Re: [PATCH-4.16] arm/efi: Improve performance requesting filesystem handle

[Ian Jackson <iwj@xxxxxxxxxxxxxx>]

Luca Fancellu writes ("[PATCH-4.16] arm/efi: Improve performance requesting 
filesystem handle"):
> Currently, the code used to handle and possibly load from the filesystem
> modules defined in the DT is allocating and closing the filesystem handle
> for each module to be loaded.
> 
> To improve the performance, the filesystem handle pointer is passed
> through the call stack, requested when it's needed only once and closed
> if it was allocated.

Thsnks for submitting this formally.  The clarity is very helpful.

> Signed-off-by: Luca Fancellu <luca.fancellu@xxxxxxx>
> ---
> Justification for integration in 4.16:
> Upside:   Fixes a performance issue only on arm that would be really useful
>           for ARM users using the release, no functional changes.

The size of this perf issue is not discussed here.  At this stage of
the release I will usually give release acks for fixes to serious
problems.  Does this meet that test ?  I think probably not.

Thanks to Bertrand who, when I asked on IRC, confirmed that this perf
issue only affects users of the new feature in a1743fc3a9 (so is not a
regression), and the perf issue is only a consquenceof 9bc9fff04ba0
which was itself a fix to something wrong with a1743fc3a9.  It seems
that one interpretation is that the problem here is simply a
less-than-optimal new feature.

> Downside: It's affecting the EFI boot path (only on ARM).
> Risk:     Risk is low given the small changes.

I did find this risk analysis at all convincing, I'm afraid.

So I'm afraid I'm saying "no".

Ian.