[PATCH v4] Fix potential access beyond array bounds in m1np
Andreas Jaeger
aj@suse.com
Tue Jan 1 13:34:00 GMT 2013
On 12/29/2012 02:43 AM, Siddhesh Poyarekar wrote:
> On Fri, Dec 28, 2012 at 06:02:51PM +0100, Andreas Jaeger wrote:
>>> + if (__glibc_unlikely (p < 18 && m <= 0)) {
>>
>> I would rather have an assert (p < 18) here - or is there a specific
>> reason for not having it?
>>
>
> I was just being conservative; I don't mind an assert either if you
> prefer that. Here's v4 then - I have cleaned up the code styling in
> that block since the earlier 'style' is ugly.
This is fine, thanks,
Andreas
> Siddhesh
>
> * sysdeps/ieee754/dbl-64/mpexp.c (__mpexp): Add assert to
> check access beyond bounds of m1np.
>
> diff --git a/sysdeps/ieee754/dbl-64/mpexp.c b/sysdeps/ieee754/dbl-64/mpexp.c
> index c5a0283..53e8b74 100644
> --- a/sysdeps/ieee754/dbl-64/mpexp.c
> +++ b/sysdeps/ieee754/dbl-64/mpexp.c
> @@ -31,6 +31,7 @@
> #include "endian.h"
> #include "mpa.h"
> #include "mpexp.h"
> +#include <assert.h>
>
> #ifndef SECTION
> # define SECTION
> @@ -71,10 +72,22 @@ __mpexp(mp_no *x, mp_no *y, int p) {
> for (i=2; i<=p; i++) { if (X[i]!=ZERO) break; }
> if (i==p+1) { m2--; a *= TWO; }
> }
> - if ((m=m1+m2) <= 0) {
> - m=0; a=ONE;
> - for (i=n-1; i>0; i--,n--) { if (m1np[i][p]+m2>0) break; }
> - }
> +
> + m = m1 + m2;
> + if (__glibc_unlikely (m <= 0))
> + {
> + /* The m1np array which is used to determine if we can reduce the
> + polynomial expansion iterations, has only 18 elements. Besides,
> + numbers smaller than those required by p >= 18 should not come here
> + at all since the fast phase of exp returns 1.0 for anything less
> + than 2^-55. */
> + assert (p < 18);
> + m = 0;
> + a = ONE;
> + for (i = n - 1; i > 0; i--, n--)
> + if (m1np[i][p] + m2 > 0)
> + break;
> + }
>
> /* Compute s=x*2**(-m). Put result in mps */
> __dbl_mp(a,&mpt1,p);
>
--
Andreas Jaeger aj@{suse.com,opensuse.org} Twitter/Identica: jaegerandi
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn,Jennifer Guild,Felix Imendörffer,HRB16746 (AG Nürnberg)
GPG fingerprint = 93A3 365E CE47 B889 DF7F FED1 389A 563C C272 A126
More information about the Libc-alpha
mailing list