[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7993b83e-c53f-6b16-3c3a-53601e0ffe14@intel.com>
Date: Fri, 9 Dec 2022 09:53:46 -0800
From: Jesse Brandeburg <jesse.brandeburg@...el.com>
To: Michal Kubecek <mkubecek@...e.cz>
CC: <netdev@...r.kernel.org>
Subject: Re: [PATCH ethtool v2 09/13] ethtool: merge uapi changes to implement
BIT and friends
On 12/7/2022 10:44 PM, Michal Kubecek wrote:
> On Wed, Dec 07, 2022 at 05:11:18PM -0800, Jesse Brandeburg wrote:
>> I was looking into some errors reported by the runtime sanitizers and
>> found a couple of places where (1 << 31) was being used. This is a shift
>> of a bit into the sign-bit of an integer. This is undefined behavior for
>> the C-specification, and can be easily fixed with using (1UL << 31)
>> instead. A better way to do this is to use the BIT() macro, which
>> already has the 1UL in it (see future patch in series).
>>
>> Convert and sync with the same changes made upstream to the uapi file,
>> to implement ethtool use BIT() and friends.
>
> Please follow the guidelines on updating UAPI header copies in
> "Submitting patches" section of
>
> https://www.kernel.org/pub/software/network/ethtool/devel.html
>
> Fixing fsl_enetc.c within the UAPI update is OK (and definitely better
> than trying to avoid it) but please update all UAPI header copies to the
> state of the same kernel tree commit which will be indicated in the
> commit message.
I'll respin the uapi changes in the kernel to use uapi/linux/const.h and
wait to see how that series goes then.
Powered by blists - more mailing lists