[Patch] shmem truncate optimizations and cleanup

From: Christoph Rohland (cr@sap.com)
Date: Wed Jan 10 2001 - 10:22:13 EST


Hi Alan,

The appended patch speeds up the truncate logic of shmem.c
considerably and makes it more readable.

Would you apply it to your -ac series?

I will go on with some cache lookup optimizations and probably
read/write support.

Greetings
                Christoph

diff -uNr 4-13-5/include/linux/shmem_fs.h c/include/linux/shmem_fs.h
--- 4-13-5/include/linux/shmem_fs.h Thu Dec 28 23:19:27 2000
+++ c/include/linux/shmem_fs.h Thu Dec 28 23:22:58 2000
@@ -19,6 +19,7 @@
 
 struct shmem_inode_info {
         spinlock_t lock;
+ unsigned long max_index;
         swp_entry_t i_direct[SHMEM_NR_DIRECT]; /* for the first blocks */
         swp_entry_t **i_indirect; /* doubly indirect blocks */
         unsigned long swapped;
diff -uNr 4-13-5/mm/shmem.c c/mm/shmem.c
--- 4-13-5/mm/shmem.c Thu Dec 28 23:19:27 2000
+++ c/mm/shmem.c Thu Dec 28 23:31:20 2000
@@ -51,11 +51,16 @@
 
 static swp_entry_t * shmem_swp_entry (struct shmem_inode_info *info, unsigned long index)
 {
+ unsigned long offset;
+
         if (index < SHMEM_NR_DIRECT)
                 return info->i_direct+index;
 
         index -= SHMEM_NR_DIRECT;
- if (index >= ENTRIES_PER_PAGE*ENTRIES_PER_PAGE)
+ offset = index % ENTRIES_PER_PAGE;
+ index /= ENTRIES_PER_PAGE;
+
+ if (index >= ENTRIES_PER_PAGE)
                 return NULL;
 
         if (!info->i_indirect) {
@@ -63,13 +68,13 @@
                 if (!info->i_indirect)
                         return NULL;
         }
- if(!(info->i_indirect[index/ENTRIES_PER_PAGE])) {
- info->i_indirect[index/ENTRIES_PER_PAGE] = (swp_entry_t *) get_zeroed_page(GFP_USER);
- if (!info->i_indirect[index/ENTRIES_PER_PAGE])
+ if(!(info->i_indirect[index])) {
+ info->i_indirect[index] = (swp_entry_t *) get_zeroed_page(GFP_USER);
+ if (!info->i_indirect[index])
                         return NULL;
         }
         
- return info->i_indirect[index/ENTRIES_PER_PAGE]+index%ENTRIES_PER_PAGE;
+ return info->i_indirect[index]+offset;
 }
 
 static int shmem_free_swp(swp_entry_t *dir, unsigned int count)
@@ -99,7 +104,6 @@
  * @dir: pointer to swp_entries
  * @size: number of entries in dir
  * @start: offset to start from
- * @inode: inode for statistics
  * @freed: counter for freed pages
  *
  * It frees the swap entries from dir+start til dir+size
@@ -109,7 +113,7 @@
 
 static unsigned long
 shmem_truncate_part (swp_entry_t * dir, unsigned long size,
- unsigned long start, struct inode * inode, unsigned long *freed) {
+ unsigned long start, unsigned long *freed) {
         if (start > size)
                 return start - size;
         if (dir)
@@ -121,21 +125,27 @@
 static void shmem_truncate (struct inode * inode)
 {
         int clear_base;
- unsigned long start;
+ unsigned long index, start;
         unsigned long mmfreed, freed = 0;
- swp_entry_t **base, **ptr;
+ swp_entry_t **base, **ptr, **last;
         struct shmem_inode_info * info = &inode->u.shmem_i;
 
         spin_lock (&info->lock);
- start = (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+ index = (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+ if (index >= info->max_index) {
+ info->max_index = index;
+ spin_unlock (&info->lock);
+ return;
+ }
 
- start = shmem_truncate_part (info->i_direct, SHMEM_NR_DIRECT, start, inode, &freed);
+ start = shmem_truncate_part (info->i_direct, SHMEM_NR_DIRECT, index, &freed);
 
         if (!(base = info->i_indirect))
- goto out;;
+ goto out;
 
         clear_base = 1;
- for (ptr = base; ptr < base + ENTRIES_PER_PAGE; ptr++) {
+ last = base + ((info->max_index - SHMEM_NR_DIRECT + ENTRIES_PER_PAGE - 1) / ENTRIES_PER_PAGE);
+ for (ptr = base; ptr < last; ptr++) {
                 if (!start) {
                         if (!*ptr)
                                 continue;
@@ -145,16 +155,16 @@
                         continue;
                 }
                 clear_base = 0;
- start = shmem_truncate_part (*ptr, ENTRIES_PER_PAGE, start, inode, &freed);
+ start = shmem_truncate_part (*ptr, ENTRIES_PER_PAGE, start, &freed);
         }
 
- if (!clear_base)
- goto out;
-
- free_page ((unsigned long)base);
- info->i_indirect = 0;
+ if (clear_base) {
+ free_page ((unsigned long)base);
+ info->i_indirect = 0;
+ }
 
 out:
+ info->max_index = index;
 
         /*
          * We have to calculate the free blocks since we do not know
@@ -209,16 +219,16 @@
                 return 1;
 
         spin_lock(&info->lock);
- entry = shmem_swp_entry (info, page->index);
+ entry = shmem_swp_entry(info, page->index);
         if (!entry) /* this had been allocted on page allocation */
                 BUG();
         error = -EAGAIN;
         if (entry->val) {
- __swap_free(swap, 2);
+ __swap_free(swap, 2);
                 goto out;
- }
+ }
 
- *entry = swap;
+ *entry = swap;
         error = 0;
         /* Remove the from the page cache */
         lru_cache_del(page);

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon Jan 15 2001 - 21:00:26 EST