path: root/fs/exofs/inode.c

/*
 * Copyright (C) 2005, 2006
 * Avishay Traeger (avishay@gmail.com)
 * Copyright (C) 2008, 2009
 * Boaz Harrosh <ooo@electrozaur.com>
 *
 * Copyrights for code taken from ext2:
 *     Copyright (C) 1992, 1993, 1994, 1995
 *     Remy Card (card@masi.ibp.fr)
 *     Laboratoire MASI - Institut Blaise Pascal
 *     Universite Pierre et Marie Curie (Paris VI)
 *     from
 *     linux/fs/minix/inode.c
 *     Copyright (C) 1991, 1992  Linus Torvalds
 *
 * This file is part of exofs.
 *
 * exofs is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation.  Since it is based on ext2, and the only
 * valid version of GPL for the Linux kernel is version 2, the only valid
 * version of GPL for exofs is version 2.
 *
 * exofs is distributed in the hope that 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 exofs; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */

#include <linux/slab.h>

#include "exofs.h"

#define EXOFS_DBGMSG2(M...) do {} while (0)

unsigned exofs_max_io_pages(struct ore_layout *layout,
			    unsigned expected_pages)
{
	unsigned pages = min_t(unsigned, expected_pages,
			       layout->max_io_length / PAGE_SIZE);

	return pages;
}

struct page_collect {
	struct exofs_sb_info *sbi;
	struct inode *inode;
	unsigned expected_pages;
	struct ore_io_state *ios;

	struct page **pages;
	unsigned alloc_pages;
	unsigned nr_pages;
	unsigned long length;
	loff_t pg_first; /* keep 64bit also in 32-arches */
	bool read_4_write; /* This means two things: that the read is sync
			    * And the pages should not be unlocked.
			    */
	struct page *that_locked_page;
};

static void _pcol_init(struct page_collect *pcol, unsigned expected_pages,
		       struct inode *inode)
{
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;

	pcol->sbi = sbi;
	pcol->inode = inode;
	pcol->expected_pages = expected_pages;

	pcol->ios = NULL;
	pcol->pages = NULL;
	pcol->alloc_pages = 0;
	pcol->nr_pages = 0;
	pcol->length = 0;
	pcol->pg_first = -1;
	pcol->read_4_write = false;
	pcol->that_locked_page = NULL;
}

static void _pcol_reset(struct page_collect *pcol)
{
	pcol->expected_pages -= min(pcol->nr_pages, pcol->expected_pages);

	pcol->pages = NULL;
	pcol->alloc_pages = 0;
	pcol->nr_pages = 0;
	pcol->length = 0;
	pcol->pg_first = -1;
	pcol->ios = NULL;
	pcol->that_locked_page = NULL;

	/* this is probably the end of the loop but in writes
	 * it might not end here. don't be left with nothing
	 */
	if (!pcol->expected_pages)
		pcol->expected_pages =
				exofs_max_io_pages(&pcol->sbi->layout, ~0);
}

static int pcol_try_alloc(struct page_collect *pcol)
{
	unsigned pages;

	/* TODO: easily support bio chaining */
	pages =  exofs_max_io_pages(&pcol->sbi->layout, pcol->expected_pages);

	for (; pages; pages >>= 1) {
		pcol->pages = kmalloc(pages * sizeof(struct page *),
				      GFP_KERNEL);
		if (likely(pcol->pages)) {
			pcol->alloc_pages = pages;
			return 0;
		}
	}

	EXOFS_ERR("Failed to kmalloc expected_pages=%u\n",
		  pcol->expected_pages);
	return -ENOMEM;
}

static void pcol_free(struct page_collect *pcol)
{
	kfree(pcol->pages);
	pcol->pages = NULL;

	if (pcol->ios) {
		ore_put_io_state(pcol->ios);
		pcol->ios = NULL;
	}
}

static int pcol_add_page(struct page_collect *pcol, struct page *page,
			 unsigned len)
{
	if (unlikely(pcol->nr_pages >= pcol->alloc_pages))
		return -ENOMEM;

	pcol->pages[pcol->nr_pages++] = page;
	pcol->length += len;
	return 0;
}

enum {PAGE_WAS_NOT_IN_IO = 17};
static int update_read_page(struct page *page, int ret)
{
	switch (ret) {
	case 0:
		/* Everything is OK */
		SetPageUptodate(page);
		if (PageError(page))
			ClearPageError(page);
		break;
	case -EFAULT:
		/* In this case we were trying to read something that wasn't on
		 * disk yet - return a page full of zeroes.  This should be OK,
		 * because the object should be empty (if there was a write
		 * before this read, the read would be waiting with the page
		 * locked */
		clear_highpage(page);

		SetPageUptodate(page);
		if (PageError(page))
			ClearPageError(page);
		EXOFS_DBGMSG("recovered read error\n");
		/* fall through */
	case PAGE_WAS_NOT_IN_IO:
		ret = 0; /* recovered error */
		break;
	default:
		SetPageError(page);
	}
	return ret;
}

static void update_write_page(struct page *page, int ret)
{
	if (unlikely(ret == PAGE_WAS_NOT_IN_IO))
		return; /* don't pass start don't collect $200 */

	if (ret) {
		mapping_set_error(page->mapping, ret);
		SetPageError(page);
	}
	end_page_writeback(page);
}

/* Called at the end of reads, to optionally unlock pages and update their
 * status.
 */
static int __readpages_done(struct page_collect *pcol)
{
	int i;
	u64 good_bytes;
	u64 length = 0;
	int ret = ore_check_io(pcol->ios, NULL);

	if (likely(!ret)) {
		good_bytes = pcol->length;
		ret = PAGE_WAS_NOT_IN_IO;
	} else {
		good_bytes = 0;
	}

	EXOFS_DBGMSG2("readpages_done(0x%lx) good_bytes=0x%llx"
		     " length=0x%lx nr_pages=%u\n",
		     pcol->inode->i_ino, _LLU(good_bytes), pcol->length,
		     pcol->nr_pages);

	for (i = 0; i < pcol->nr_pages; i++) {
		struct page *page = pcol->pages[i];
		struct inode *inode = page->mapping->host;
		int page_stat;

		if (inode != pcol->inode)
			continue; /* osd might add more pages at end */

		if (likely(length < good_bytes))
			page_stat = 0;
		else
			page_stat = ret;

		EXOFS_DBGMSG2("    readpages_done(0x%lx, 0x%lx) %s\n",
			  inode->i_ino, page->index,
			  page_stat ? "bad_bytes" : "good_bytes");

		ret = update_read_page(page, page_stat);
		if (!pcol->read_4_write)
			unlock_page(page);
		length += PAGE_SIZE;
	}

	pcol_free(pcol);
	EXOFS_DBGMSG2("readpages_done END\n");
	return ret;
}

/* callback of async reads */
static void readpages_done(struct ore_io_state *ios, void *p)
{
	struct page_collect *pcol = p;

	__readpages_done(pcol);
	atomic_dec(&pcol->sbi->s_curr_pending);
	kfree(pcol);
}

static void _unlock_pcol_pages(struct page_collect *pcol, int ret, int rw)
{
	int i;

	for (i = 0; i < pcol->nr_pages; i++) {
		struct page *page = pcol->pages[i];

		if (rw == READ)
			update_read_page(page, ret);
		else
			update_write_page(page, ret);

		unlock_page(page);
	}
}

static int _maybe_not_all_in_one_io(struct ore_io_state *ios,
	struct page_collect *pcol_src, struct page_collect *pcol)
{
	/* length was wrong or offset was not page aligned */
	BUG_ON(pcol_src->nr_pages < ios->nr_pages);

	if (pcol_src->nr_pages > ios->nr_pages) {
		struct page **src_page;
		unsigned pages_less = pcol_src->nr_pages - ios->nr_pages;
		unsigned long len_less = pcol_src->length - ios->length;
		unsigned i;
		int ret;

		/* This IO was trimmed */
		pcol_src->nr_pages = ios->nr_pages;
		pcol_src->length = ios->length;

		/* Left over pages are passed to the next io */
		pcol->expected_pages += pages_less;
		pcol->nr_pages = pages_less;
		pcol->length = len_less;
		src_page = pcol_src->pages + pcol_src->nr_pages;
		pcol->pg_first = (*src_page)->index;

		ret = pcol_try_alloc(pcol);
		if (unlikely(ret))
			return ret;

		for (i = 0; i < pages_less; ++i)
			pcol->pages[i] = *src_page++;

		EXOFS_DBGMSG("Length was adjusted nr_pages=0x%x "
			"pages_less=0x%x expected_pages=0x%x "
			"next_offset=0x%llx next_len=0x%lx\n",
			pcol_src->nr_pages, pages_less, pcol->expected_pages,
			pcol->pg_first * PAGE_SIZE, pcol->length);
	}
	return 0;
}

static int read_exec(struct pa