From: Rusty Russell <rusty@rustcorp.com.au>
Date: Sat, 25 Sep 2010 07:54:50 +0000 (+0930)
Subject: tdb: separate key and data size summary in tdb_summary().
X-Git-Url: https://git.ozlabs.org/?p=ccan;a=commitdiff_plain;h=43d881a232a4999e0d23db8e4335604dd5f3474a

tdb: separate key and data size summary in tdb_summary().
---

diff --git a/ccan/tdb/summary.c b/ccan/tdb/summary.c
index a9dc4f80..b8f7f976 100644
--- a/ccan/tdb/summary.c
+++ b/ccan/tdb/summary.c
@@ -19,9 +19,10 @@
 #include <ccan/tally/tally.h>
 
 #define SUMMARY_FORMAT \
-	"Size of file: %zu\n" \
+	"Size of file/data: %zu/%zu\n" \
 	"Number of records: %zu\n" \
-	"Smallest/average/largest records: %zu/%zu/%zu\n%s" \
+	"Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
+	"Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 	"Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 	"Number of dead records: %zu\n" \
 	"Smallest/average/largest dead records: %zu/%zu/%zu\n%s" \
@@ -29,7 +30,9 @@
 	"Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 	"Number of hash chains: %zu\n" \
 	"Smallest/average/largest hash chains: %zu/%zu/%zu\n%s" \
-	"Total data = %zu (%.0f%%)\n"
+	"Number of uncoalesced records: %zu\n" \
+	"Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
+	"Percentage keys/data/padding/free/dead/rechdrs&tailers/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 
 #define HISTO_WIDTH 70
 #define HISTO_HEIGHT 20
@@ -71,12 +74,12 @@ static size_t get_hash_length(struct tdb_context *tdb, unsigned int i)
 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 {
 	tdb_off_t off;
-	struct tally *freet, *used, *dead, *extra, *hash;
-	char *freeg, *usedg, *deadg, *extrag, *hashg;
+	struct tally *freet, *keys, *data, *dead, *extra, *hash, *uncoal;
+	char *freeg, *keysg, *datag, *deadg, *extrag, *hashg, *uncoalg;
 	struct tdb_record rec;
 	char *ret = NULL;
 	bool locked;
-	size_t len;
+	size_t len, unc = 0;
 
 	/* Read-only databases use no locking at all: it's best-effort.
 	 * We may have a write lock already, so skip that case too. */
@@ -89,11 +92,13 @@ char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 	}
 
 	freet = tally_new(HISTO_HEIGHT);
-	used = tally_new(HISTO_HEIGHT);
+	keys = tally_new(HISTO_HEIGHT);
+	data = tally_new(HISTO_HEIGHT);
 	dead = tally_new(HISTO_HEIGHT);
 	extra = tally_new(HISTO_HEIGHT);
 	hash = tally_new(HISTO_HEIGHT);
-	if (!freet || !used || !dead || !extra || !hash) {
+	uncoal = tally_new(HISTO_HEIGHT);
+	if (!freet || !keys || !data || !dead || !extra || !hash || !uncoal) {
 		tdb->ecode = TDB_ERR_OOM;
 		goto unlock;
 	}
@@ -106,20 +111,23 @@ char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 			goto unlock;
 		switch (rec.magic) {
 		case TDB_MAGIC:
-			tally_add(used, rec.key_len + rec.data_len);
+			tally_add(keys, rec.key_len);
+			tally_add(data, rec.data_len);
 			tally_add(extra, rec.rec_len - (rec.key_len
 							+ rec.data_len));
 			break;
 		case TDB_FREE_MAGIC:
-			tally_add(freet, sizeof(rec) + rec.rec_len);
+			tally_add(freet, rec.rec_len);
+			unc++;
 			break;
 		/* If we crash after ftruncate, we can get zeroes or fill. */
 		case TDB_RECOVERY_INVALID_MAGIC:
 		case 0x42424242:
+			unc++;
 			rec.rec_len = dead_space(tdb, off) - sizeof(rec);
 			/* Fall through */
 		case TDB_DEAD_MAGIC:
-			tally_add(dead, sizeof(rec) + rec.rec_len);
+			tally_add(dead, rec.rec_len);
 			break;
 		default:
 			TDB_LOG((tdb, TDB_DEBUG_ERROR,
@@ -127,37 +135,51 @@ char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 				 rec.magic, off));
 			goto unlock;
 		}
+
+		if (unc &&
+		    (rec.magic == TDB_MAGIC || rec.magic == TDB_DEAD_MAGIC)) {
+			tally_add(uncoal, unc);
+			unc = 0;
+		}
 	}
+	if (unc)
+		tally_add(uncoal, unc);
 
 	for (off = 0; off < tdb->header.hash_size; off++)
 		tally_add(hash, get_hash_length(tdb, off));
 
 	if (flags & TDB_SUMMARY_HISTOGRAMS) {
 		freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
-		usedg = tally_histogram(used, HISTO_WIDTH, HISTO_HEIGHT);
+		keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
+		datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 		deadg = tally_histogram(dead, HISTO_WIDTH, HISTO_HEIGHT);
 		extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 		hashg = tally_histogram(hash, HISTO_WIDTH, HISTO_HEIGHT);
+		uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 	} else {
-		freeg = usedg = deadg = extrag = hashg = NULL;
+		freeg = keysg = datag = deadg = extrag = hashg = NULL;
 	}
 
 	/* 20 is max length of a %zu. */
-	len = strlen(SUMMARY_FORMAT) + 22*20 + 1
+	len = strlen(SUMMARY_FORMAT) + 29*20 + 1
 		+ (freeg ? strlen(freeg) : 0)
-		+ (usedg ? strlen(usedg) : 0)
+		+ (keysg ? strlen(keysg) : 0)
+		+ (datag ? strlen(datag) : 0)
 		+ (deadg ? strlen(deadg) : 0)
 		+ (extrag ? strlen(extrag) : 0)
-		+ (hashg ? strlen(hashg) : 0);
+		+ (hashg ? strlen(hashg) : 0)
+		+ (uncoalg ? strlen(uncoalg) : 0);
 	ret = malloc(len);
 	if (!ret)
 		goto unlock;
 
 	sprintf(ret, SUMMARY_FORMAT,
-		tdb->map_size,
-		tally_num(used),
-		tally_min(used), tally_mean(used), tally_max(used),
-		usedg ? usedg : "",
+		tdb->map_size, tally_total(keys, NULL)+tally_total(data, NULL),
+		tally_num(keys),
+		tally_min(keys), tally_mean(keys), tally_max(keys),
+		keysg ? keysg : "",
+		tally_min(data), tally_mean(data), tally_max(data),
+		datag ? datag : "",
 		tally_min(extra), tally_mean(extra), tally_max(extra),
 		extrag ? extrag : "",
 		tally_num(dead),
@@ -169,20 +191,35 @@ char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 		tally_num(hash),
 		tally_min(hash), tally_mean(hash), tally_max(hash),
 		hashg ? hashg : "",
-		tally_total(used, NULL),
-		tally_total(used, NULL) * 100.0 / tdb->map_size);
+		tally_total(uncoal, NULL),
+		tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
+		uncoalg ? uncoalg : "",
+		tally_total(keys, NULL) * 100.0 / tdb->map_size,
+		tally_total(data, NULL) * 100.0 / tdb->map_size,
+		tally_total(extra, NULL) * 100.0 / tdb->map_size,
+		tally_total(freet, NULL) * 100.0 / tdb->map_size,
+		tally_total(dead, NULL) * 100.0 / tdb->map_size,
+		(tally_num(keys) + tally_num(freet) + tally_num(dead))
+		* (sizeof(struct tdb_record) + sizeof(uint32_t))
+		* 100.0 / tdb->map_size,
+		tdb->header.hash_size * sizeof(tdb_off_t)
+		* 100.0 / tdb->map_size);
 
 unlock:
 	free(freeg);
-	free(usedg);
+	free(keysg);
+	free(datag);
 	free(deadg);
 	free(extrag);
 	free(hashg);
+	free(uncoalg);
 	free(freet);
-	free(used);
+	free(keys);
+	free(data);
 	free(dead);
 	free(extra);
 	free(hash);
+	free(uncoal);
 	if (locked) {
 		tdb_unlockall_read(tdb);
 	}