mirror of
https://github.com/torvalds/linux.git
synced 2024-12-23 11:21:33 +00:00
09cbfeaf1a
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
381 lines
10 KiB
C
381 lines
10 KiB
C
/* -*- linux-c -*- ------------------------------------------------------- *
|
|
*
|
|
* Copyright 2001 H. Peter Anvin - All Rights Reserved
|
|
*
|
|
* This program 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, Inc., 675 Mass Ave, Cambridge MA 02139,
|
|
* USA; either version 2 of the License, or (at your option) any later
|
|
* version; incorporated herein by reference.
|
|
*
|
|
* ----------------------------------------------------------------------- */
|
|
|
|
/*
|
|
* linux/fs/isofs/compress.c
|
|
*
|
|
* Transparent decompression of files on an iso9660 filesystem
|
|
*/
|
|
|
|
#include <linux/module.h>
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/vmalloc.h>
|
|
#include <linux/zlib.h>
|
|
|
|
#include "isofs.h"
|
|
#include "zisofs.h"
|
|
|
|
/* This should probably be global. */
|
|
static char zisofs_sink_page[PAGE_SIZE];
|
|
|
|
/*
|
|
* This contains the zlib memory allocation and the mutex for the
|
|
* allocation; this avoids failures at block-decompression time.
|
|
*/
|
|
static void *zisofs_zlib_workspace;
|
|
static DEFINE_MUTEX(zisofs_zlib_lock);
|
|
|
|
/*
|
|
* Read data of @inode from @block_start to @block_end and uncompress
|
|
* to one zisofs block. Store the data in the @pages array with @pcount
|
|
* entries. Start storing at offset @poffset of the first page.
|
|
*/
|
|
static loff_t zisofs_uncompress_block(struct inode *inode, loff_t block_start,
|
|
loff_t block_end, int pcount,
|
|
struct page **pages, unsigned poffset,
|
|
int *errp)
|
|
{
|
|
unsigned int zisofs_block_shift = ISOFS_I(inode)->i_format_parm[1];
|
|
unsigned int bufsize = ISOFS_BUFFER_SIZE(inode);
|
|
unsigned int bufshift = ISOFS_BUFFER_BITS(inode);
|
|
unsigned int bufmask = bufsize - 1;
|
|
int i, block_size = block_end - block_start;
|
|
z_stream stream = { .total_out = 0,
|
|
.avail_in = 0,
|
|
.avail_out = 0, };
|
|
int zerr;
|
|
int needblocks = (block_size + (block_start & bufmask) + bufmask)
|
|
>> bufshift;
|
|
int haveblocks;
|
|
blkcnt_t blocknum;
|
|
struct buffer_head *bhs[needblocks + 1];
|
|
int curbh, curpage;
|
|
|
|
if (block_size > deflateBound(1UL << zisofs_block_shift)) {
|
|
*errp = -EIO;
|
|
return 0;
|
|
}
|
|
/* Empty block? */
|
|
if (block_size == 0) {
|
|
for ( i = 0 ; i < pcount ; i++ ) {
|
|
if (!pages[i])
|
|
continue;
|
|
memset(page_address(pages[i]), 0, PAGE_SIZE);
|
|
flush_dcache_page(pages[i]);
|
|
SetPageUptodate(pages[i]);
|
|
}
|
|
return ((loff_t)pcount) << PAGE_SHIFT;
|
|
}
|
|
|
|
/* Because zlib is not thread-safe, do all the I/O at the top. */
|
|
blocknum = block_start >> bufshift;
|
|
memset(bhs, 0, (needblocks + 1) * sizeof(struct buffer_head *));
|
|
haveblocks = isofs_get_blocks(inode, blocknum, bhs, needblocks);
|
|
ll_rw_block(READ, haveblocks, bhs);
|
|
|
|
curbh = 0;
|
|
curpage = 0;
|
|
/*
|
|
* First block is special since it may be fractional. We also wait for
|
|
* it before grabbing the zlib mutex; odds are that the subsequent
|
|
* blocks are going to come in in short order so we don't hold the zlib
|
|
* mutex longer than necessary.
|
|
*/
|
|
|
|
if (!bhs[0])
|
|
goto b_eio;
|
|
|
|
wait_on_buffer(bhs[0]);
|
|
if (!buffer_uptodate(bhs[0])) {
|
|
*errp = -EIO;
|
|
goto b_eio;
|
|
}
|
|
|
|
stream.workspace = zisofs_zlib_workspace;
|
|
mutex_lock(&zisofs_zlib_lock);
|
|
|
|
zerr = zlib_inflateInit(&stream);
|
|
if (zerr != Z_OK) {
|
|
if (zerr == Z_MEM_ERROR)
|
|
*errp = -ENOMEM;
|
|
else
|
|
*errp = -EIO;
|
|
printk(KERN_DEBUG "zisofs: zisofs_inflateInit returned %d\n",
|
|
zerr);
|
|
goto z_eio;
|
|
}
|
|
|
|
while (curpage < pcount && curbh < haveblocks &&
|
|
zerr != Z_STREAM_END) {
|
|
if (!stream.avail_out) {
|
|
if (pages[curpage]) {
|
|
stream.next_out = page_address(pages[curpage])
|
|
+ poffset;
|
|
stream.avail_out = PAGE_SIZE - poffset;
|
|
poffset = 0;
|
|
} else {
|
|
stream.next_out = (void *)&zisofs_sink_page;
|
|
stream.avail_out = PAGE_SIZE;
|
|
}
|
|
}
|
|
if (!stream.avail_in) {
|
|
wait_on_buffer(bhs[curbh]);
|
|
if (!buffer_uptodate(bhs[curbh])) {
|
|
*errp = -EIO;
|
|
break;
|
|
}
|
|
stream.next_in = bhs[curbh]->b_data +
|
|
(block_start & bufmask);
|
|
stream.avail_in = min_t(unsigned, bufsize -
|
|
(block_start & bufmask),
|
|
block_size);
|
|
block_size -= stream.avail_in;
|
|
block_start = 0;
|
|
}
|
|
|
|
while (stream.avail_out && stream.avail_in) {
|
|
zerr = zlib_inflate(&stream, Z_SYNC_FLUSH);
|
|
if (zerr == Z_BUF_ERROR && stream.avail_in == 0)
|
|
break;
|
|
if (zerr == Z_STREAM_END)
|
|
break;
|
|
if (zerr != Z_OK) {
|
|
/* EOF, error, or trying to read beyond end of input */
|
|
if (zerr == Z_MEM_ERROR)
|
|
*errp = -ENOMEM;
|
|
else {
|
|
printk(KERN_DEBUG
|
|
"zisofs: zisofs_inflate returned"
|
|
" %d, inode = %lu,"
|
|
" page idx = %d, bh idx = %d,"
|
|
" avail_in = %ld,"
|
|
" avail_out = %ld\n",
|
|
zerr, inode->i_ino, curpage,
|
|
curbh, stream.avail_in,
|
|
stream.avail_out);
|
|
*errp = -EIO;
|
|
}
|
|
goto inflate_out;
|
|
}
|
|
}
|
|
|
|
if (!stream.avail_out) {
|
|
/* This page completed */
|
|
if (pages[curpage]) {
|
|
flush_dcache_page(pages[curpage]);
|
|
SetPageUptodate(pages[curpage]);
|
|
}
|
|
curpage++;
|
|
}
|
|
if (!stream.avail_in)
|
|
curbh++;
|
|
}
|
|
inflate_out:
|
|
zlib_inflateEnd(&stream);
|
|
|
|
z_eio:
|
|
mutex_unlock(&zisofs_zlib_lock);
|
|
|
|
b_eio:
|
|
for (i = 0; i < haveblocks; i++)
|
|
brelse(bhs[i]);
|
|
return stream.total_out;
|
|
}
|
|
|
|
/*
|
|
* Uncompress data so that pages[full_page] is fully uptodate and possibly
|
|
* fills in other pages if we have data for them.
|
|
*/
|
|
static int zisofs_fill_pages(struct inode *inode, int full_page, int pcount,
|
|
struct page **pages)
|
|
{
|
|
loff_t start_off, end_off;
|
|
loff_t block_start, block_end;
|
|
unsigned int header_size = ISOFS_I(inode)->i_format_parm[0];
|
|
unsigned int zisofs_block_shift = ISOFS_I(inode)->i_format_parm[1];
|
|
unsigned int blockptr;
|
|
loff_t poffset = 0;
|
|
blkcnt_t cstart_block, cend_block;
|
|
struct buffer_head *bh;
|
|
unsigned int blkbits = ISOFS_BUFFER_BITS(inode);
|
|
unsigned int blksize = 1 << blkbits;
|
|
int err;
|
|
loff_t ret;
|
|
|
|
BUG_ON(!pages[full_page]);
|
|
|
|
/*
|
|
* We want to read at least 'full_page' page. Because we have to
|
|
* uncompress the whole compression block anyway, fill the surrounding
|
|
* pages with the data we have anyway...
|
|
*/
|
|
start_off = page_offset(pages[full_page]);
|
|
end_off = min_t(loff_t, start_off + PAGE_SIZE, inode->i_size);
|
|
|
|
cstart_block = start_off >> zisofs_block_shift;
|
|
cend_block = (end_off + (1 << zisofs_block_shift) - 1)
|
|
>> zisofs_block_shift;
|
|
|
|
WARN_ON(start_off - (full_page << PAGE_SHIFT) !=
|
|
((cstart_block << zisofs_block_shift) & PAGE_MASK));
|
|
|
|
/* Find the pointer to this specific chunk */
|
|
/* Note: we're not using isonum_731() here because the data is known aligned */
|
|
/* Note: header_size is in 32-bit words (4 bytes) */
|
|
blockptr = (header_size + cstart_block) << 2;
|
|
bh = isofs_bread(inode, blockptr >> blkbits);
|
|
if (!bh)
|
|
return -EIO;
|
|
block_start = le32_to_cpu(*(__le32 *)
|
|
(bh->b_data + (blockptr & (blksize - 1))));
|
|
|
|
while (cstart_block < cend_block && pcount > 0) {
|
|
/* Load end of the compressed block in the file */
|
|
blockptr += 4;
|
|
/* Traversed to next block? */
|
|
if (!(blockptr & (blksize - 1))) {
|
|
brelse(bh);
|
|
|
|
bh = isofs_bread(inode, blockptr >> blkbits);
|
|
if (!bh)
|
|
return -EIO;
|
|
}
|
|
block_end = le32_to_cpu(*(__le32 *)
|
|
(bh->b_data + (blockptr & (blksize - 1))));
|
|
if (block_start > block_end) {
|
|
brelse(bh);
|
|
return -EIO;
|
|
}
|
|
err = 0;
|
|
ret = zisofs_uncompress_block(inode, block_start, block_end,
|
|
pcount, pages, poffset, &err);
|
|
poffset += ret;
|
|
pages += poffset >> PAGE_SHIFT;
|
|
pcount -= poffset >> PAGE_SHIFT;
|
|
full_page -= poffset >> PAGE_SHIFT;
|
|
poffset &= ~PAGE_MASK;
|
|
|
|
if (err) {
|
|
brelse(bh);
|
|
/*
|
|
* Did we finish reading the page we really wanted
|
|
* to read?
|
|
*/
|
|
if (full_page < 0)
|
|
return 0;
|
|
return err;
|
|
}
|
|
|
|
block_start = block_end;
|
|
cstart_block++;
|
|
}
|
|
|
|
if (poffset && *pages) {
|
|
memset(page_address(*pages) + poffset, 0,
|
|
PAGE_SIZE - poffset);
|
|
flush_dcache_page(*pages);
|
|
SetPageUptodate(*pages);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* When decompressing, we typically obtain more than one page
|
|
* per reference. We inject the additional pages into the page
|
|
* cache as a form of readahead.
|
|
*/
|
|
static int zisofs_readpage(struct file *file, struct page *page)
|
|
{
|
|
struct inode *inode = file_inode(file);
|
|
struct address_space *mapping = inode->i_mapping;
|
|
int err;
|
|
int i, pcount, full_page;
|
|
unsigned int zisofs_block_shift = ISOFS_I(inode)->i_format_parm[1];
|
|
unsigned int zisofs_pages_per_cblock =
|
|
PAGE_SHIFT <= zisofs_block_shift ?
|
|
(1 << (zisofs_block_shift - PAGE_SHIFT)) : 0;
|
|
struct page *pages[max_t(unsigned, zisofs_pages_per_cblock, 1)];
|
|
pgoff_t index = page->index, end_index;
|
|
|
|
end_index = (inode->i_size + PAGE_SIZE - 1) >> PAGE_SHIFT;
|
|
/*
|
|
* If this page is wholly outside i_size we just return zero;
|
|
* do_generic_file_read() will handle this for us
|
|
*/
|
|
if (index >= end_index) {
|
|
SetPageUptodate(page);
|
|
unlock_page(page);
|
|
return 0;
|
|
}
|
|
|
|
if (PAGE_SHIFT <= zisofs_block_shift) {
|
|
/* We have already been given one page, this is the one
|
|
we must do. */
|
|
full_page = index & (zisofs_pages_per_cblock - 1);
|
|
pcount = min_t(int, zisofs_pages_per_cblock,
|
|
end_index - (index & ~(zisofs_pages_per_cblock - 1)));
|
|
index -= full_page;
|
|
} else {
|
|
full_page = 0;
|
|
pcount = 1;
|
|
}
|
|
pages[full_page] = page;
|
|
|
|
for (i = 0; i < pcount; i++, index++) {
|
|
if (i != full_page)
|
|
pages[i] = grab_cache_page_nowait(mapping, index);
|
|
if (pages[i]) {
|
|
ClearPageError(pages[i]);
|
|
kmap(pages[i]);
|
|
}
|
|
}
|
|
|
|
err = zisofs_fill_pages(inode, full_page, pcount, pages);
|
|
|
|
/* Release any residual pages, do not SetPageUptodate */
|
|
for (i = 0; i < pcount; i++) {
|
|
if (pages[i]) {
|
|
flush_dcache_page(pages[i]);
|
|
if (i == full_page && err)
|
|
SetPageError(pages[i]);
|
|
kunmap(pages[i]);
|
|
unlock_page(pages[i]);
|
|
if (i != full_page)
|
|
put_page(pages[i]);
|
|
}
|
|
}
|
|
|
|
/* At this point, err contains 0 or -EIO depending on the "critical" page */
|
|
return err;
|
|
}
|
|
|
|
const struct address_space_operations zisofs_aops = {
|
|
.readpage = zisofs_readpage,
|
|
/* No sync_page operation supported? */
|
|
/* No bmap operation supported */
|
|
};
|
|
|
|
int __init zisofs_init(void)
|
|
{
|
|
zisofs_zlib_workspace = vmalloc(zlib_inflate_workspacesize());
|
|
if ( !zisofs_zlib_workspace )
|
|
return -ENOMEM;
|
|
|
|
return 0;
|
|
}
|
|
|
|
void zisofs_cleanup(void)
|
|
{
|
|
vfree(zisofs_zlib_workspace);
|
|
}
|