[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <noxbrmequ4inpxgnwna6drnzuymcn3k6ik2kvlbzacxvr2udgf@pximrevgcgmu>
Date: Tue, 18 Jun 2024 15:20:20 +0300
From: "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>
To: Tom Lendacky <thomas.lendacky@....com>
Cc: Borislav Petkov <bp@...en8.de>, Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...hat.com>, Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
"Rafael J. Wysocki" <rafael@...nel.org>, Peter Zijlstra <peterz@...radead.org>,
Adrian Hunter <adrian.hunter@...el.com>,
Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>, Elena Reshetova <elena.reshetova@...el.com>,
Jun Nakajima <jun.nakajima@...el.com>, Rick Edgecombe <rick.p.edgecombe@...el.com>,
"Kalra, Ashish" <ashish.kalra@....com>, Sean Christopherson <seanjc@...gle.com>,
"Huang, Kai" <kai.huang@...el.com>, Ard Biesheuvel <ardb@...nel.org>, Baoquan He <bhe@...hat.com>,
"H. Peter Anvin" <hpa@...or.com>, "K. Y. Srinivasan" <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>, kexec@...ts.infradead.org, linux-hyperv@...r.kernel.org,
linux-acpi@...r.kernel.org, linux-coco@...ts.linux.dev, linux-kernel@...r.kernel.org,
Tao Liu <ltao@...hat.com>
Subject: Re: [PATCHv11 18/19] x86/acpi: Add support for CPU offlining for
ACPI MADT wakeup method
On Fri, Jun 14, 2024 at 09:06:30AM -0500, Tom Lendacky wrote:
> On 6/13/24 09:56, Borislav Petkov wrote:
> > On Thu, Jun 13, 2024 at 04:41:00PM +0300, Kirill A. Shutemov wrote:
> > > It is easy enough to do. See the patch below.
> >
> > Thanks, will have a look.
> >
> > > But I am not sure if I can justify it properly. If someone doesn't really
> > > need 5-level paging, disabling it at compile-time would save ~34K of
> > > kernel code with the configuration.
> > >
> > > Is it worth saving ~100 lines of code?
> >
> > Well, it goes both ways: is it worth saving ~34K kernel text and for that make
> > the code a lot less conditional, more readable, contain less ugly ifdeffery,
>
> Won't getting rid of the config option cause 5-level to be used by default
> on all platforms that support it? The no5lvl command line option would have
> to be used to get 4-level paging at that point.
Yes, there won't be compile-time option to disable 5-level paging.
Is it a problem?
We benchmarked it back when 5-level paging got introduced and were not able
to see a measurable difference between 4- and 5-level paging on the same
machine. There's some memory overhead on more page table, but it shouldn't
be a show stopper.
I would prefer to get 5-level paging enabled if the machine supports it.
"no5lvl" cmdline option can be useful for debug or if your workload is
somehow special.
--
Kiryl Shutsemau / Kirill A. Shutemov
Powered by blists - more mailing lists