Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux

hexdump: fix ascii column for the tail of a dump

In the current implementation we have a floating ascii column in the tail
of the dump.

For example, for row size equal to 16 the ascii column as in following
table

group size \ length 8 12 16
1 50 50 50
2 22 32 42
4 20 29 38
8 19 - 36

This patch makes it the same independently of amount of bytes dumped.

The change is safe since all current users, which use ASCII part of the
dump, rely on the group size equal to 1. The patch doesn't change
behaviour for such group size (see the table above).

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

authored by

Andy Shevchenko and committed by
Linus Torvalds
6f6f3fcb 64d1d77a

+3 -3
+3 -3
lib/hexdump.c
··· 126 126 lx += scnprintf(linebuf + lx, linebuflen - lx, 127 127 "%s%16.16llx", j ? " " : "", 128 128 (unsigned long long)*(ptr8 + j)); 129 - ascii_column = 17 * ngroups + 2; 129 + ascii_column = rowsize * 2 + rowsize / 8 + 2; 130 130 break; 131 131 } 132 132 ··· 137 137 for (j = 0; j < ngroups; j++) 138 138 lx += scnprintf(linebuf + lx, linebuflen - lx, 139 139 "%s%8.8x", j ? " " : "", *(ptr4 + j)); 140 - ascii_column = 9 * ngroups + 2; 140 + ascii_column = rowsize * 2 + rowsize / 4 + 2; 141 141 break; 142 142 } 143 143 ··· 148 148 for (j = 0; j < ngroups; j++) 149 149 lx += scnprintf(linebuf + lx, linebuflen - lx, 150 150 "%s%4.4x", j ? " " : "", *(ptr2 + j)); 151 - ascii_column = 5 * ngroups + 2; 151 + ascii_column = rowsize * 2 + rowsize / 2 + 2; 152 152 break; 153 153 } 154 154