netdev - Re: [PATCH net-next v2 00/10] make drivers/net/ethernet W=1 clean

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200915101124.00004146@intel.com>
Date:   Tue, 15 Sep 2020 10:11:24 -0700
From:   Jesse Brandeburg <jesse.brandeburg@...el.com>
To:     Saeed Mahameed <saeed@...nel.org>
Cc:     netdev@...r.kernel.org, intel-wired-lan@...ts.osuosl.org
Subject: Re: [PATCH net-next v2 00/10] make drivers/net/ethernet W=1 clean

Saeed Mahameed wrote:
> Reviewed-by: Saeed Mahameed <saeedm@...dia.com>

Thanks! In all fairness, Jakub reviewed this in v1 too but I made enough
changes in v2 that I felt I had to drop the previous review ACKs.

> Hi Jesse, 
> What was the criteria to select which drivers to enable in your .config
> ?

As Jakub said, I'm using allmodconfig on x86_64, but just yesterday
discovered there was much more to fix because I ran the kernel-doc
script directly on the source (those other things come from different
ARCH= builds which limit allmodconfig)

> I think we need some automation here and have a well known .config that
> enables as many drivers as we can for static + compilation testing,
> otherwise we are going to need to repeat this patch every 2-3 months.

Totally agree! Jakub already has some cobbled together and is regularly
running W=1 C=1 builds on all new patches. I found I could cross compile
different ARCH targets to get (some of) the other warnings, or better
yet just run the scripts/kernel-doc script directly in automation.

> I know Jakub and Dave do some compilation testing before merging but i
> don't know how much driver coverage they have and if they use a
> specific .config or they just manually create one on demand..
> 
> bottom line, we need a bot after this series is applied.
> All we need is to daily apply all ongoing patches to some testing
> branch and let 0-DAY kernel test [1] run on it with whatever make
> command we define and with all drivers enabled.

Yes, that's the end goal and I think this moves us closer to that. A
little more work remains before we go and turn all warnings on - as
Andrew suggested in another reply. (it's also sometimes a losing game
fighting against many compiler versions, etc).  However, the zero-day
bot reporting more results from W=1 compiles would *really* help (I
looked at , but am having some troubles verifying that)

> [1] https://lists.01.org/pipermail/kbuild-all 
> 
> > ---
> > 
> > Q: Maybe I can fix the remaining warnings in a followup patch? If
> > I try to put it on this series it will make it much larger
> > (double).