On Sat, Feb 20, 2021 at 11:51:00PM +0100, Peter Zijlstra wrote:
> > 11 patches and one beer later, it even boots :-)
Yah, beer and coding sometimes works. But only sometimes, ask rostedt
and tglx.
:-P
> > Saves more than 6k on a defconfig build.
Uuh, niice. And that will be a lot more on a all{
On Sun, Feb 21, 2021 at 06:45:54AM +0100, Jürgen Groß wrote:
> If you want I can cook up a patch and include it in my paravirt
> cleanup series.
Sure, Linus already pulled the first part of your cleanup so you can
base off the rest ontop.
Thx.
--
Regards/Gruss,
Boris.
https://people.kernel
On 20.02.21 23:32, Peter Zijlstra wrote:
On Sat, Feb 20, 2021 at 06:41:01PM +0100, Borislav Petkov wrote:
- if we had negative alternatives objtool doesn't need to actually
rewrite code in this case. It could simply emit alternative entries
and call it a day.
I don't mind the negativ
On Sat, Feb 20, 2021 at 11:28:02PM +0100, Peter Zijlstra wrote:
> On Sat, Feb 20, 2021 at 06:41:01PM +0100, Borislav Petkov wrote:
> > > - I have more cases for objtool to rewrite code (I'll see if I can
> > >rebase and post that this weekend -- no promises).
> >
> > Oh noes.
>
> 11 patches
On Sat, Feb 20, 2021 at 06:41:01PM +0100, Borislav Petkov wrote:
> > - if we had negative alternatives objtool doesn't need to actually
> >rewrite code in this case. It could simply emit alternative entries
> >and call it a day.
>
> I don't mind the negative alt per se - I mind the implem
On Sat, Feb 20, 2021 at 06:41:01PM +0100, Borislav Petkov wrote:
> > - I have more cases for objtool to rewrite code (I'll see if I can
> >rebase and post that this weekend -- no promises).
>
> Oh noes.
11 patches and one beer later, it even boots :-)
Saves more than 6k on a defconfig build
On Sat, Feb 20, 2021 at 05:48:46PM +0100, Peter Zijlstra wrote:
> - straight line execution is always better than a round-trip to
>somewhere else, no matter how trivial.
Sure, but not at that price. Especially not if it is waaay down in perf
profiles.
> - supposely EIBRS (yeah, I know, ther
On Sat, Feb 20, 2021 at 01:39:20AM +0100, Borislav Petkov wrote:
> On Fri, Feb 19, 2021 at 11:01:58PM +0100, Peter Zijlstra wrote:
> > We could, but it so happens Joerg is also wanting negative features.
>
> Juergen.
Argh! I should stick to jgross. Sorry.
> > So I was thikning that perhaps we c
On Fri, Feb 19, 2021 at 11:01:58PM +0100, Peter Zijlstra wrote:
> We could, but it so happens Joerg is also wanting negative features.
Juergen.
> So I was thikning that perhaps we can convince Boris they're not
> really all that aweful after all :-)
Well, I'm not crazy about this, TBH - I total
On Fri, Feb 19, 2021 at 03:55:30PM -0600, Josh Poimboeuf wrote:
> On Fri, Feb 19, 2021 at 09:43:06PM +0100, Peter Zijlstra wrote:
> > Arguably it would be simpler to do the other way around, but
> > unfortunately alternatives don't work that way, we cannot say:
> >
> > ALTERNATIVE "call __x86_
On Fri, Feb 19, 2021 at 09:43:06PM +0100, Peter Zijlstra wrote:
> Arguably it would be simpler to do the other way around, but
> unfortunately alternatives don't work that way, we cannot say:
>
> ALTERNATIVE "call __x86_indirect_thunk_\reg",
> "call *%reg", ~X86_FEATURE_RET
When the compiler emits: "CALL __x86_indirect_thunk_\reg" for an
indirect call, have objtool rewrite it to:
ALTERNATIVE "call *%reg",
"call __x86_indirect_thunk_\reg", X86_FEATURE_RETPOLINE
That is, rewrite the thunk calls into actual indirect calls and emit
alternativ
12 matches
Mail list logo