linux-kernel - Re: revert "fs/befs/linuxvfs.c: replace strncpy by strlcpy"

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20150428160523.GJ889@ZenIV.linux.org.uk>
Date:	Tue, 28 Apr 2015 17:05:23 +0100
From:	Al Viro <viro@...IV.linux.org.uk>
To:	Fabian Frederick <fabf@...net.be>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	linux-kernel@...r.kernel.org
Subject: Re: revert "fs/befs/linuxvfs.c: replace strncpy by strlcpy"

On Tue, Apr 28, 2015 at 07:35:10AM +0200, Fabian Frederick wrote:

> > Al, very unhappy about the prospect of looking through ~2000 calls of
> > strlcpy()
> > we have in the tree...
> 
> Sorry Al, I thought it was more secure.

It's not just you, unfortunately, and dumping all that annoyance on you
as a proxy for everyone who does that kind of thing had been unfair.
My apologies...

> I guess the 2 following patches should be reversed as well :
> 
> 6cb103b6f45a
> "fs/befs/btree.c: replace strncpy by strlcpy + coding style fixing"
> 
> 69201bb11327
> "fs/ocfs2/super.c: use OCFS2_MAX_VOL_LABEL_LEN and strlcpy"

AFAICS, they should.

Unfortunately, we _can't_ make strlcpy() never look past src + size - 1 -
not without changing its semantics.  Its callers expect it to return
the length of source; one of the intended uses is
	wanted = strlcpy(dst, src, dst_size);
	if (wanted >= dst_size) {
		p = realloc(dst, wanted + 1);
		if (!p) {
			// too bad
		} else {
			dst = p;
			memcpy(dst, src, wanted + 1);
		}
	}
and that really wants the accurate length.  Now, the absolute majority of
strlcpy() users in the kernel completely ignore the return value, i.e. go for
silent truncation.  About 1% do not.

Out of those, some are correctly implemented "fail if truncated" uses.
The rest...  Some are weirdly open-coded snprintf() (series of strlcpy and
strlcat) and some are _very_ dubious.  Either they really never get
truncated, or we have a problem.  For example, this
kernel/module.c:2349:			s += strlcpy(s, &mod->strtab[src[i].st_name],
smells really bad.  We are truncating a bunch of strings dowsn to KSYM_NAME_LEN
there and storing them in a buffer, with spacing that matches _untruncated_
strings.

drivers/s390/scsi/zfcp_fc.c:825:	len = strlcpy(rspn_req->rspn.fr_name, fc_host_symbolic_name(shost),
also looks fishy - what happens there is
        len = strlcpy(rspn_req->rspn.fr_name, fc_host_symbolic_name(shost),
                      FC_SYMBOLIC_NAME_SIZE);
        rspn_req->rspn.fr_name_len = len;

drivers/usb/gadget/function/f_midi.c:976:	result = strlcpy(page, opts->id, PAGE_SIZE);
drivers/usb/gadget/function/f_printer.c:1172:	result = strlcpy(page, opts->pnp_string + 2, PNP_STRING_LEN - 2);
drivers/usb/gadget/function/f_printer.c:1184:	result = strlcpy(opts->pnp_string + 2, page, PNP_STRING_LEN - 2);

are also strange...
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/