mailing list of musl libc
 help / color / mirror / code / Atom feed
From: Rich Felker <dalias@libc.org>
To: Kaihang Zhang <kaihang.zhang@smartx.com>
Cc: musl@lists.openwall.com, 2010267516@qq.com
Subject: Re: [musl] [PATCH v2] fix: Truncate the too-long mntent in function getmntent_r
Date: Sat, 8 Jan 2022 22:12:51 -0500	[thread overview]
Message-ID: <20220109031250.GN7074@brightrain.aerifal.cx> (raw)
In-Reply-To: <20211015122000.2490-1-kaihang.zhang@smartx.com>

On Fri, Oct 15, 2021 at 08:20:00AM -0400, Kaihang Zhang wrote:
> In function getmntent_r in source misc/mntent.c, entry that is too long
> will be truncated rather than discarded. The caller can tell by errno
> whether the supplied buffer is too small, and retry from the beginning
> of the file.
> ---
>  src/misc/mntent.c | 53 +++++++++++++++++++++++++++++------------------
>  1 file changed, 33 insertions(+), 20 deletions(-)
> 
> diff --git a/src/misc/mntent.c b/src/misc/mntent.c
> index eabb8200..085ce45d 100644
> --- a/src/misc/mntent.c
> +++ b/src/misc/mntent.c
> @@ -21,12 +21,12 @@ int endmntent(FILE *f)
>  
>  struct mntent *getmntent_r(FILE *f, struct mntent *mnt, char *linebuf, int buflen)
>  {
> -	int cnt, n[8], use_internal = (linebuf == SENTINEL);
> -
> -	mnt->mnt_freq = 0;
> -	mnt->mnt_passno = 0;
> +	int use_internal = (linebuf == SENTINEL);
> +	char *sub;
>  
>  	do {
> +		char *end_ptr;
> +
>  		if (use_internal) {
>  			getline(&internal_buf, &internal_bufsize, f);
>  			linebuf = internal_buf;
> @@ -34,25 +34,38 @@ struct mntent *getmntent_r(FILE *f, struct mntent *mnt, char *linebuf, int bufle
>  			fgets(linebuf, buflen, f);
>  		}
>  		if (feof(f) || ferror(f)) return 0;
> -		if (!strchr(linebuf, '\n')) {
> +
> +		end_ptr = strchr(linebuf, '\n');
> +		if (end_ptr != NULL) {
> +			while ((end_ptr[-1] == ' ' || end_ptr[-1] == '\t') && end_ptr != linebuf) end_ptr--;
> +			*end_ptr = '\0';

Unless I'm misreading, this seems to invoke UB by reading the [-1]
index before checking if that's valid. It could be fixed by just
changing the order of the comparison expressions, but I'm not clear
why this needs to be done anyway.


> +		} else {
>  			fscanf(f, "%*[^\n]%*[\n]");
>  			errno = ERANGE;
> -			return 0;
>  		}
> -		cnt = sscanf(linebuf, " %n%*s%n %n%*s%n %n%*s%n %n%*s%n %d %d",
> -			n, n+1, n+2, n+3, n+4, n+5, n+6, n+7,
> -			&mnt->mnt_freq, &mnt->mnt_passno);
> -	} while (cnt < 2 || linebuf[n[0]] == '#');
> -
> -	linebuf[n[1]] = 0;
> -	linebuf[n[3]] = 0;
> -	linebuf[n[5]] = 0;
> -	linebuf[n[7]] = 0;
> -
> -	mnt->mnt_fsname = linebuf+n[0];
> -	mnt->mnt_dir = linebuf+n[2];
> -	mnt->mnt_type = linebuf+n[4];
> -	mnt->mnt_opts = linebuf+n[6];
> +
> +		linebuf += strspn(linebuf, " \t");
> +	} while (linebuf[0] == '\0' || linebuf[0] == '#');
> +
> +	mnt->mnt_fsname = strsep(&linebuf, " \t");
> +
> +	if (linebuf) linebuf += strspn(linebuf, " \t");
> +	sub = strsep(&linebuf, " \t");
> +	mnt->mnt_dir = sub ? sub : (char *) "";

"" already has type char[1] and decays to char *; no cast is needed
here.

> +
> +	if (linebuf) linebuf += strspn(linebuf, " \t");
> +	sub = strsep (&linebuf, " \t");
> +	mnt->mnt_type = sub ? sub : (char *) "";
> +
> +	if (linebuf) linebuf += strspn(linebuf, " \t");
> +	sub = strsep(&linebuf, " \t");
> +	mnt->mnt_opts = sub ? sub : (char *) "";
> +
> +	switch (linebuf ? sscanf(linebuf, " %d %d", &mnt->mnt_freq, &mnt->mnt_passno) : 0) {
> +	case 0: mnt->mnt_freq = 0;
> +	case 1: mnt->mnt_passno = 0;
> +	case 2: break;
> +	}
>  
>  	return mnt;
>  }
> -- 
> 2.25.4

This is gratuitously rewriting a lot of parsing logic in a form that
doesn't seem like it's better, and even if it were, the change is
orthogonal to fixing the behavior. I'm sorry for taking so long to get
back to you and say this clearly.

I do want to move forward on this because I know folks have been
waiting on an upstream fix for a long time. But I need a patch
accompanied by a clear explanation of the behavioral changes it's
making, and just those changes. Or if we're in agreement on what the
behavioral changes should be, I can just write the patch.

The patch by Alyssa Ross is more minimal and better documented, but
I'm not sure it covers everything you're concerned about. Could you
let me know whether it does? I'll follow up on that thread about
whether there are open issues with it.

Rich

  parent reply	other threads:[~2022-01-09  3:13 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-15 12:20 Kaihang Zhang
2021-12-01 12:33 ` [musl] " Kaihang Zhang
2021-12-01 15:24 ` [musl] " Rich Felker
2022-01-09  3:12 ` Rich Felker [this message]
2022-01-18 10:17   ` Kaihang Zhang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220109031250.GN7074@brightrain.aerifal.cx \
    --to=dalias@libc.org \
    --cc=2010267516@qq.com \
    --cc=kaihang.zhang@smartx.com \
    --cc=musl@lists.openwall.com \
    --subject='Re: [musl] [PATCH v2] fix: Truncate the too-long mntent in function getmntent_r' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Code repositories for project(s) associated with this inbox:

	https://git.vuxu.org/mirror/musl/

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).