[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-raid
Subject:    Re: [PATCH mdadm] Makefile: support latest gcc: address-of-packed-member
From:       NeilBrown <neilb () suse ! de>
Date:       2019-10-31 3:18:30
Message-ID: 87h83peh6h.fsf () notabene ! neil ! brown ! name
[Download RAW message or body]


On Wed, Oct 30 2019, Jes Sorensen wrote:

> On 10/30/19 7:58 PM, NeilBrown wrote:
>> 
>> super-intel often takes the address of a packed member,
>> and seems to work.
>> So suppress this warning.
>> (Earlier gcc ignore the new flag)
>> 
>> Signed-off-by: NeilBrown <neilb@suse.de>
>
> I am kinda in two minds about this. I started cleaning up some of the 
> newer gcc stuff a while ago, but then got stuck on super-intel.c
>
> I want the code to build, but I also feel super-intel.c needs cleaning 
> up and made to use accessor functions or something like that to deal 
> with these accesses in a better way?
>
> Thoughts?
>
I wonder how much we really need the __attribute((packed)) ...
The structures look well organized and dense to me.
Maybe remove the attribute and add a BUILD_BUG_ON() if the sizeof()
is wrong??

NeilBrown

["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]