From 59bc055211b8d266ab6089158058bf8268e02006 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Wed, 23 Sep 2009 14:44:56 +0200 Subject: zisofs: Implement reading of compressed files when PAGE_CACHE_SIZE > compress block size Also split and cleanup zisofs_readpage() when we are changing it anyway. Signed-off-by: Jan Kara --- fs/isofs/compress.c | 533 ++++++++++++++++++++++++++++------------------------ fs/isofs/rock.c | 3 +- 2 files changed, 286 insertions(+), 250 deletions(-) (limited to 'fs/isofs') diff --git a/fs/isofs/compress.c b/fs/isofs/compress.c index defb932..0b3fa79 100644 --- a/fs/isofs/compress.c +++ b/fs/isofs/compress.c @@ -36,286 +36,323 @@ static void *zisofs_zlib_workspace; static DEFINE_MUTEX(zisofs_zlib_lock); /* - * 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. + * 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 int zisofs_readpage(struct file *file, struct page *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) { - struct inode *inode = file->f_path.dentry->d_inode; - struct address_space *mapping = inode->i_mapping; - unsigned int maxpage, xpage, fpage, blockindex; - unsigned long offset; - unsigned long blockptr, blockendptr, cstart, cend, csize; - struct buffer_head *bh, *ptrbh[2]; - unsigned long bufsize = ISOFS_BUFFER_SIZE(inode); - unsigned int bufshift = ISOFS_BUFFER_BITS(inode); - unsigned long bufmask = bufsize - 1; - int err = -EIO; - int i; - unsigned int header_size = ISOFS_I(inode)->i_format_parm[0]; unsigned int zisofs_block_shift = ISOFS_I(inode)->i_format_parm[1]; - /* unsigned long zisofs_block_size = 1UL << zisofs_block_shift; */ - unsigned int zisofs_block_page_shift = zisofs_block_shift-PAGE_CACHE_SHIFT; - unsigned long zisofs_block_pages = 1UL << zisofs_block_page_shift; - unsigned long zisofs_block_page_mask = zisofs_block_pages-1; - struct page *pages[zisofs_block_pages]; - unsigned long index = page->index; - int indexblocks; - - /* We have already been given one page, this is the one - we must do. */ - xpage = index & zisofs_block_page_mask; - pages[xpage] = page; - - /* The remaining pages need to be allocated and inserted */ - offset = index & ~zisofs_block_page_mask; - blockindex = offset >> zisofs_block_page_shift; - maxpage = (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; - - /* - * If this page is wholly outside i_size we just return zero; - * do_generic_file_read() will handle this for us - */ - if (page->index >= maxpage) { - SetPageUptodate(page); - unlock_page(page); + 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; } - - maxpage = min(zisofs_block_pages, maxpage-offset); - - for ( i = 0 ; i < maxpage ; i++, offset++ ) { - if ( i != xpage ) { - pages[i] = grab_cache_page_nowait(mapping, offset); - } - page = pages[i]; - if ( page ) { - ClearPageError(page); - kmap(page); + /* Empty block? */ + if (block_size == 0) { + for ( i = 0 ; i < pcount ; i++ ) { + if (!pages[i]) + continue; + memset(page_address(pages[i]), 0, PAGE_CACHE_SIZE); + flush_dcache_page(pages[i]); + SetPageUptodate(pages[i]); } + return ((loff_t)pcount) << PAGE_CACHE_SHIFT; } - /* This is the last page filled, plus one; used in case of abort. */ - fpage = 0; + /* 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); - /* 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 + blockindex) << 2; - blockendptr = blockptr + 4; + 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. + */ - indexblocks = ((blockptr^blockendptr) >> bufshift) ? 2 : 1; - ptrbh[0] = ptrbh[1] = NULL; + if (!bhs[0]) + goto b_eio; - if ( isofs_get_blocks(inode, blockptr >> bufshift, ptrbh, indexblocks) != indexblocks ) { - if ( ptrbh[0] ) brelse(ptrbh[0]); - printk(KERN_DEBUG "zisofs: Null buffer on reading block table, inode = %lu, block = %lu\n", - inode->i_ino, blockptr >> bufshift); - goto eio; - } - ll_rw_block(READ, indexblocks, ptrbh); - - bh = ptrbh[0]; - if ( !bh || (wait_on_buffer(bh), !buffer_uptodate(bh)) ) { - printk(KERN_DEBUG "zisofs: Failed to read block table, inode = %lu, block = %lu\n", - inode->i_ino, blockptr >> bufshift); - if ( ptrbh[1] ) - brelse(ptrbh[1]); - goto eio; - } - cstart = le32_to_cpu(*(__le32 *)(bh->b_data + (blockptr & bufmask))); - - if ( indexblocks == 2 ) { - /* We just crossed a block boundary. Switch to the next block */ - brelse(bh); - bh = ptrbh[1]; - if ( !bh || (wait_on_buffer(bh), !buffer_uptodate(bh)) ) { - printk(KERN_DEBUG "zisofs: Failed to read block table, inode = %lu, block = %lu\n", - inode->i_ino, blockendptr >> bufshift); - goto eio; - } + wait_on_buffer(bhs[0]); + if (!buffer_uptodate(bhs[0])) { + *errp = -EIO; + goto b_eio; } - cend = le32_to_cpu(*(__le32 *)(bh->b_data + (blockendptr & bufmask))); - brelse(bh); - if (cstart > cend) - goto eio; + stream.workspace = zisofs_zlib_workspace; + mutex_lock(&zisofs_zlib_lock); - csize = cend-cstart; - - if (csize > deflateBound(1UL << zisofs_block_shift)) - goto eio; - - /* Now page[] contains an array of pages, any of which can be NULL, - and the locks on which we hold. We should now read the data and - release the pages. If the pages are NULL the decompressed data - for that particular page should be discarded. */ - - if ( csize == 0 ) { - /* This data block is empty. */ - - for ( fpage = 0 ; fpage < maxpage ; fpage++ ) { - if ( (page = pages[fpage]) != NULL ) { - memset(page_address(page), 0, PAGE_CACHE_SIZE); - - flush_dcache_page(page); - SetPageUptodate(page); - kunmap(page); - unlock_page(page); - if ( fpage == xpage ) - err = 0; /* The critical page */ - else - page_cache_release(page); + 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_CACHE_SIZE - poffset; + poffset = 0; + } else { + stream.next_out = (void *)&zisofs_sink_page; + stream.avail_out = PAGE_CACHE_SIZE; } } - } else { - /* This data block is compressed. */ - z_stream stream; - int bail = 0, left_out = -1; - int zerr; - int needblocks = (csize + (cstart & bufmask) + bufmask) >> bufshift; - int haveblocks; - struct buffer_head *bhs[needblocks+1]; - struct buffer_head **bhptr; - - /* Because zlib is not thread-safe, do all the I/O at the top. */ - - blockptr = cstart >> bufshift; - memset(bhs, 0, (needblocks+1)*sizeof(struct buffer_head *)); - haveblocks = isofs_get_blocks(inode, blockptr, bhs, needblocks); - ll_rw_block(READ, haveblocks, bhs); - - bhptr = &bhs[0]; - bh = *bhptr++; - - /* 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 ( !bh || (wait_on_buffer(bh), !buffer_uptodate(bh)) ) { - printk(KERN_DEBUG "zisofs: Hit null buffer, fpage = %d, xpage = %d, csize = %ld\n", - fpage, xpage, csize); - goto b_eio; - } - stream.next_in = bh->b_data + (cstart & bufmask); - stream.avail_in = min(bufsize-(cstart & bufmask), csize); - csize -= stream.avail_in; - - stream.workspace = zisofs_zlib_workspace; - mutex_lock(&zisofs_zlib_lock); - - zerr = zlib_inflateInit(&stream); - if ( zerr != Z_OK ) { - if ( err && zerr == Z_MEM_ERROR ) - err = -ENOMEM; - printk(KERN_DEBUG "zisofs: zisofs_inflateInit returned %d\n", - zerr); - goto z_eio; + 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 ( !bail && fpage < maxpage ) { - page = pages[fpage]; - if ( page ) - stream.next_out = page_address(page); - else - stream.next_out = (void *)&zisofs_sink_page; - stream.avail_out = PAGE_CACHE_SIZE; - - while ( stream.avail_out ) { - int ao, ai; - if ( stream.avail_in == 0 && left_out ) { - if ( !csize ) { - printk(KERN_WARNING "zisofs: ZF read beyond end of input\n"); - bail = 1; - break; - } else { - bh = *bhptr++; - if ( !bh || - (wait_on_buffer(bh), !buffer_uptodate(bh)) ) { - /* Reached an EIO */ - printk(KERN_DEBUG "zisofs: Hit null buffer, fpage = %d, xpage = %d, csize = %ld\n", - fpage, xpage, csize); - - bail = 1; - break; - } - stream.next_in = bh->b_data; - stream.avail_in = min(csize,bufsize); - csize -= stream.avail_in; - } - } - ao = stream.avail_out; ai = stream.avail_in; - zerr = zlib_inflate(&stream, Z_SYNC_FLUSH); - left_out = stream.avail_out; - if ( zerr == Z_BUF_ERROR && stream.avail_in == 0 ) - continue; - if ( zerr != Z_OK ) { - /* EOF, error, or trying to read beyond end of input */ - if ( err && zerr == Z_MEM_ERROR ) - err = -ENOMEM; - if ( zerr != Z_STREAM_END ) - printk(KERN_DEBUG "zisofs: zisofs_inflate returned %d, inode = %lu, index = %lu, fpage = %d, xpage = %d, avail_in = %d, avail_out = %d, ai = %d, ao = %d\n", - zerr, inode->i_ino, index, - fpage, xpage, - stream.avail_in, stream.avail_out, - ai, ao); - bail = 1; - break; + 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 = %d," + " avail_out = %d\n", + zerr, inode->i_ino, curpage, + curbh, stream.avail_in, + stream.avail_out); + *errp = -EIO; } + goto inflate_out; } + } - if ( stream.avail_out && zerr == Z_STREAM_END ) { - /* Fractional page written before EOF. This may - be the last page in the file. */ - memset(stream.next_out, 0, stream.avail_out); - stream.avail_out = 0; + 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); - if ( !stream.avail_out ) { - /* This page completed */ - if ( page ) { - flush_dcache_page(page); - SetPageUptodate(page); - kunmap(page); - unlock_page(page); - if ( fpage == xpage ) - err = 0; /* The critical page */ - else - page_cache_release(page); - } - fpage++; - } +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_CACHE_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_CACHE_SHIFT) != + ((cstart_block << zisofs_block_shift) & PAGE_CACHE_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_CACHE_SHIFT; + pcount -= poffset >> PAGE_CACHE_SHIFT; + full_page -= poffset >> PAGE_CACHE_SHIFT; + poffset &= ~PAGE_CACHE_MASK; + + if (err) { + brelse(bh); + /* + * Did we finish reading the page we really wanted + * to read? + */ + if (full_page < 0) + return 0; + return err; } - zlib_inflateEnd(&stream); - z_eio: - mutex_unlock(&zisofs_zlib_lock); + block_start = block_end; + cstart_block++; + } + + if (poffset && *pages) { + memset(page_address(*pages) + poffset, 0, + PAGE_CACHE_SIZE - poffset); + flush_dcache_page(*pages); + SetPageUptodate(*pages); + } + return 0; +} - b_eio: - for ( i = 0 ; i < haveblocks ; i++ ) { - if ( bhs[i] ) - brelse(bhs[i]); +/* + * 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->f_path.dentry->d_inode; + 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_CACHE_SHIFT <= zisofs_block_shift ? + (1 << (zisofs_block_shift - PAGE_CACHE_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_CACHE_SIZE - 1) >> PAGE_CACHE_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_CACHE_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]); } } -eio: + err = zisofs_fill_pages(inode, full_page, pcount, pages); /* Release any residual pages, do not SetPageUptodate */ - while ( fpage < maxpage ) { - page = pages[fpage]; - if ( page ) { - flush_dcache_page(page); - if ( fpage == xpage ) - SetPageError(page); - kunmap(page); - unlock_page(page); - if ( fpage != xpage ) - page_cache_release(page); + 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) + page_cache_release(pages[i]); } - fpage++; } /* At this point, err contains 0 or -EIO depending on the "critical" page */ diff --git a/fs/isofs/rock.c b/fs/isofs/rock.c index c2fb2dd..96a685c 100644 --- a/fs/isofs/rock.c +++ b/fs/isofs/rock.c @@ -518,8 +518,7 @@ repeat: if (algo == SIG('p', 'z')) { int block_shift = isonum_711(&rr->u.ZF.parms[1]); - if (block_shift < PAGE_CACHE_SHIFT - || block_shift > 17) { + if (block_shift > 17) { printk(KERN_WARNING "isofs: " "Can't handle ZF block " "size of 2^%d\n", -- cgit v1.1