Re: [ RFC, PATCH - 1/2, v2 ] x86-microcode: refactor microcode output messages

[Ingo Molnar <mingo@elte.hu>]

* dimm <dmitry.adamushko@gmail.com> wrote:

> [ resending ]
> 
> 
> Hi,
> 
> 
> this is in response to Mike's patch "Limit the number of microcode
> messages".
> 
> What's about the following (yet preliminary and not thoroughly tested)
> approach?
> 
> patch-1:
> 
> simplify 'struct ucode_cpu_info' and related functional logic.
> 
> 
> patch-2: 
> 
> reduce a number of similar 'microcode version' messages by printing a
> single message for all cpus with equal microcode version, like:
> 
> (1)
> [   96.589437] microcode: original microcode versions...
> [   96.589451] microcode: CPU0-1: sig=0x6f2, pf=0x20, revision=0x57
> 
> (2)
> [   96.603176] microcode: microcode versions after update...
> [   96.603193] microcode: CPU0-1: sig=0x6f2, pf=0x20, revision=0x57
> 
> 
> The new approach is used in microcode_init() [ i.e. when loading a
> module ] and microcode_write(), that's when we update all the cpus at
> once.
> 
> reload_for_cpu() and update-all-cpus-upon-resuming() use the old
> approach - a new microcode version is being reported upon applying it.
> 
> The latter might employ the similar 'report-for-all' approach as above
> but that would somewhat complicate the logic. Anyways, there are plenty
> of per-cpu messages upon system resuming so having some more
> update-microcode related ones won't harm that muc, I guess :-)

Seems sensible to me.

	Ingo