| 
  | 
Message-ID: <20210113143809.608ea269@ncopa-desktop.lan>
Date: Wed, 13 Jan 2021 14:38:09 +0100
From: Natanael Copa <ncopa@...inelinux.org>
To: musl@...ts.openwall.com
Subject: Re: [PATCH] make realpath replace leading // with a single /
On Wed, 13 Jan 2021 14:28:35 +0100
Natanael Copa <ncopa@...inelinux.org> wrote:
> On some systems a leading double slash may have special meaning, so
> POSIX[1] says that "If a pathname begins with two successive <slash>
> characters, the first component following the leading <slash> characters
> may be interpreted in an implementation-defined manner"
> 
> While current musl implementation is technically correct, most other
> systems' (at least GNU libc, freebsd, openbsd, netbsd macOS)
> implementations will replace a leading // with a single /. Make musl
> do the same to avoid surprises.
> 
> [1]: https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_13
> ---
>  src/misc/realpath.c | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/src/misc/realpath.c b/src/misc/realpath.c
> index db8b74dc..414b4741 100644
> --- a/src/misc/realpath.c
> +++ b/src/misc/realpath.c
> @@ -46,9 +46,6 @@ restart:
>  			q=0;
>  			output[q++] = '/';
>  			p++;
> -			/* Initial // is special. */
> -			if (stack[p] == '/' && stack[p+1] != '/')
> -				output[q++] = '/';
>  			continue;
>  		}
>  
This fixes gettext's (gnulib) testsuite, which tests if realpath("//",
NULL) return "/" and fails if it doesn't.
I ran this testcase on multiple systems:
#include <stdio.h>
#include <stdlib.h>
int main() {
	printf("%s\n", realpath("//", NULL));
	return 0;
}
musl 1.1.24:	/
ubuntu 20.03:	/
macOS Big Sur:	/
OpenBSD 6.8:	/
FreeBSD 12.2:	/
NetBSD 9.1:	/
musl (current)	//
I don't know why this behavior was introduced in musl, but I think
this only adds meaningless friction to downstream users.
-nc
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.