#include "lib/mempool.h"
#include "lib/sorter/common.h"
+#include <fcntl.h>
+
void *
sorter_alloc(struct sort_context *ctx, uns size)
{
struct sort_bucket *
sbuck_new(struct sort_context *ctx)
{
- return sorter_alloc(ctx, sizeof(struct sort_bucket));
+ struct sort_bucket *b = sorter_alloc(ctx, sizeof(struct sort_bucket));
+ b->ctx = ctx;
+ return b;
}
void
{
if (b)
{
+ ASSERT(!(b->flags & SBF_DESTROYED));
if (b->n.prev)
clist_remove(&b->n);
bclose(b->fb);
bzero(b, sizeof(*b));
+ b->flags = SBF_DESTROYED;
}
}
+sh_off_t
+sbuck_size(struct sort_bucket *b)
+{
+ if ((b->flags & SBF_OPEN_WRITE) && !(b->flags & SBF_SWAPPED_OUT))
+ return btell(b->fb);
+ else
+ return b->size;
+}
+
int
-sbuck_can_read(struct sort_bucket *b)
+sbuck_have(struct sort_bucket *b)
{
- return b && b->size;
+ return b && sbuck_size(b);
}
-struct fastbuf *
-sbuck_open_read(struct sort_bucket *b)
+static int
+sbuck_has_file(struct sort_bucket *b)
{
- /* FIXME: These functions should handle buckets with no fb and only name. */
- ASSERT(b->fb);
- return b->fb;
+ return (b->fb || (b->flags & SBF_SWAPPED_OUT));
+}
+
+static void
+sbuck_swap_in(struct sort_bucket *b)
+{
+ if (b->flags & SBF_SWAPPED_OUT)
+ {
+ b->fb = bopen(b->filename, O_RDWR, sorter_stream_bufsize);
+ if (b->flags & SBF_OPEN_WRITE)
+ bseek(b->fb, 0, SEEK_END);
+ bconfig(b->fb, BCONFIG_IS_TEMP_FILE, 1);
+ b->flags &= ~SBF_SWAPPED_OUT;
+ SORT_XTRACE("Swapped in %s", b->filename);
+ }
}
struct fastbuf *
-sbuck_open_write(struct sort_bucket *b)
+sbuck_read(struct sort_bucket *b)
{
- if (!b->fb)
- b->fb = bopen_tmp(sorter_stream_bufsize);
- return b->fb;
+ sbuck_swap_in(b);
+ if (b->flags & SBF_OPEN_READ)
+ return b->fb;
+ else if (b->flags & SBF_OPEN_WRITE)
+ {
+ b->size = btell(b->fb);
+ b->flags = (b->flags & ~SBF_OPEN_WRITE) | SBF_OPEN_READ;
+ brewind(b->fb);
+ return b->fb;
+ }
+ else
+ ASSERT(0);
}
-void
-sbuck_close_read(struct sort_bucket *b)
+struct fastbuf *
+sbuck_write(struct sort_bucket *b)
{
- if (!b)
- return;
- ASSERT(b->fb);
- bclose(b->fb);
- b->fb = NULL;
+ sbuck_swap_in(b);
+ if (b->flags & SBF_OPEN_WRITE)
+ ASSERT(b->fb);
+ else
+ {
+ ASSERT(!(b->flags & (SBF_OPEN_READ | SBF_DESTROYED)));
+ b->fb = bopen_tmp(sorter_stream_bufsize);
+ b->flags |= SBF_OPEN_WRITE;
+ b->filename = mp_strdup(b->ctx->pool, b->fb->name);
+ }
+ return b->fb;
}
void
-sbuck_close_write(struct sort_bucket *b)
+sbuck_swap_out(struct sort_bucket *b)
{
- if (b->fb)
+ if ((b->flags & (SBF_OPEN_READ | SBF_OPEN_WRITE)) && b->fb)
{
- b->size = btell(b->fb);
- brewind(b->fb);
+ if (b->flags & SBF_OPEN_WRITE)
+ b->size = btell(b->fb);
+ bconfig(b->fb, BCONFIG_IS_TEMP_FILE, 0);
+ bclose(b->fb);
+ b->fb = NULL;
+ b->flags |= SBF_SWAPPED_OUT;
+ SORT_XTRACE("Swapped out %s", b->filename);
}
}
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);
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;
}
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;
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
// 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
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);
}
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);
}
projects / libucw.git / blobdiff