacity and there is room for expansion.
+ * Return <0 error code on error; 0 when nothing done; 1 when files were
+ * expanded and execution may have blocked.
+ * The files->file_lock should be held on entry, and will be held on exit.
*/
int expand_files(struct files_struct *files, int nr)
{
- int err, expand = 0;
struct fdtable *fdt;
fdt = files_fdtable(files);
- if (nr >= fdt->max_fdset || nr >= fdt->max_fds) {
- if (fdt->max_fdset >= NR_OPEN ||
- fdt->max_fds >= NR_OPEN || nr >= NR_OPEN) {
- err = -EMFILE;
- goto out;
- }
- expand = 1;
- if ((err = expand_fdtable(files, nr)))
- goto out;
- }
- err = expand;
-out:
- return err;
+ /* Do we need to expand? */
+ if (nr < fdt->max_fds)
+ return 0;
+ /* Can we expand? */
+ if (nr >= NR_OPEN)
+ return -EMFILE;
+
+ /* All good, so we try */
+ return expand_fdtable(files, nr);
}
static void __devinit fdtable_defer_list_init(int cpu)
{
struct fdtable_defer *fddef = &per_cpu(fdtable_defer_list, cpu);
spin_lock_init(&fddef->lock);
- INIT_WORK(&fddef->wq, (void (*)(void *))free_fdtable_work, fddef);
- init_timer(&fddef->timer);
- fddef->timer.data = (unsigned long)fddef;
- fddef->timer.function = fdtable_timer;
+ INIT_WORK(&fddef->wq, free_fdtable_work);
fddef->next = NULL;
}
git.openpandora.org / pandora-kernel.git / blobdiff