Re: [PATCH 5/7] public / x86: introduce __HYPERCALL_iommu_op

[Paul Durrant <Paul.Durrant@citrix.com>]

> -----Original Message-----
> From: George Dunlap [mailto:george.dunlap@citrix.com]
> Sent: 07 June 2018 14:45
> To: Jan Beulich <JBeulich@suse.com>; Paul Durrant
> <Paul.Durrant@citrix.com>
> Cc: Andrew Cooper <Andrew.Cooper3@citrix.com>; Ian Jackson
> <Ian.Jackson@citrix.com>; Wei Liu <wei.liu2@citrix.com>; Stefano Stabellini
> <sstabellini@kernel.org>; xen-devel <xen-devel@lists.xenproject.org>;
> Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>; Daniel de Graaf
> <dgdegra@tycho.nsa.gov>; Tim (Xen.org) <tim@xen.org>
> Subject: Re: [PATCH 5/7] public / x86: introduce __HYPERCALL_iommu_op
> 
> On 06/07/2018 02:21 PM, Jan Beulich wrote:
> >>>> On 07.06.18 at 13:42, <Paul.Durrant@citrix.com> wrote:
> >>> From: Jan Beulich [mailto:JBeulich@suse.com]
> >>> Sent: 16 March 2018 12:25
> >>>>>> On 12.02.18 at 11:47, <paul.durrant@citrix.com> wrote:
> >>>> --- a/xen/arch/x86/Makefile
> >>>> +++ b/xen/arch/x86/Makefile
> >>>> @@ -33,6 +33,7 @@ obj-$(CONFIG_CRASH_DEBUG) += gdbstub.o
> >>>>  obj-y += hypercall.o
> >>>>  obj-y += i387.o
> >>>>  obj-y += i8259.o
> >>>> +obj-y += iommu_op.o
> >>>
> >>> As mentioned in other contexts, I'd prefer if we stopped using
> >>> underscores in places where dashes (or other separators not
> >>> usable in C identifiers) are fine.
> >>
> >> I don't see any guidance in CODING_STYLE or elsewhere, and also the
> majority
> >> of the codebase seems to prefer using underscores in module names.
> Personally
> >> I'd prefer new code remain consistent.
> >
> > The lack of statement to this effect is why I've said "I'd prefer". See
> > alternative-asm.h, x86-defns.h, or x86-vendors.h for _recent_
> > examples of moving into the other direction. On all keyboards I've
> > seen or used, an underscore requires two keys to be pressed, while
> > a dash takes only one. This isn't much for an individual instance, but
> > it sums up. It's the same reason why I'm advocating against the use
> > of underscores in new command line option names.
> >
> > In the end, looking at the history of typography, I think underscore
> > is a relatively late (and presumably artificial) addition; in particular I
> > don't recall mechanical type writers to even have a key for it.
> 
> <pedantic>The mechanical typewriters I learned on had an underscore to
> allow you to go back and underline words.</pedantic>
> 
> > It's
> > use as a visual separator is necessary in e.g. programming
> > languages, as commonly dash designated the operator for "minus"
> > there. Extending such naming to non-identifiers (file system names
> > and command line options are just prominent examples) is simply
> > misguided imo.
> 
> Well in any case, maybe this should be discussed in a patch to
> CODING_STYLE, rather than in the middle of a patch series about
> something completely different.
> 
> >>>> +#ifndef __XEN_PUBLIC_IOMMU_OP_H__
> >>>> +#define __XEN_PUBLIC_IOMMU_OP_H__
> >>>
> >>> Please can you avoid introducing further name space violations
> >>> into the public headers?
> >>
> >> I assume you mean the leading '__'? Again, I chose the name based on
> >> consistency with other code and I'd prefer to remain consistent. Could
> you
> >> explain why having a leading '__' is problematic?
> >
> > Names starting with double underscores are reserved (as are, btw,
> > names starting with a single underscore and an upper case letter).
> > While it's unlikely for a compiler to ever want to use
> > __XEN_PUBLIC_IOMMU_OP_H__ for its internal purposes, we couldn't
> > validly complain if one did.
> 
> I'm with Jan on this one.  At the moment I'm not sure about using dashes
> instead of underscores for filenames, but in this case the extra
> underscores at the beginning and end are redundant; the "XEN_..._H" is
> sufficient to make the contents unique.
> 

FWIW Linux appears to use a single '_' prefix and no suffix.

  Paul

>  -George
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel