Re: [PATCH 2/3] x86/entry: Make #PF/NMI/INT0x82 more amenable to livepatching

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Tue, Jan 23, 2024 at 02:43:15PM +0100, Jan Beulich wrote:
> On 23.01.2024 14:37, Roger Pau Monné wrote:
> > On Tue, Jan 23, 2024 at 10:22:10AM +0100, Jan Beulich wrote:
> >> On 22.01.2024 19:17, Andrew Cooper wrote:
> >>> It is bad form to have inter-function fallthrough.  It only functions 
> >>> right
> >>> now because alignment padding bytes are NOPs.
> >>
> >> But that's a requirement anyway in executable sections.
> > 
> > Really?  I was under the impression we wanted to replace the padding
> > nops with rets maybe, or even poison the padding with int3 or ud2.
> 
> Well, that would be a decision of ours. Which then imo can't be described as
> "only functions right now because ..." The assembler can't[1] use other than
> NOPs by default, as it can't know whether fall-through is intended.

So it's not a strict requirement of ELF that padding is done using
nops, it's just the default decision of the assembler because it
doesn't know better.

Thanks, Roger.