fixed bug that caused number of inodes to make estimation to be terrible

in most cases

diff --git a/commands/simple/mkfs.c b/commands/simple/mkfs.c
index 25278294e35b4d89e89b34bb58a4e6ca4b5a1b4d..c5d0b4c8009e12708d33049aac19aac6920ad261 100755 (executable)
--- a/commands/simple/mkfs.c
+++ b/commands/simple/mkfs.c
@@ -39,8 +39,8 @@
 #include "../../servers/fs/super.h"
 #include <minix/fslib.h>
 
-#ifndef min
-#define min(a,b) ((a) < (b) ? (a) : (b))
+#ifndef max
+#define max(a,b) ((a) > (b) ? (a) : (b))
 #endif
 
 #ifndef DOS
@@ -267,7 +267,7 @@ char *argv[];
        }
        if (i == 0) {
                int kb;
-               kb = blocks * (min(block_size,1024) / 1024);
+               kb = blocks * (max(block_size,1024) / 1024);
                /* The default for inodes is 2 blocks per kb, rounded up
                 * to fill an inode block.  Above 20M, the average files are
                 * sure to be larger because it is hard to fill up 20M with
@@ -280,9 +280,12 @@ char *argv[];
                if (kb >= 60000) i = kb / 5;
                if (kb >= 80000) i = kb / 6;
                if (kb >= 100000) i = kb / 7;
+
+               /* round up to fill inode block */
                i += inodes_per_block - 1;
                i = i / inodes_per_block * inodes_per_block;
                if (i > INODE_MAX) i = INODE_MAX;
+
        }
        if (blocks < 5) pexit("Block count too small");
        if (blocks > max_nrblocks) pexit("Block count too large");