On Monday 14 July 2003 20:25, [EMAIL PROTECTED] wrote:
> The patch below is just cleanup.
>
> 1) page_offset as calculated in __get_metapage is always 0.
> Please go through the variable substitution/calculation
> to verify.
If the PAGE_CACHE_SIZE is greater than 4096, page_offset may be greater
than zero. However, JFS is currently broken with a larger
PAGE_CACHE_SIZE, but I mean to get that working some day.
> 2) mp = NULL isn't necessary because that statement is surrounded
> by "if (!mp) {".
True.
> 3) if statement re-ordered so else path is usually not taken.
Right. I like this better. Testing for the negation and then having an
else clause is ugly.
> Tested successfully with dbench/tiobench/bonnie++ on 2.5.75
> uniprocessor x86.
>
>
>
> --- jfs_metapage.c.orig 2003-02-26 21:46:13.000000000 -0500
> +++ jfs_metapage.c 2003-07-14 20:54:15.000000000 -0400
> @@ -213,12 +213,8 @@
> unsigned long new)
> {
> struct metapage **hash_ptr;
> - int l2BlocksPerPage;
> - int l2bsize;
> struct address_space *mapping;
> struct metapage *mp;
> - unsigned long page_index;
> - unsigned long page_offset;
>
> jfs_info("__get_metapage: inode = 0x%p, lblock = 0x%lx", inode,
> lblock);
>
> @@ -244,12 +240,7 @@
> lock_metapage(mp);
> spin_unlock(&meta_lock);
> } else {
> - l2bsize = inode->i_blkbits;
> - l2BlocksPerPage = PAGE_CACHE_SHIFT - l2bsize;
> - page_index = lblock >> l2BlocksPerPage;
> - page_offset = (lblock - (page_index << l2BlocksPerPage)) <<
> - l2bsize;
> - if ((page_offset + size) > PAGE_CACHE_SIZE) {
> + if (size > PAGE_CACHE_SIZE) {
> spin_unlock(&meta_lock);
> jfs_err("MetaData crosses page boundary!!");
> return NULL;
> @@ -263,7 +254,6 @@
> * Attempt to get metapage without blocking, tapping into
> * reserves if necessary.
> */
> - mp = NULL;
> if (JFS_IP(inode)->fileset == AGGREGATE_I) {
> mp = mempool_alloc(metapage_mempool, GFP_ATOMIC);
> if (!mp) {
> @@ -311,13 +301,13 @@
>
> if (new) {
> jfs_info("__get_metapage: Calling grab_cache_page");
> - mp->page = grab_cache_page(mapping, page_index);
> - if (!mp->page) {
> - jfs_err("grab_cache_page failed!");
> - goto freeit;
> - } else {
> + mp->page = grab_cache_page(mapping, lblock >>
> (PAGE_CACHE_SHIFT -
> (inode->i_blkbits))); + if (mp->page) {
> INCREMENT(mpStat.pagealloc);
> unlock_page(mp->page);
> + } else {
> + jfs_err("grab_cache_page failed!");
> + goto freeit;
> }
> } else {
> jfs_info("__get_metapage: Calling read_cache_page");
> @@ -329,7 +319,7 @@
> } else
> INCREMENT(mpStat.pagealloc);
> }
> - mp->data = kmap(mp->page) + page_offset;
> + mp->data = kmap(mp->page);
> }
> jfs_info("__get_metapage: returning = 0x%p", mp);
> return mp;
--
David Kleikamp
IBM Linux Technology Center
_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion
