Re: [PATCH 03/10] efi/x86: Use pr_efi_err for error messages

[Arvind Sankar]

On Wed, Apr 29, 2020 at 11:55:04PM +0200, Ard Biesheuvel wrote:
> On Wed, 29 Apr 2020 at 23:53, Joe Perches <joe@xxxxxxxxxxx> wrote:
> >
> > On Wed, 2020-04-29 at 17:43 -0400, Arvind Sankar wrote:
> > > On Wed, Apr 29, 2020 at 08:49:21PM +0200, Ard Biesheuvel wrote:
> > > > On Wed, 29 Apr 2020 at 20:47, Joe Perches <joe@xxxxxxxxxxx> wrote:
> > > > > On Wed, 2020-04-29 at 13:41 -0400, Arvind Sankar wrote:
> > > > > > Use pr_efi_err instead of bare efi_printk for error messages.
> > > > >
> > > > > Perhaps it'd be better to rename pr_efi_err to eri_err
> > > > > so it's clearer it's a typical efi_ logging function.
> > > > >
> > > > > $ git grep -w --name-only pr_efi_err | \
> > > > >   xargs sed -i 's/\bpr_efi_err\b/efi_err/g'
> > > > >
> > > >
> > > > Yeah, pr_efi_err() is probably not the best name
> > >
> > > Should I rename pr_efi/pr_efi_err to, say, efi_pr_info/efi_pr_error?
> >
> > Perhaps not use pr_ in the name at all.
> >
> > I suggest:
> >
> > pr_efi          -> efi_info (or efi_debug or efi_dbg)
> >                    (it is guarded by an efi_quiet flag, default: on)
> > pr_efi_err      -> efi_err
> >
> 
> Agreed. Shorter is better if there is no risk of confusion..

Ok, I'll use efi_info/efi_err. We could add debugging output as
efi_debug later, enabled if efi=debug is specified.

While we're here: most of the existing cases of pr_efi look like notice
or info level, except maybe these two, which probably should be at least
warnings?

drivers/firmware/efi/libstub/arm64-stub.c
62: pr_efi("EFI_RNG_PROTOCOL unavailable, no randomness supplied\n");

drivers/firmware/efi/libstub/efi-stub.c
254: pr_efi("Ignoring DTB from command line.\n");