From: Rich Felker <dalias@libc.org>
To: Markus Wichmann <nullplan@gmx.net>
Cc: musl@lists.openwall.com
Subject: Re: [musl] [PATCH] fix return value of wcs{,n}cmp for near-limits signed wchar_t values
Date: Fri, 6 Jan 2023 06:17:24 -0500 [thread overview]
Message-ID: <20230106111724.GA4163@brightrain.aerifal.cx> (raw)
In-Reply-To: <20230106092010.GA2032@voyager>
On Fri, Jan 06, 2023 at 10:20:10AM +0100, Markus Wichmann wrote:
> On Wed, Jan 04, 2023 at 04:07:19PM +0100, Gabriel Ravier wrote:
> > int wcscmp(const wchar_t *l, const wchar_t *r)
> > {
> > for (; *l==*r && *l && *r; l++, r++);
>
> I just noticed this line. Isn't the "&& *r" part superfluous? If r is a
> prefix of l, then *r and *l will be unequal, and the loop will terminate
> because of the first condition alone. (If l is a prefix of r, we need
> the second condition to terminate the loop.)
Yes, but I would assume the compiler would make the same optimization
anyway. The original motivation here may have just been writing it
symmetrically in l and r.
> > - return *l - *r;
> > + return *l < *r ? -1 : *l > *r;
> > }
>
> Ah, that old bug. The problem is that the difference between two 32-bit
> values takes up 33 bits to save. I wonder if it would be worth it to
> just cast the values to 64 bits, then dividing the result by two. You
> know, like
>
> return ((int64_t)*l - *r) >> 1;
>
> Although that does presuppose that wchar_t is smaller than 64 bits,
> which the proposed change does not require.
As you noted later this doesn't work but I think the core flaw here is
not the same as the classic bug. Rather, I probably had a wrong
initial assumption that the function was intended only to work for
valid wide character values of wchar_t not arbitrary integers that fit
in wchar_t arrays. In that case they would be 21-bit and the
difference would never overflow.
Rich
next prev parent reply other threads:[~2023-01-06 11:17 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-04 15:07 Gabriel Ravier
2023-01-06 9:20 ` Markus Wichmann
2023-01-06 9:31 ` Markus Wichmann
2023-01-06 11:17 ` Rich Felker [this message]
2023-01-06 16:57 ` Markus Wichmann
2023-01-07 1:51 ` Gabriel Ravier
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=20230106111724.GA4163@brightrain.aerifal.cx \
--to=dalias@libc.org \
--cc=musl@lists.openwall.com \
--cc=nullplan@gmx.net \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public 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).