>>> On 07.03.16 at 15:41, <kon...@kernel.org> wrote: > On Mon, Mar 7, 2016 at 5:44 AM, Jan Beulich <jbeul...@suse.com> wrote: >>>>> On 04.03.16 at 21:55, <konrad.w...@oracle.com> wrote: >>>> + case XEN_SYSCTL_LOGLVL_set: >>>> + if ( (op->host.lower_thresh >= 0 && op->host.upper_thresh >= 0 && >>>> + op->host.lower_thresh > op->host.upper_thresh) || >>>> + (op->guest.lower_thresh >= 0 && op->guest.upper_thresh >= 0 >>>> && >>>> + op->guest.lower_thresh > op->guest.upper_thresh) ) >>>> + return -EINVAL; >>>> + >>>> + do_loglvl_op(&op->host, &xenlog_lower_thresh, >>>> + &xenlog_upper_thresh, "standard"); >>> >>> >>> The keyboard and the sysctl both allow the user to go beyound the XENLOG_ >>> values we have. That is you could set the lower and upper threshold to be >>> at 9 (or more) say. It will have the same effect as XENLOG_DEBUG (which is >>> 4) >>> as printk_prefix_check seems to have a simple < check. >>> >>> But perhaps to be correct only accept only proper values? Not allow >>> the system admin to set the level to say 31415? >> >> Since there's no bad side effect from doing so I opted for not >> adding respective extra checks, keeping the code easier to read. >> > > Fair enough. Could you perhaps just add that in the commit description?
Sure. > Also I noticed that this patch is missing an XSM check in flask_sysctl > - could that be added please? Of course; it's pretty ugly that one doesn't notice the lack thereof via a build failure. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
