From: Linus Torvalds <torvalds@linux-foundation.org>
To: "Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>
Cc: linux-kernel@vger.kernel.org, Michael Buesch <mb@bu3sch.de>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] strcmp: fix overflow error
Date: Tue, 17 Nov 2009 09:41:58 -0800 (PST) [thread overview]
Message-ID: <alpine.LFD.2.01.0911170922190.9384@localhost.localdomain> (raw)
In-Reply-To: <1258476700-21323-1-git-send-email-u.kleine-koenig@pengutronix.de>
On Tue, 17 Nov 2009, Uwe Kleine-König wrote:
>
> strcmp("\x01", "\xef") returns 18 but it should return something < 0.
> The reason is that the variable holding the result of the subtraction is
> too small and overflows.
No. The reason is that whoever wrote that function is a moron and doesn't
know the standard. And your fix is not correct _either_
The comparison should be done as *unsigned char*. As specified by POSIX
"The sign of a non-zero return value shall be determined by the sign of
the difference between the values of the first pair of bytes (both
interpreted as type unsigned char) that differ in the strings being
compared."
and both the original code and your change gets it wrong in different
ways.
> int strcmp(const char *cs, const char *ct)
> {
> - signed char __res;
> + int __res;
>
> while (1) {
> if ((__res = *cs - *ct++) != 0 || !*cs++)
So this is fundamentally incorrect both with "signed char __res" _and_
with "int __res", because '*cs' and '*ct' are both (possibly - it depends
on the compiler and architecture) signed chars.
So in the case you mention, strcmp() _should_ return a negative value,
because "\x01" is smaller than "\xef", but you have:
- *cs = 1, *ct = (char) 0xef = -17 _OR_ 239 depending on sign of 'char'
and as a result:
- signed char __res = 18 (incorrect, regardless: ct is larger)
- int __res = 18 (incorrect) or -238 (correct) depending on sign of char
so your patch doesn't actually help at all.
What would help is something like the appended, but I have not tested it
AT ALL. It may be total and utter crap too. Maybe it doesn't compile,
maybe it buggers your pet hedgehog. I just don't know.
Linus
---
lib/string.c | 20 ++++++++++++++------
1 files changed, 14 insertions(+), 6 deletions(-)
diff --git a/lib/string.c b/lib/string.c
index b19b87a..e96421a 100644
--- a/lib/string.c
+++ b/lib/string.c
@@ -246,13 +246,17 @@ EXPORT_SYMBOL(strlcat);
#undef strcmp
int strcmp(const char *cs, const char *ct)
{
- signed char __res;
+ unsigned char c1, c2;
while (1) {
- if ((__res = *cs - *ct++) != 0 || !*cs++)
+ c1 = *cs++;
+ c2 = *ct++;
+ if (c1 != c2)
+ return c1 < c2 ? -1 : 1;
+ if (!c1)
break;
}
- return __res;
+ return 0;
}
EXPORT_SYMBOL(strcmp);
#endif
@@ -266,14 +270,18 @@ EXPORT_SYMBOL(strcmp);
*/
int strncmp(const char *cs, const char *ct, size_t count)
{
- signed char __res = 0;
+ unsigned char c1, c2;
while (count) {
- if ((__res = *cs - *ct++) != 0 || !*cs++)
+ c1 = *cs++;
+ c2 = *ct++;
+ if (c1 != c2)
+ return c1 < c2 ? -1 : 1;
+ if (!c1)
break;
count--;
}
- return __res;
+ return 0;
}
EXPORT_SYMBOL(strncmp);
#endif
next prev parent reply other threads:[~2009-11-17 17:42 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-17 16:51 Uwe Kleine-König
2009-11-17 17:36 ` Andreas Schwab
2009-11-17 17:41 ` Linus Torvalds [this message]
2009-11-17 18:16 ` Michael Buesch
2009-11-17 18:40 ` Linus Torvalds
2009-11-17 20:34 ` Andreas Schwab
2009-11-17 18:55 ` Uwe Kleine-König
2009-11-17 19:02 ` Linus Torvalds
2009-11-17 19:12 ` Linus Torvalds
2009-11-17 19:19 ` Joe Perches
2009-11-18 21:31 ` [PATCH] strcmp: fix overflow and possibly signedness error Uwe Kleine-König
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=alpine.LFD.2.01.0911170922190.9384@localhost.localdomain \
--to=torvalds@linux-foundation.org \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mb@bu3sch.de \
--cc=u.kleine-koenig@pengutronix.de \
/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.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®