[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20171115.095427.136250865702817393.davem@davemloft.net>
Date: Wed, 15 Nov 2017 09:54:27 +0900 (KST)
From: David Miller <davem@...emloft.net>
To: ecree@...arflare.com
Cc: amsalam20@...il.com, david.lebrun@...ouvain.be,
netdev@...r.kernel.org
Subject: Re: [PATCH] ipv6: sr: update the struct ipv6_sr_hdr
From: Edward Cree <ecree@...arflare.com>
Date: Tue, 14 Nov 2017 14:14:01 +0000
> On 14/11/17 12:37, David Miller wrote:
>> From: Ahmed Abdelsalam <amsalam20@...il.com>
>> Date: Sun, 12 Nov 2017 21:37:01 +0100
>>
>>> diff --git a/include/uapi/linux/seg6.h b/include/uapi/linux/seg6.h
>>> index 2f6fb0d..3f4b3ab 100644
>>> --- a/include/uapi/linux/seg6.h
>>> +++ b/include/uapi/linux/seg6.h
>>> @@ -26,9 +26,9 @@ struct ipv6_sr_hdr {
>>> __u8 hdrlen;
>>> __u8 type;
>>> __u8 segments_left;
>>> - __u8 first_segment;
>>> + __u8 last_entry;
>>
>> This is user ABI and cannot be changed.
>>
>> Sorry, folks should have considered these issues when the SR
>> changes were submitted. This field must keep the name 'first_segment'
>> forever.
>
> Surely renaming struct fields only changes the API, not the ABI?
> Binaries compiled against the old struct definition will still behave the
> same (AFAICT the patch doesn't change how these fields are used), while
> sources being recompiled (so they care about the name change) can be
> changed.
Yes, it will stop existing applications from compiling and we can't
do that.
Powered by blists - more mailing lists