dax: replace XIP documentation with DAX documentation
[linux-drm-fsl-dcu.git] / mm / filemap_xip.c
index 0d105aeff82fe258f06eaacbb8205ade160436cb..8e3f99b619599d40e01c75aad43d3c470d19a386 100644 (file)
@@ -9,6 +9,7 @@
  */
 
 #include <linux/fs.h>
+#include <linux/backing-dev.h>
 #include <linux/pagemap.h>
 #include <linux/export.h>
 #include <linux/uio.h>
 #include <asm/tlbflush.h>
 #include <asm/io.h>
 
-/*
- * We do use our own empty page to avoid interference with other users
- * of ZERO_PAGE(), such as /dev/zero
- */
-static DEFINE_MUTEX(xip_sparse_mutex);
-static seqcount_t xip_sparse_seq = SEQCNT_ZERO(xip_sparse_seq);
-static struct page *__xip_sparse_page;
-
-/* called under xip_sparse_mutex */
-static struct page *xip_sparse_page(void)
-{
-       if (!__xip_sparse_page) {
-               struct page *page = alloc_page(GFP_HIGHUSER | __GFP_ZERO);
-
-               if (page)
-                       __xip_sparse_page = page;
-       }
-       return __xip_sparse_page;
-}
-
-/*
- * This is a file read routine for execute in place files, and uses
- * the mapping->a_ops->get_xip_mem() function for the actual low-level
- * stuff.
- *
- * Note the struct file* is not used at all.  It may be NULL.
- */
-static ssize_t
-do_xip_mapping_read(struct address_space *mapping,
-                   struct file_ra_state *_ra,
-                   struct file *filp,
-                   char __user *buf,
-                   size_t len,
-                   loff_t *ppos)
-{
-       struct inode *inode = mapping->host;
-       pgoff_t index, end_index;
-       unsigned long offset;
-       loff_t isize, pos;
-       size_t copied = 0, error = 0;
-
-       BUG_ON(!mapping->a_ops->get_xip_mem);
-
-       pos = *ppos;
-       index = pos >> PAGE_CACHE_SHIFT;
-       offset = pos & ~PAGE_CACHE_MASK;
-
-       isize = i_size_read(inode);
-       if (!isize)
-               goto out;
-
-       end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
-       do {
-               unsigned long nr, left;
-               void *xip_mem;
-               unsigned long xip_pfn;
-               int zero = 0;
-
-               /* nr is the maximum number of bytes to copy from this page */
-               nr = PAGE_CACHE_SIZE;
-               if (index >= end_index) {
-                       if (index > end_index)
-                               goto out;
-                       nr = ((isize - 1) & ~PAGE_CACHE_MASK) + 1;
-                       if (nr <= offset) {
-                               goto out;
-                       }
-               }
-               nr = nr - offset;
-               if (nr > len - copied)
-                       nr = len - copied;
-
-               error = mapping->a_ops->get_xip_mem(mapping, index, 0,
-                                                       &xip_mem, &xip_pfn);
-               if (unlikely(error)) {
-                       if (error == -ENODATA) {
-                               /* sparse */
-                               zero = 1;
-                       } else
-                               goto out;
-               }
-
-               /* If users can be writing to this page using arbitrary
-                * virtual addresses, take care about potential aliasing
-                * before reading the page on the kernel side.
-                */
-               if (mapping_writably_mapped(mapping))
-                       /* address based flush */ ;
-
-               /*
-                * Ok, we have the mem, so now we can copy it to user space...
-                *
-                * The actor routine returns how many bytes were actually used..
-                * NOTE! This may not be the same as how much of a user buffer
-                * we filled up (we may be padding etc), so we can only update
-                * "pos" here (the actor routine has to update the user buffer
-                * pointers and the remaining count).
-                */
-               if (!zero)
-                       left = __copy_to_user(buf+copied, xip_mem+offset, nr);
-               else
-                       left = __clear_user(buf + copied, nr);
-
-               if (left) {
-                       error = -EFAULT;
-                       goto out;
-               }
-
-               copied += (nr - left);
-               offset += (nr - left);
-               index += offset >> PAGE_CACHE_SHIFT;
-               offset &= ~PAGE_CACHE_MASK;
-       } while (copied < len);
-
-out:
-       *ppos = pos + copied;
-       if (filp)
-               file_accessed(filp);
-
-       return (copied ? copied : error);
-}
-
-ssize_t
-xip_file_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos)
-{
-       if (!access_ok(VERIFY_WRITE, buf, len))
-               return -EFAULT;
-
-       return do_xip_mapping_read(filp->f_mapping, &filp->f_ra, filp,
-                           buf, len, ppos);
-}
-EXPORT_SYMBOL_GPL(xip_file_read);
-
-/*
- * __xip_unmap is invoked from xip_unmap and xip_write
- *
- * This function walks all vmas of the address_space and unmaps the
- * __xip_sparse_page when found at pgoff.
- */
-static void __xip_unmap(struct address_space * mapping, unsigned long pgoff)
-{
-       struct vm_area_struct *vma;
-       struct page *page;
-       unsigned count;
-       int locked = 0;
-
-       count = read_seqcount_begin(&xip_sparse_seq);
-
-       page = __xip_sparse_page;
-       if (!page)
-               return;
-
-retry:
-       i_mmap_lock_read(mapping);
-       vma_interval_tree_foreach(vma, &mapping->i_mmap, pgoff, pgoff) {
-               pte_t *pte, pteval;
-               spinlock_t *ptl;
-               struct mm_struct *mm = vma->vm_mm;
-               unsigned long address = vma->vm_start +
-                       ((pgoff - vma->vm_pgoff) << PAGE_SHIFT);
-
-               BUG_ON(address < vma->vm_start || address >= vma->vm_end);
-               pte = page_check_address(page, mm, address, &ptl, 1);
-               if (pte) {
-                       /* Nuke the page table entry. */
-                       flush_cache_page(vma, address, pte_pfn(*pte));
-                       pteval = ptep_clear_flush(vma, address, pte);
-                       page_remove_rmap(page);
-                       dec_mm_counter(mm, MM_FILEPAGES);
-                       BUG_ON(pte_dirty(pteval));
-                       pte_unmap_unlock(pte, ptl);
-                       /* must invalidate_page _before_ freeing the page */
-                       mmu_notifier_invalidate_page(mm, address);
-                       page_cache_release(page);
-               }
-       }
-       i_mmap_unlock_read(mapping);
-
-       if (locked) {
-               mutex_unlock(&xip_sparse_mutex);
-       } else if (read_seqcount_retry(&xip_sparse_seq, count)) {
-               mutex_lock(&xip_sparse_mutex);
-               locked = 1;
-               goto retry;
-       }
-}
-
-/*
- * xip_fault() is invoked via the vma operations vector for a
- * mapped memory region to read in file data during a page fault.
- *
- * This function is derived from filemap_fault, but used for execute in place
- */
-static int xip_file_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
-{
-       struct file *file = vma->vm_file;
-       struct address_space *mapping = file->f_mapping;
-       struct inode *inode = mapping->host;
-       pgoff_t size;
-       void *xip_mem;
-       unsigned long xip_pfn;
-       struct page *page;
-       int error;
-
-       /* XXX: are VM_FAULT_ codes OK? */
-again:
-       size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
-       if (vmf->pgoff >= size)
-               return VM_FAULT_SIGBUS;
-
-       error = mapping->a_ops->get_xip_mem(mapping, vmf->pgoff, 0,
-                                               &xip_mem, &xip_pfn);
-       if (likely(!error))
-               goto found;
-       if (error != -ENODATA)
-               return VM_FAULT_OOM;
-
-       /* sparse block */
-       if ((vma->vm_flags & (VM_WRITE | VM_MAYWRITE)) &&
-           (vma->vm_flags & (VM_SHARED | VM_MAYSHARE)) &&
-           (!(mapping->host->i_sb->s_flags & MS_RDONLY))) {
-               int err;
-
-               /* maybe shared writable, allocate new block */
-               mutex_lock(&xip_sparse_mutex);
-               error = mapping->a_ops->get_xip_mem(mapping, vmf->pgoff, 1,
-                                                       &xip_mem, &xip_pfn);
-               mutex_unlock(&xip_sparse_mutex);
-               if (error)
-                       return VM_FAULT_SIGBUS;
-               /* unmap sparse mappings at pgoff from all other vmas */
-               __xip_unmap(mapping, vmf->pgoff);
-
-found:
-               err = vm_insert_mixed(vma, (unsigned long)vmf->virtual_address,
-                                                       xip_pfn);
-               if (err == -ENOMEM)
-                       return VM_FAULT_OOM;
-               /*
-                * err == -EBUSY is fine, we've raced against another thread
-                * that faulted-in the same page
-                */
-               if (err != -EBUSY)
-                       BUG_ON(err);
-               return VM_FAULT_NOPAGE;
-       } else {
-               int err, ret = VM_FAULT_OOM;
-
-               mutex_lock(&xip_sparse_mutex);
-               write_seqcount_begin(&xip_sparse_seq);
-               error = mapping->a_ops->get_xip_mem(mapping, vmf->pgoff, 0,
-                                                       &xip_mem, &xip_pfn);
-               if (unlikely(!error)) {
-                       write_seqcount_end(&xip_sparse_seq);
-                       mutex_unlock(&xip_sparse_mutex);
-                       goto again;
-               }
-               if (error != -ENODATA)
-                       goto out;
-               /* not shared and writable, use xip_sparse_page() */
-               page = xip_sparse_page();
-               if (!page)
-                       goto out;
-               err = vm_insert_page(vma, (unsigned long)vmf->virtual_address,
-                                                       page);
-               if (err == -ENOMEM)
-                       goto out;
-
-               ret = VM_FAULT_NOPAGE;
-out:
-               write_seqcount_end(&xip_sparse_seq);
-               mutex_unlock(&xip_sparse_mutex);
-
-               return ret;
-       }
-}
-
-static const struct vm_operations_struct xip_file_vm_ops = {
-       .fault  = xip_file_fault,
-       .page_mkwrite   = filemap_page_mkwrite,
-       .remap_pages = generic_file_remap_pages,
-};
-
-int xip_file_mmap(struct file * file, struct vm_area_struct * vma)
-{
-       BUG_ON(!file->f_mapping->a_ops->get_xip_mem);
-
-       file_accessed(file);
-       vma->vm_ops = &xip_file_vm_ops;
-       vma->vm_flags |= VM_MIXEDMAP;
-       return 0;
-}
-EXPORT_SYMBOL_GPL(xip_file_mmap);
-
-static ssize_t
-__xip_file_write(struct file *filp, const char __user *buf,
-                 size_t count, loff_t pos, loff_t *ppos)
-{
-       struct address_space * mapping = filp->f_mapping;
-       const struct address_space_operations *a_ops = mapping->a_ops;
-       struct inode    *inode = mapping->host;
-       long            status = 0;
-       size_t          bytes;
-       ssize_t         written = 0;
-
-       BUG_ON(!mapping->a_ops->get_xip_mem);
-
-       do {
-               unsigned long index;
-               unsigned long offset;
-               size_t copied;
-               void *xip_mem;
-               unsigned long xip_pfn;
-
-               offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
-               index = pos >> PAGE_CACHE_SHIFT;
-               bytes = PAGE_CACHE_SIZE - offset;
-               if (bytes > count)
-                       bytes = count;
-
-               status = a_ops->get_xip_mem(mapping, index, 0,
-                                               &xip_mem, &xip_pfn);
-               if (status == -ENODATA) {
-                       /* we allocate a new page unmap it */
-                       mutex_lock(&xip_sparse_mutex);
-                       status = a_ops->get_xip_mem(mapping, index, 1,
-                                                       &xip_mem, &xip_pfn);
-                       mutex_unlock(&xip_sparse_mutex);
-                       if (!status)
-                               /* unmap page at pgoff from all other vmas */
-                               __xip_unmap(mapping, index);
-               }
-
-               if (status)
-                       break;
-
-               copied = bytes -
-                       __copy_from_user_nocache(xip_mem + offset, buf, bytes);
-
-               if (likely(copied > 0)) {
-                       status = copied;
-
-                       if (status >= 0) {
-                               written += status;
-                               count -= status;
-                               pos += status;
-                               buf += status;
-                       }
-               }
-               if (unlikely(copied != bytes))
-                       if (status >= 0)
-                               status = -EFAULT;
-               if (status < 0)
-                       break;
-       } while (count);
-       *ppos = pos;
-       /*
-        * No need to use i_size_read() here, the i_size
-        * cannot change under us because we hold i_mutex.
-        */
-       if (pos > inode->i_size) {
-               i_size_write(inode, pos);
-               mark_inode_dirty(inode);
-       }
-
-       return written ? written : status;
-}
-
-ssize_t
-xip_file_write(struct file *filp, const char __user *buf, size_t len,
-              loff_t *ppos)
-{
-       struct address_space *mapping = filp->f_mapping;
-       struct inode *inode = mapping->host;
-       size_t count;
-       loff_t pos;
-       ssize_t ret;
-
-       mutex_lock(&inode->i_mutex);
-
-       if (!access_ok(VERIFY_READ, buf, len)) {
-               ret=-EFAULT;
-               goto out_up;
-       }
-
-       pos = *ppos;
-       count = len;
-
-       /* We can write back this queue in page reclaim */
-       current->backing_dev_info = mapping->backing_dev_info;
-
-       ret = generic_write_checks(filp, &pos, &count, S_ISBLK(inode->i_mode));
-       if (ret)
-               goto out_backing;
-       if (count == 0)
-               goto out_backing;
-
-       ret = file_remove_suid(filp);
-       if (ret)
-               goto out_backing;
-
-       ret = file_update_time(filp);
-       if (ret)
-               goto out_backing;
-
-       ret = __xip_file_write (filp, buf, count, pos, ppos);
-
- out_backing:
-       current->backing_dev_info = NULL;
- out_up:
-       mutex_unlock(&inode->i_mutex);
-       return ret;
-}
-EXPORT_SYMBOL_GPL(xip_file_write);
-
-/*
- * truncate a page used for execute in place
- * functionality is analog to block_truncate_page but does use get_xip_mem
- * to get the page instead of page cache
- */
-int
-xip_truncate_page(struct address_space *mapping, loff_t from)
-{
-       pgoff_t index = from >> PAGE_CACHE_SHIFT;
-       unsigned offset = from & (PAGE_CACHE_SIZE-1);
-       unsigned blocksize;
-       unsigned length;
-       void *xip_mem;
-       unsigned long xip_pfn;
-       int err;
-
-       BUG_ON(!mapping->a_ops->get_xip_mem);
-
-       blocksize = 1 << mapping->host->i_blkbits;
-       length = offset & (blocksize - 1);
-
-       /* Block boundary? Nothing to do */
-       if (!length)
-               return 0;
-
-       length = blocksize - length;
-
-       err = mapping->a_ops->get_xip_mem(mapping, index, 0,
-                                               &xip_mem, &xip_pfn);
-       if (unlikely(err)) {
-               if (err == -ENODATA)
-                       /* Hole? No need to truncate */
-                       return 0;
-               else
-                       return err;
-       }
-       memset(xip_mem + offset, 0, length);
-       return 0;
-}
-EXPORT_SYMBOL_GPL(xip_truncate_page);