X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;ds=sidebyside;f=lib%2Fsorter%2Fcommon.h;h=b4b24ab865af01890d71552d368f702a47750389;hb=240457dec3ea3ef6d758e86e6bbf68c530f3c056;hp=99bc6a5f7e68497bfac047bdfddc010ecbcab6ef;hpb=6652d8d95e5f89eeb4c16163e708583e2f8f9799;p=libucw.git

diff --git a/lib/sorter/common.h b/lib/sorter/common.h
index 99bc6a5f..b4b24ab8 100644
--- a/lib/sorter/common.h
+++ b/lib/sorter/common.h
@@ -14,16 +14,19 @@
 
 /* Configuration, some of the variables are used by the old sorter, too. */
 extern uns sorter_trace, sorter_presort_bufsize, sorter_stream_bufsize;
-extern uns sorter_debug;
+extern uns sorter_debug, sorter_min_radix_bits, sorter_max_radix_bits;
 extern u64 sorter_bufsize;
+extern struct fb_params sorter_fb_params;
 
-#define SORT_TRACE(x...) do { if (sorter_trace) log(L_DEBUG, x); } while(0)
-#define SORT_XTRACE(x...) do { if (sorter_trace > 1) log(L_DEBUG, x); } while(0)
+#define SORT_TRACE(x...) do { if (sorter_trace) msg(L_DEBUG, x); } while(0)
+#define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) msg(L_DEBUG, x); } while(0)
 
 enum sort_debug {
   SORT_DEBUG_NO_PRESORT = 1,
   SORT_DEBUG_NO_JOIN = 2,
   SORT_DEBUG_KEEP_BUCKETS = 4,
+  SORT_DEBUG_NO_RADIX = 8,
+  SORT_DEBUG_NO_MULTIWAY = 16,
 };
 
 struct sort_bucket;
@@ -32,23 +35,40 @@ struct sort_context {
   struct fastbuf *in_fb;
   struct fastbuf *out_fb;
   uns hash_bits;
+  u64 in_size;
 
   struct mempool *pool;
   clist bucket_list;
-  void *big_buf, *big_buf_half;
-  size_t big_buf_size, big_buf_half_size;
+  void *big_buf;
+  size_t big_buf_size;
 
   int (*custom_presort)(struct fastbuf *dest, void *buf, size_t bufsize);
+
   // Take as much as possible from the source bucket, sort it in memory and dump to destination bucket.
   // Return 1 if there is more data available in the source bucket.
   int (*internal_sort)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only);
+
+  // Estimate how much input data from `b' will fit in the internal sorting buffer.
+  u64 (*internal_estimate)(struct sort_context *ctx, struct sort_bucket *b);
+
   // Two-way split/merge: merge up to 2 source buckets to up to 2 destination buckets.
   // Bucket arrays are NULL-terminated.
   void (*twoway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket **outs);
 
+  // Multi-way merge: merge an arbitrary number of source buckets to a single destination bucket.
+  void (*multiway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket *out);
+
+  // Radix split according to hash function
+  void (*radix_split)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket **outs, uns bitpos, uns numbits);
+
   // State variables of internal_sort
   void *key_buf;
   int more_keys;
+
+  // Timing
+  timestamp_t start_time;
+  uns last_pass_time;
+  uns total_int_time, total_pre_time, total_ext_time;
 };
 
 void sorter_run(struct sort_context *ctx);
@@ -56,6 +76,7 @@ void sorter_run(struct sort_context *ctx);
 /* Buffers */
 
 void *sorter_alloc(struct sort_context *ctx, uns size);
+void sorter_prepare_buf(struct sort_context *ctx);
 void sorter_alloc_buf(struct sort_context *ctx);
 void sorter_free_buf(struct sort_context *ctx);