Re: Fix built-in early-load Intel microcode alignment

Re: Fix built-in early-load Intel microcode alignment

[Luis Chamberlain]

On Mon, Jan 13, 2020 at 11:44:25AM -0800, Linus Torvalds wrote:
> On Mon, Jan 13, 2020 at 7:47 AM Luis Chamberlain <mcgrof@kernel.org> wrote:
> >
> > So I'd like to determine first if we really need this. Then if so,
> > either add a new global config option, and worst comes to worst
> > figure out a way to do it per driver. I don't think we'd need it
> > per driver.
> 
> I really don't think we need to have a config option for some small
> alignment. Increasing the alignment unconditionally to 16 bytes won't
> hurt anybody.

Since you are confident in that, then simply bumping it to 16 bytes
seems fine by me.

> Now, whether there might be other firmware loaders that need even more
> alignment, that might be an interesting question, and if such an
> alignment would be _huge_ we might want to worry about actual memory
> waste.

I can only envision waste being considered due to alignent for remote
proc folks, who I *doubt* use the built-in stuff given the large size of
their blobs... but since you never know, better poke. So I've CC'd them.

> But 16-byte alignment for a fw blob? That's nothing.

Fine by me if we are sure it won't break anything and we hear no
complaints by remote proc folks.

  Luis

Re: Fix built-in early-load Intel microcode alignment

[Bjorn Andersson]

On Tue 14 Jan 18:27 PST 2020, Luis Chamberlain wrote:

> On Mon, Jan 13, 2020 at 11:44:25AM -0800, Linus Torvalds wrote:
> > On Mon, Jan 13, 2020 at 7:47 AM Luis Chamberlain <mcgrof@kernel.org> wrote:
> > >
> > > So I'd like to determine first if we really need this. Then if so,
> > > either add a new global config option, and worst comes to worst
> > > figure out a way to do it per driver. I don't think we'd need it
> > > per driver.
> > 
> > I really don't think we need to have a config option for some small
> > alignment. Increasing the alignment unconditionally to 16 bytes won't
> > hurt anybody.
> 
> Since you are confident in that, then simply bumping it to 16 bytes
> seems fine by me.
> 
> > Now, whether there might be other firmware loaders that need even more
> > alignment, that might be an interesting question, and if such an
> > alignment would be _huge_ we might want to worry about actual memory
> > waste.
> 
> I can only envision waste being considered due to alignent for remote
> proc folks, who I *doubt* use the built-in stuff given the large size of
> their blobs... but since you never know, better poke. So I've CC'd them.
> 

I've not heard of anyone using built-in firmware with remoteproc, but as
you say firmware used with remoteproc is large. So I can't see there
being a problem of potentially wasting 8 bytes...

> > But 16-byte alignment for a fw blob? That's nothing.
> 
> Fine by me if we are sure it won't break anything and we hear no
> complaints by remote proc folks.
> 

Go for it.

Regards,
Bjorn