static int ocfs2_local_alloc_find_clear_bits(struct ocfs2_super *osb,
struct ocfs2_dinode *alloc,
- u32 numbits,
+ u32 *numbits,
struct ocfs2_alloc_reservation *resv);
static void ocfs2_clear_local_alloc(struct ocfs2_dinode *alloc);
static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
struct inode *local_alloc_inode);
+/*
+ * ocfs2_la_default_mb() - determine a default size, in megabytes of
+ * the local alloc.
+ *
+ * Generally, we'd like to pick as large a local alloc as
+ * possible. Performance on large workloads tends to scale
+ * proportionally to la size. In addition to that, the reservations
+ * code functions more efficiently as it can reserve more windows for
+ * write.
+ *
+ * Some things work against us when trying to choose a large local alloc:
+ *
+ * - We need to ensure our sizing is picked to leave enough space in
+ * group descriptors for other allocations (such as block groups,
+ * etc). Picking default sizes which are a multiple of 4 could help
+ * - block groups are allocated in 2mb and 4mb chunks.
+ *
+ * - Likewise, we don't want to starve other nodes of bits on small
+ * file systems. This can easily be taken care of by limiting our
+ * default to a reasonable size (256M) on larger cluster sizes.
+ *
+ * - Some file systems can't support very large sizes - 4k and 8k in
+ * particular are limited to less than 128 and 256 megabytes respectively.
+ *
+ * The following reference table shows group descriptor and local
+ * alloc maximums at various cluster sizes (4k blocksize)
+ *
+ * csize: 4K group: 126M la: 121M
+ * csize: 8K group: 252M la: 243M
+ * csize: 16K group: 504M la: 486M
+ * csize: 32K group: 1008M la: 972M
+ * csize: 64K group: 2016M la: 1944M
+ * csize: 128K group: 4032M la: 3888M
+ * csize: 256K group: 8064M la: 7776M
+ * csize: 512K group: 16128M la: 15552M
+ * csize: 1024K group: 32256M la: 31104M
+ */
+#define OCFS2_LA_MAX_DEFAULT_MB 256
+#define OCFS2_LA_OLD_DEFAULT 8
+unsigned int ocfs2_la_default_mb(struct ocfs2_super *osb)
+{
+ unsigned int la_mb;
+ unsigned int gd_mb;
+ unsigned int megs_per_slot;
+ struct super_block *sb = osb->sb;
+
+ gd_mb = ocfs2_clusters_to_megabytes(osb->sb,
+ 8 * ocfs2_group_bitmap_size(sb, 0, osb->s_feature_incompat));
+
+ /*
+ * This takes care of files systems with very small group
+ * descriptors - 512 byte blocksize at cluster sizes lower
+ * than 16K and also 1k blocksize with 4k cluster size.
+ */
+ if ((sb->s_blocksize == 512 && osb->s_clustersize <= 8192)
+ || (sb->s_blocksize == 1024 && osb->s_clustersize == 4096))
+ return OCFS2_LA_OLD_DEFAULT;
+
+ /*
+ * Leave enough room for some block groups and make the final
+ * value we work from a multiple of 4.
+ */
+ gd_mb -= 16;
+ gd_mb &= 0xFFFFFFFB;
+
+ la_mb = gd_mb;
+
+ /*
+ * Keep window sizes down to a reasonable default
+ */
+ if (la_mb > OCFS2_LA_MAX_DEFAULT_MB) {
+ /*
+ * Some clustersize / blocksize combinations will have
+ * given us a larger than OCFS2_LA_MAX_DEFAULT_MB
+ * default size, but get poor distribution when
+ * limited to exactly 256 megabytes.
+ *
+ * As an example, 16K clustersize at 4K blocksize
+ * gives us a cluster group size of 504M. Paring the
+ * local alloc size down to 256 however, would give us
+ * only one window and around 200MB left in the
+ * cluster group. Instead, find the first size below
+ * 256 which would give us an even distribution.
+ *
+ * Larger cluster group sizes actually work out pretty
+ * well when pared to 256, so we don't have to do this
+ * for any group that fits more than two
+ * OCFS2_LA_MAX_DEFAULT_MB windows.
+ */
+ if (gd_mb > (2 * OCFS2_LA_MAX_DEFAULT_MB))
+ la_mb = 256;
+ else {
+ unsigned int gd_mult = gd_mb;
+
+ while (gd_mult > 256)
+ gd_mult = gd_mult >> 1;
+
+ la_mb = gd_mult;
+ }
+ }
+
+ megs_per_slot = osb->osb_clusters_at_boot / osb->max_slots;
+ megs_per_slot = ocfs2_clusters_to_megabytes(osb->sb, megs_per_slot);
+ /* Too many nodes, too few disk clusters. */
+ if (megs_per_slot < la_mb)
+ la_mb = megs_per_slot;
+
+ return la_mb;
+}
+
+void ocfs2_la_set_sizes(struct ocfs2_super *osb, int requested_mb)
+{
+ struct super_block *sb = osb->sb;
+ unsigned int la_default_mb = ocfs2_la_default_mb(osb);
+ unsigned int la_max_mb;
+
+ la_max_mb = ocfs2_clusters_to_megabytes(sb,
+ ocfs2_local_alloc_size(sb) * 8);
+
+ mlog(0, "requested: %dM, max: %uM, default: %uM\n",
+ requested_mb, la_max_mb, la_default_mb);
+
+ if (requested_mb == -1) {
+ /* No user request - use defaults */
+ osb->local_alloc_default_bits =
+ ocfs2_megabytes_to_clusters(sb, la_default_mb);
+ } else if (requested_mb > la_max_mb) {
+ /* Request is too big, we give the maximum available */
+ osb->local_alloc_default_bits =
+ ocfs2_megabytes_to_clusters(sb, la_max_mb);
+ } else {
+ osb->local_alloc_default_bits =
+ ocfs2_megabytes_to_clusters(sb, requested_mb);
+ }
+
+ osb->local_alloc_bits = osb->local_alloc_default_bits;
+}
+
static inline int ocfs2_la_state_enabled(struct ocfs2_super *osb)
{
return (osb->local_alloc_state == OCFS2_LA_THROTTLED ||
osb->local_alloc_bits, (osb->bitmap_cpg - 1));
osb->local_alloc_bits =
ocfs2_megabytes_to_clusters(osb->sb,
- OCFS2_DEFAULT_LOCAL_ALLOC_SIZE);
+ ocfs2_la_default_mb(osb));
}
/* read the alloc off disk */
}
ocfs2_clear_local_alloc(alloc);
-
- status = ocfs2_journal_dirty(handle, bh);
- if (status < 0) {
- mlog_errno(status);
- goto out_commit;
- }
+ ocfs2_journal_dirty(handle, bh);
brelse(bh);
osb->local_alloc_bh = NULL;
return status;
}
-/* Check to see if the local alloc window is within ac->ac_max_block */
-static int ocfs2_local_alloc_in_range(struct inode *inode,
- struct ocfs2_alloc_context *ac,
- u32 bits_wanted)
-{
- struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
- struct ocfs2_dinode *alloc;
- struct ocfs2_local_alloc *la;
- int start;
- u64 block_off;
-
- if (!ac->ac_max_block)
- return 1;
-
- alloc = (struct ocfs2_dinode *) osb->local_alloc_bh->b_data;
- la = OCFS2_LOCAL_ALLOC(alloc);
-
- start = ocfs2_local_alloc_find_clear_bits(osb, alloc, bits_wanted, NULL);
- if (start == -1) {
- mlog_errno(-ENOSPC);
- return 0;
- }
-
- /*
- * Converting (bm_off + start + bits_wanted) to blocks gives us
- * the blkno just past our actual allocation. This is perfect
- * to compare with ac_max_block.
- */
- block_off = ocfs2_clusters_to_blocks(inode->i_sb,
- le32_to_cpu(la->la_bm_off) +
- start + bits_wanted);
- mlog(0, "Checking %llu against %llu\n",
- (unsigned long long)block_off,
- (unsigned long long)ac->ac_max_block);
- if (block_off > ac->ac_max_block)
- return 0;
-
- return 1;
-}
-
/*
* make sure we've got at least bits_wanted contiguous bits in the
* local alloc. You lose them when you drop i_mutex.
mlog(0, "Calling in_range for max block %llu\n",
(unsigned long long)ac->ac_max_block);
- if (!ocfs2_local_alloc_in_range(local_alloc_inode, ac,
- bits_wanted)) {
- /*
- * The window is outside ac->ac_max_block.
- * This errno tells the caller to keep localalloc enabled
- * but to get the allocation from the main bitmap.
- */
- status = -EFBIG;
- goto bail;
- }
-
ac->ac_inode = local_alloc_inode;
/* We should never use localalloc from another slot */
ac->ac_alloc_slot = osb->slot_num;
alloc = (struct ocfs2_dinode *) osb->local_alloc_bh->b_data;
la = OCFS2_LOCAL_ALLOC(alloc);
- start = ocfs2_local_alloc_find_clear_bits(osb, alloc, bits_wanted,
+ start = ocfs2_local_alloc_find_clear_bits(osb, alloc, &bits_wanted,
ac->ac_resv);
if (start == -1) {
/* TODO: Shouldn't we just BUG here? */
bitmap = la->la_bitmap;
*bit_off = le32_to_cpu(la->la_bm_off) + start;
- /* local alloc is always contiguous by nature -- we never
- * delete bits from it! */
*num_bits = bits_wanted;
status = ocfs2_journal_access_di(handle,
ocfs2_set_bit(start++, bitmap);
le32_add_cpu(&alloc->id1.bitmap1.i_used, *num_bits);
+ ocfs2_journal_dirty(handle, osb->local_alloc_bh);
- status = ocfs2_journal_dirty(handle, osb->local_alloc_bh);
- if (status < 0) {
- mlog_errno(status);
- goto bail;
- }
-
- status = 0;
bail:
mlog_exit(status);
return status;
static int ocfs2_local_alloc_find_clear_bits(struct ocfs2_super *osb,
struct ocfs2_dinode *alloc,
- u32 numbits,
+ u32 *numbits,
struct ocfs2_alloc_reservation *resv)
{
int numfound, bitoff, left, startoff, lastzero;
+ int local_resv = 0;
+ struct ocfs2_alloc_reservation r;
void *bitmap = NULL;
struct ocfs2_reservation_map *resmap = &osb->osb_la_resmap;
- mlog_entry("(numbits wanted = %u)\n", numbits);
+ mlog_entry("(numbits wanted = %u)\n", *numbits);
if (!alloc->id1.bitmap1.i_total) {
mlog(0, "No bits in my window!\n");
goto bail;
}
- bitmap = OCFS2_LOCAL_ALLOC(alloc)->la_bitmap;
+ if (!resv) {
+ local_resv = 1;
+ ocfs2_resv_init_once(&r);
+ ocfs2_resv_set_type(&r, OCFS2_RESV_FLAG_TMP);
+ resv = &r;
+ }
+
+ numfound = *numbits;
+ if (ocfs2_resmap_resv_bits(resmap, resv, &bitoff, &numfound) == 0) {
+ if (numfound < *numbits)
+ *numbits = numfound;
+ goto bail;
+ }
/*
- * Ask the reservations code first whether this request can be
- * easily fulfilled. No errors here are fatal - if we didn't
- * find the number of bits needed, we'll just take the slow
- * path.
+ * Code error. While reservations are enabled, local
+ * allocation should _always_ go through them.
*/
- if (ocfs2_resmap_resv_bits(resmap, resv, bitmap, &bitoff, &numfound)
- == 0) {
- if (numfound >= numbits) {
- numfound = numbits;
- goto bail;
- }
- }
+ BUG_ON(osb->osb_resv_level != 0);
+
+ /*
+ * Reservations are disabled. Handle this the old way.
+ */
+
+ bitmap = OCFS2_LOCAL_ALLOC(alloc)->la_bitmap;
numfound = bitoff = startoff = 0;
lastzero = -1;
startoff = bitoff+1;
}
/* we got everything we needed */
- if (numfound == numbits) {
+ if (numfound == *numbits) {
/* mlog(0, "Found it all!\n"); */
break;
}
mlog(0, "Exiting loop, bitoff = %d, numfound = %d\n", bitoff,
numfound);
- if (numfound == numbits)
+ if (numfound == *numbits)
bitoff = startoff - numfound;
- else {
- numfound = 0;
+ else
bitoff = -1;
- }
bail:
+ if (local_resv)
+ ocfs2_resv_discard(resmap, resv);
+
mlog_exit(bitoff);
return bitoff;
}
/* we used the generic suballoc reserve function, but we set
* everything up nicely, so there's no reason why we can't use
* the more specific cluster api to claim bits. */
- status = ocfs2_claim_clusters(osb, handle, ac, osb->local_alloc_bits,
+ status = ocfs2_claim_clusters(handle, ac, osb->local_alloc_bits,
&cluster_off, &cluster_count);
if (status == -ENOSPC) {
retry_enospc:
goto bail;
ac->ac_bits_wanted = osb->local_alloc_default_bits;
- status = ocfs2_claim_clusters(osb, handle, ac,
+ status = ocfs2_claim_clusters(handle, ac,
osb->local_alloc_bits,
&cluster_off,
&cluster_count);
memset(OCFS2_LOCAL_ALLOC(alloc)->la_bitmap, 0,
le16_to_cpu(la->la_size));
- ocfs2_resmap_restart(&osb->osb_la_resmap, cluster_count);
+ ocfs2_resmap_restart(&osb->osb_la_resmap, cluster_count,
+ OCFS2_LOCAL_ALLOC(alloc)->la_bitmap);
mlog(0, "New window allocated:\n");
mlog(0, "window la_bm_off = %u\n",
}
ocfs2_clear_local_alloc(alloc);
-
- status = ocfs2_journal_dirty(handle, osb->local_alloc_bh);
- if (status < 0) {
- mlog_errno(status);
- goto bail;
- }
+ ocfs2_journal_dirty(handle, osb->local_alloc_bh);
status = ocfs2_sync_local_to_main(osb, handle, alloc_copy,
main_bm_inode, main_bm_bh);
atomic_inc(&osb->alloc_stats.moves);
- status = 0;
bail:
if (handle)
ocfs2_commit_trans(osb, handle);
projects / linux-flexiantxendom0-3.2.10.git / blobdiff