running leaking_addresses.pl

running leaking_addresses.pl

[Tobin C. Harding]

Hi,

Recently scripts/leaking_addresses.pl was merged into the mainline with
the hope of catching leaking kernel addresses.

Would it be in scope for this script to be run by the kbuild test robot?
Excuse my very little knowledge of the kbuild test robot but would this
lead to the script being run on a number of kernels with varying
configuration?

In the event that this may be a possibility can I add that I do not have
a suggestion for what to do with the output, can it go to LKML? I'm
happy to be CC'd on the output to help investigate.

thanks,
Tobin.

Re: running leaking_addresses.pl

[Fengguang Wu]

Hi Tobin,

On Tue, Dec 05, 2017 at 10:36:13AM +1100, Tobin C. Harding wrote:
>Hi,
>
>Recently scripts/leaking_addresses.pl was merged into the mainline with
>the hope of catching leaking kernel addresses.
>
>Would it be in scope for this script to be run by the kbuild test robot?
>Excuse my very little knowledge of the kbuild test robot but would this
>lead to the script being run on a number of kernels with varying
>configuration?

Yes it's in our plan. Bill Roberts also requested testing leaking_addresses.pl
and Philip has scheduled it to next quarter.

>In the event that this may be a possibility can I add that I do not have
>a suggestion for what to do with the output, can it go to LKML? I'm
>happy to be CC'd on the output to help investigate.

That'd be great! In general we are mainly responsible for catching
regressions. Due to limit of time and expertise, we leave the analyze
and bug fixing works to the first bad commit's authors and anyone who
are interested in looking into a class of problems. For example, we'll
CC some smatch and coccinelle regressions (possibly false positive) to
Dan and Julia.

Thanks,
Fengguang

Re: running leaking_addresses.pl

[Tobin C. Harding]

On Tue, Dec 05, 2017 at 11:39:12AM +0800, Fengguang Wu wrote:
> Hi Tobin,
> 
> On Tue, Dec 05, 2017 at 10:36:13AM +1100, Tobin C. Harding wrote:
> >Hi,
> >
> >Recently scripts/leaking_addresses.pl was merged into the mainline with
> >the hope of catching leaking kernel addresses.
> >
> >Would it be in scope for this script to be run by the kbuild test robot?
> >Excuse my very little knowledge of the kbuild test robot but would this
> >lead to the script being run on a number of kernels with varying
> >configuration?
> 
> Yes it's in our plan. Bill Roberts also requested testing leaking_addresses.pl
> and Philip has scheduled it to next quarter.

Awesome.

> >In the event that this may be a possibility can I add that I do not have
> >a suggestion for what to do with the output, can it go to LKML? I'm
> >happy to be CC'd on the output to help investigate.
> 
> That'd be great! In general we are mainly responsible for catching
> regressions. Due to limit of time and expertise, we leave the analyze
> and bug fixing works to the first bad commit's authors and anyone who
> are interested in looking into a class of problems. For example, we'll
> CC some smatch and coccinelle regressions (possibly false positive) to
> Dan and Julia.

Ok cool, sign me up. I'm keen as mustard.

thanks,
Tobin