Re: [PATCH] xen/decompress: make helper symbols static

[Jan Beulich <jbeulich@xxxxxxxx>]

On 19.01.2021 15:58, Ian Jackson wrote:
> Jan Beulich writes ("Re: [PATCH] xen/decompress: make helper symbols static"):
>> On 18.01.2021 17:09, Ian Jackson wrote:
>>> AFAICT, this patch
>>>
>>> * was first posted after the last posting date for Xen 4.15.
>>> * is not a bugfix.
>>>
>>> In which case it has missed 4.15.
>>>
>>> Please correct me if you think I am wrong.
>>
>> You aren't, and I don't view this as a big problem - its more of
>> an aid to be sure no bad (unintended) references exist (Arm's
>> unhelpful re-use of xz's CRC32 function triggered the change).
> 
> Ah I see.  Thanks for that explanation.  Well, I might be inclined to
> grant a freeze exception on the basis that the point of this is to
> remove rather than add risk.
> 
> If you think that might be sensible, would you care to provide a frank
> assessment of the risks to 4.15 of taking this patch, vs. the risks to
> 4.15 of postponing it ?

Considering that we've been building fine without this adjustment,
I don't think the risk of not taking the patch is non-negligible.
The added safety is more for future work, to prevent anyone
mistakenly using any of the so far globally available symbols.
(There's actually more to do I think, in a separate patch, further
reducing exposure of the just introduced zstd decompressor
functions. I have this on my list of things to look into.)

Taking the patch has a minimal risk of breaking the build, in
case in some configuration a bad reference actually exists. I
can't see any other risk, as the actual code doesn't get changed
at all.

Jan