summaryrefslogtreecommitdiff
path: root/sys/man/2
diff options
context:
space:
mode:
authorcinap_lenrek <cinap_lenrek@felloff.net>2017-12-31 09:06:42 +0100
committercinap_lenrek <cinap_lenrek@felloff.net>2017-12-31 09:06:42 +0100
commitc039b52fc3d7c225fe8b6f88f21df83ac26d5984 (patch)
treebed796d3111679d359851521644fbbee2378f200 /sys/man/2
parentee89c82dd35475b93dd67773b7d10e461921a734 (diff)
libc: constant time implementation for encode(2) routines, fix base32
the string encoding functions touch secret key material in a bunch of places (devtls, devcap), so make sure we do not leak information by cache timing side channels, making the encoding and decoding routines constant time. we also expose the alphabets through encXchr()/decXchr() functions so caller can find the end of a encoded string before calling decode function (for libmp). the base32 encoding was broken in several ways. inputs lengths of len%5 == [2,3,4] had output truncated and it was using non-standard alphabet. documenting the alphabet change in the manpage.
Diffstat (limited to 'sys/man/2')
-rw-r--r--sys/man/2/encode52
1 files changed, 50 insertions, 2 deletions
diff --git a/sys/man/2/encode b/sys/man/2/encode
index 3a8ca61e4..db7944945 100644
--- a/sys/man/2/encode
+++ b/sys/man/2/encode
@@ -1,6 +1,8 @@
.TH ENCODE 2
.SH NAME
-dec64, enc64, dec32, enc32, dec16, enc16, encodefmt \- encoding byte arrays as strings
+dec64, enc64, dec32, enc32, dec16, enc16, \
+dec64chr, enc64chr, dec32chr, enc32chr, dec16chr, enc16chr, \
+encodefmt \- encoding byte arrays as strings
.SH SYNOPSIS
.B #include <u.h>
.br
@@ -25,8 +27,28 @@ int dec16(uchar *out, int lim, char *in, int n)
int enc16(char *out, int lim, uchar *in, int n)
.PP
.B
+int dec64chr(int c)
+.PP
+.B
+int enc64chr(int o)
+.PP
+.B
+int dec32chr(int c)
+.PP
+.B
+int enc32chr(int o)
+.PP
+.B
+int dec16chr(int c)
+.PP
+.B
+int enc16chr(int o)
+.PP
+.B
int encodefmt(Fmt*)
.SH DESCRIPTION
+The functions described here handle encoding and decoding of
+bytes to printable ASCII strings as specified by RFC4648.
.PP
.IR Enc16 ,
.I enc32
@@ -47,6 +69,20 @@ The decoding fails if the output buffer is not large enough or,
for base 32, if the input buffer length is not a multiple
of 8.
.PP
+.IR Dec16chr ,
+.I dec32chr
+and
+.I dec64chr
+return the value for a symbol of the alphabet or -1 when the
+symbol is not in the alphabet.
+.PP
+.IR Enc16chr ,
+.I enc32chr
+and
+.I enc64chr
+encode a symbol of the alphabet given a value.
+if the value is out of range then zero is returned.
+.PP
.I Encodefmt
can be used with
.IR fmtinstall (2)
@@ -62,7 +98,8 @@ in upper case. The
flag forces lower case.
.TP
.B <
-base 32
+base 32. The default is upper case, same as
+.BR H .
.TP
.B [
base 64 (same as MIME)
@@ -80,8 +117,19 @@ For example, to display a 15 byte array as hex:
.EE
.SH SOURCE
+.B /sys/src/libc/port/u16.c
+.br
.B /sys/src/libc/port/u32.c
.br
.B /sys/src/libc/port/u64.c
.br
.B /sys/src/libc/port/encodefmt.c
+.SH HISTORY
+In Jan 2018, base 32 encoding was changed from non-standard
+to standard RFC4648 alphabet.
+.TP
+old:
+.B "23456789abcdefghijkmnpqrstuvwxyz"
+.TP
+new:
+.B "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567"