dedup

commit b32c5c5b3c6b92b0a10ddc694e7df2d233c43c52
parent 0b721adea7ed4e7d16054d439bb092a2905bade8
Author: sin <sin@2f30.org>
Date:   Sun, 17 Feb 2019 09:54:19 +0000

Print some stats about blk size in verbose output

Tweak sizes again.

Diffstat:
M
dedup.c
 | 
27
++++++++++++++++++++-------

1 file changed, 20 insertions(+), 7 deletions(-)
diff --git a/dedup.c b/dedup.c
@@ -16,8 +16,8 @@
 #define STOREF ".store"
 #define CACHEF ".cache"
 
-#define BLKSIZ (8*1024*1024)
-#define WINSIZ 4096
+#define BLKSIZ (1024 * 512)
+#define WINSIZ 1024
 #define HASHMSK ((1ul << 21) - 1)
 #define MDSIZ SHA256_DIGEST_LENGTH
 
@@ -556,6 +556,11 @@ void
 init_cache(void)
 {
 	uint64_t nents, i;
+	uint64_t min, max, avg;
+
+	min = BLKSIZ;
+	max = 0;
+	avg = 0;
 
 	nents = cache_nents();
 	lseek(cfd, 0, SEEK_SET);
@@ -567,11 +572,19 @@ init_cache(void)
 			errx(1, "read: unexpected EOF");
 		add_cent(cent);
 
-		if (verbose) {
-			fprintf(stderr, "bdescr.offset: %llu bdescr.size: %llu\n",
-			       (unsigned long long)cent->bdescr.offset,
-			       (unsigned long long)cent->bdescr.size);
-		}
+		if (cent->bdescr.size > max)
+			max = cent->bdescr.size;
+		if (cent->bdescr.size < min)
+			min = cent->bdescr.size;
+		avg += cent->bdescr.size;
+	}
+	avg /= nents;
+
+	if (verbose) {
+		fprintf(stderr, "min/avg/max blksize: %llu/%llu/%llu\n",
+		        (unsigned long long)min,
+		        (unsigned long long)avg,
+		        (unsigned long long)max);
 	}
 }