X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fgovern.c;h=738186e34da8a6eebddb0384f6aa3781b4387fcb;hb=bbcbba14a32658a09d7cd44cee40c8369065822f;hp=97942dee37bbc0d7604d3b2a6ed62f0e7fb631ca;hpb=baa9f9a3368c8d318b9711340727f822d8fc8a34;p=libucw.git

diff --git a/lib/sorter/govern.c b/lib/sorter/govern.c
index 97942dee..738186e3 100644
--- a/lib/sorter/govern.c
+++ b/lib/sorter/govern.c
@@ -12,103 +12,14 @@
 #include "lib/mempool.h"
 #include "lib/sorter/common.h"
 
-void *
-sorter_alloc(struct sort_context *ctx, uns size)
-{
-  return mp_alloc_zero(ctx->pool, size);
-}
-
-struct sort_bucket *
-sbuck_new(struct sort_context *ctx)
-{
-  return sorter_alloc(ctx, sizeof(struct sort_bucket));
-}
-
-void
-sbuck_drop(struct sort_bucket *b)
-{
-  if (b)
-    {
-      if (b->n.prev)
-	clist_remove(&b->n);
-      bclose(b->fb);
-      bzero(b, sizeof(*b));
-    }
-}
-
-int
-sbuck_can_read(struct sort_bucket *b)
-{
-  return b && b->size;
-}
-
-struct fastbuf *
-sbuck_open_read(struct sort_bucket *b)
-{
-  /* FIXME: These functions should handle buckets with no fb and only name. */
-  ASSERT(b->fb);
-  return b->fb;
-}
-
-struct fastbuf *
-sbuck_open_write(struct sort_bucket *b)
-{
-  if (!b->fb)
-    b->fb = bopen_tmp(sorter_stream_bufsize);
-  return b->fb;
-}
-
-void
-sbuck_close_read(struct sort_bucket *b)
-{
-  if (!b)
-    return;
-  ASSERT(b->fb);
-  bclose(b->fb);
-  b->fb = NULL;
-}
-
-void
-sbuck_close_write(struct sort_bucket *b)
-{
-  if (b->fb)
-    {
-      b->size = btell(b->fb);
-      brewind(b->fb);
-    }
-}
-
-void
-sorter_alloc_buf(struct sort_context *ctx)
-{
-  if (ctx->big_buf)
-    return;
-  u64 bs = MAX(sorter_bufsize/2, 1);
-  bs = ALIGN_TO(bs, (u64)CPU_PAGE_SIZE);
-  ctx->big_buf = big_alloc(2*bs);
-  ctx->big_buf_size = 2*bs;
-  ctx->big_buf_half = ((byte*) ctx->big_buf) + bs;
-  ctx->big_buf_half_size = bs;
-  SORT_XTRACE("Allocated sorting buffer (%jd bytes)", (uintmax_t) bs);
-}
-
-void
-sorter_free_buf(struct sort_context *ctx)
-{
-  if (!ctx->big_buf)
-    return;
-  big_free(ctx->big_buf, ctx->big_buf_size);
-  ctx->big_buf = NULL;
-  SORT_XTRACE("Freed sorting buffer");
-}
+#include <string.h>
 
 static int sorter_presort(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only)
 {
-  /* FIXME: Mode with no presorting (mostly for debugging) */
   sorter_alloc_buf(ctx);
   if (in->flags & SBF_CUSTOM_PRESORT)
     {
-      struct fastbuf *f = sbuck_open_write(out);
+      struct fastbuf *f = sbuck_write(out);
       return ctx->custom_presort(f, ctx->big_buf, ctx->big_buf_size);	// FIXME: out_only optimization?
     }
   return ctx->internal_sort(ctx, in, out, out_only);
@@ -117,6 +28,9 @@ static int sorter_presort(struct sort_context *ctx, struct sort_bucket *in, stru
 static inline struct sort_bucket *
 sbuck_join_to(struct sort_bucket *b)
 {
+  if (sorter_debug & SORT_DEBUG_NO_JOIN)
+    return NULL;
+
   struct sort_bucket *out = (struct sort_bucket *) b->n.prev;	// Such bucket is guaranteed to exist
   return (out->flags & SBF_FINAL) ? out : NULL;
 }
@@ -124,49 +38,65 @@ sbuck_join_to(struct sort_bucket *b)
 static void
 sorter_join(struct sort_bucket *b)
 {
-  struct sort_bucket *join = sbuck_join_to(b);
-  ASSERT(join);
-
-  // FIXME: What if the final bucket doesn't contain any file yet?
+  struct sort_bucket *join = (struct sort_bucket *) b->n.prev;
+  ASSERT(join->flags & SBF_FINAL);
+  ASSERT(b->runs == 1);
 
-  SORT_TRACE("Copying %jd bytes to output file", (uintmax_t) b->size);
-  struct fastbuf *src = sbuck_open_read(b);
-  struct fastbuf *dest = sbuck_open_write(join);
-  bbcopy(src, dest, ~0U);
-  sbuck_drop(b);
+  if (!sbuck_has_file(join))
+    {
+      // The final bucket doesn't have any file associated yet, so replace
+      // it with the new bucket.
+      SORT_XTRACE("Replaced final bucket");
+      b->flags |= SBF_FINAL;
+      sbuck_drop(join);
+    }
+  else
+    {
+      SORT_TRACE("Copying %jd bytes to output file", (uintmax_t) sbuck_size(b));
+      struct fastbuf *src = sbuck_read(b);
+      struct fastbuf *dest = sbuck_write(join);
+      bbcopy(src, dest, ~0U);
+      sbuck_drop(b);
+    }
 }
 
 static void
 sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
 {
-  struct sort_bucket *ins[3], *outs[3];
+  struct sort_bucket *ins[3] = { NULL }, *outs[3] = { NULL };
+  cnode *list_pos = b->n.prev;
   struct sort_bucket *join = sbuck_join_to(b);
 
-  SORT_TRACE("Presorting");
-  ins[0] = sbuck_new(ctx);
-  sbuck_open_read(b);
-  if (!sorter_presort(ctx, b, ins[0], join ? : ins[0]))
+  if (!(sorter_debug & SORT_DEBUG_NO_PRESORT) || (b->flags & SBF_CUSTOM_PRESORT))
     {
-      if (join)
-	sbuck_drop(ins[0]);
-      else
-	clist_insert_after(&ins[0]->n, &b->n);
+      SORT_TRACE("Presorting");
+      ins[0] = sbuck_new(ctx);
+      if (!sorter_presort(ctx, b, ins[0], join ? : ins[0]))
+	{
+	  SORT_XTRACE("Sorted in memory");
+	  if (join)
+	    sbuck_drop(ins[0]);
+	  else
+	    clist_insert_after(&ins[0]->n, list_pos);
+	  sbuck_drop(b);
+	  return;
+	}
+
+      ins[1] = sbuck_new(ctx);
+      int i = 1;
+      while (sorter_presort(ctx, b, ins[i], ins[i]))
+	i = 1-i;
       sbuck_drop(b);
-      return;
     }
-
-  ins[1] = sbuck_new(ctx);
-  ins[2] = NULL;
-  int i = 1;
-  while (sorter_presort(ctx, b, ins[i], ins[i]))
-    i = 1-i;
-  sbuck_close_read(b);
-  sbuck_close_write(ins[0]);
-  sbuck_close_write(ins[1]);
+  else
+    {
+      SORT_TRACE("Skipped presorting");
+      ins[0] = b;
+    }
 
   SORT_TRACE("Main sorting");
   do {
-    if (ins[0]->runs == 1 && ins[1]->runs == 1 && join)	// FIXME: Debug switch for disabling joining optimizations
+    if (ins[0]->runs == 1 && ins[1]->runs == 1 && join)
       {
 	// This is guaranteed to produce a single run, so join if possible
 	outs[0] = join;
@@ -175,24 +105,22 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
 	ASSERT(outs[0]->runs == 2);
 	outs[0]->runs--;
 	SORT_TRACE("Pass done (joined final run)");
-	sbuck_drop(b);
+	sbuck_drop(ins[0]);
+	sbuck_drop(ins[1]);
 	return;
       }
     outs[0] = sbuck_new(ctx);
     outs[1] = sbuck_new(ctx);
     outs[2] = NULL;
     ctx->twoway_merge(ctx, ins, outs);
-    sbuck_close_write(outs[0]);
-    sbuck_close_write(outs[1]);
-    SORT_TRACE("Pass done (%d+%d runs, %jd+%jd bytes)", outs[0]->runs, outs[1]->runs, (uintmax_t) outs[0]->size, (uintmax_t) outs[1]->size);
+    SORT_TRACE("Pass done (%d+%d runs, %jd+%jd bytes)", outs[0]->runs, outs[1]->runs, (uintmax_t) sbuck_size(outs[0]), (uintmax_t) sbuck_size(outs[1]));
     sbuck_drop(ins[0]);
     sbuck_drop(ins[1]);
     memcpy(ins, outs, 3*sizeof(struct sort_bucket *));
-  } while (ins[1]->size);
+  } while (sbuck_have(ins[1]));
 
   sbuck_drop(ins[1]);
-  clist_insert_after(&ins[0]->n, &b->n);
-  sbuck_drop(b);
+  clist_insert_after(&ins[0]->n, list_pos);
 }
 
 void
@@ -206,7 +134,7 @@ sorter_run(struct sort_context *ctx)
 
   // Create bucket containing the source
   struct sort_bucket *bin = sbuck_new(ctx);
-  bin->flags = SBF_SOURCE;
+  bin->flags = SBF_SOURCE | SBF_OPEN_READ;
   if (ctx->custom_presort)
     bin->flags |= SBF_CUSTOM_PRESORT;
   else
@@ -225,9 +153,9 @@ sorter_run(struct sort_context *ctx)
   clist_add_head(&ctx->bucket_list, &bout->n);
 
   struct sort_bucket *b;
-  while (b = clist_next(&ctx->bucket_list, &bout->n))
+  while (bout = clist_head(&ctx->bucket_list), b = clist_next(&ctx->bucket_list, &bout->n))
     {
-      if (!b->size)
+      if (!sbuck_have(b))
 	sbuck_drop(b);
       else if (b->runs == 1)
 	sorter_join(b);
@@ -236,7 +164,7 @@ sorter_run(struct sort_context *ctx)
     }
 
   sorter_free_buf(ctx);
-  sbuck_close_write(bout);
-  SORT_XTRACE("Final size: %jd", (uintmax_t) bout->size);
-  ctx->out_fb = sbuck_open_read(bout);
+  sbuck_write(bout);		// Force empty bucket to a file
+  SORT_XTRACE("Final size: %jd", (uintmax_t) sbuck_size(bout));
+  ctx->out_fb = sbuck_read(bout);
 }