On Tue, Jul 22, 2008 at 04:54:13PM +0800, Kent Liu wrote:
> Hello,
>
> numademo will crash if some options are added. For example, a simple
> "numademo -c 20m memcpy" will cause a segmentation fault after memcpy
> tests completed.
>
> This issue has been there for long time as I can see it in both SuSE
> and RHEL distros, and 2.0.2-rc2 package.
>
> This patch fixes the segfault by adjusting variable ac to avoid array
> index overflow.
>
> Signed-off-by: Kent Liu <kent.liu@xxxxxxxxxxxxxxx>
> ---
> numademo.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/numademo.c b/numademo.c
> index df937d5..0dd5476 100755
> --- a/numademo.c
> +++ b/numademo.c
> @@ -448,6 +448,7 @@ int main(int ac, char **av)
> int simple_tests = 0;
>
> while (av[1] && av[1][0] == '-') {
> + ac--;
> switch (av[1][1]) {
> case 'c':
> delim = ",";
> --
> To unsubscribe from this list: send the line "unsubscribe linux-numa" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
and
> Original Min/Max in numademo output is confusing. The 'Min' meant the
> minimum time the operation have been occupied, while the printed numbers
> are bandwidth data, which has reversed meaning in terms of quantity.
>
> Reverting Min/Max in the print message is better.
I merged both of your patches.
It passes "make test" on an Altix.
If you would, please test ftp://oss.sgi.com/www/projects/libnuma/download/
numactl-2.0.2-rc3.tar.gz
Thanks.
-Cliff
--
Cliff Wickman
Silicon Graphics, Inc.
cpw@xxxxxxx
(651) 683-3824
--
To unsubscribe from this list: send the line "unsubscribe linux-numa" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
