From: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
To: "kvm@vger.kernel.org" <kvm@vger.kernel.org>,
"linux-coco@lists.linux.dev" <linux-coco@lists.linux.dev>,
"Huang, Kai" <kai.huang@intel.com>,
"Hansen, Dave" <dave.hansen@intel.com>,
"Zhao, Yan Y" <yan.y.zhao@intel.com>,
"kas@kernel.org" <kas@kernel.org>,
"seanjc@google.com" <seanjc@google.com>,
"mingo@redhat.com" <mingo@redhat.com>,
"pbonzini@redhat.com" <pbonzini@redhat.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"nik.borisov@suse.com" <nik.borisov@suse.com>,
"linux-doc@vger.kernel.org" <linux-doc@vger.kernel.org>,
"hpa@zytor.com" <hpa@zytor.com>,
"Annapurve, Vishal" <vannapurve@google.com>,
"tglx@kernel.org" <tglx@kernel.org>,
"Mehta, Sohil" <sohil.mehta@intel.com>,
"bp@alien8.de" <bp@alien8.de>, "Gao, Chao" <chao.gao@intel.com>,
"x86@kernel.org" <x86@kernel.org>
Cc: "kirill.shutemov@linux.intel.com" <kirill.shutemov@linux.intel.com>
Subject: Re: [PATCH v6 04/11] x86/virt/tdx: Allocate ref counts for Dynamic PAMT memory
Date: Wed, 8 Jul 2026 23:21:29 +0000 [thread overview]
Message-ID: <a3a878fe02496be580443e338855e6aeda65d710.camel@intel.com> (raw)
In-Reply-To: <6ab50a8c-d502-4d51-b037-3af6c4b4083e@intel.com>
On Wed, 2026-07-08 at 13:49 -0700, Sohil Mehta wrote:
> On 5/25/2026 7:35 PM, Rick Edgecombe wrote:
> > From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> >
> > The PAMT memory holds metadata for all possible TDX protected memory. Each
> > physical address range is covered by PAMT entries at three levels (1GB,
> > 2MB, 4KB). With Dynamic PAMT, the 4KB range of PAMT is allocated on
> > demand. The kernel supplies the TDX module with page pairs to store the
> > 4KB entries, which cover 2MB of host physical memory. The kernel must
> > provide this page pair before using pages from the range for TDX. If this
> > is not done, SEAMCALLs that give the pages to be protected by the TDX module
> > will fail.
> >
> > Allocate reference counters for every 2MB range to track TDX memory usage.
> > This can be used to handle concurrent get/put callers, in order to
> > accurately determine when the dynamic 4KB level of Dynamic PAMT needs to
> > be allocated and when it can be freed.
> >
> > This allocation will currently consume 2 MB for every 1 TB of address
> > space from 0 to max_pfn. The allocation size will depend on how the RAM is
> > physically laid out. In a worst case scenario where the entire 52-bit
> > address space is covered this would be 8GB. Then the DPAMT refcount
> > allocations could hypothetically cause the savings from Dynamic PAMT to go
> > negative on exotic platforms with sparse, small amounts of memory.
> >
>
> ...
>
> > +/*
> > + * On a machine with Dynamic PAMT, the kernel maintains a reference counter
> > + * for every 2M range.
>
> Commit log says every 2MB range.
Yea, I'll unify to MB, TB, KB.
>
> The counter indicates how many users there are for
> > + * the PAMT memory of the 2M range. The kernel allocates PAMT refcounts at
> > + * initialization.
> > + */
> > +static atomic_t *pamt_refcounts;
> > +
> > /* All TDX-usable memory regions. Protected by mem_hotplug_lock. */
> > static LIST_HEAD(tdx_memlist);
> >
> > @@ -254,6 +263,43 @@ static struct syscore tdx_syscore = {
> > .ops = &tdx_syscore_ops,
> > };
> >
> > +/*
> > + * Allocate PAMT reference counters for all physical memory.
> > + *
> > + * It consumes 2MiB for every 1TiB of physical memory.
>
> Commit log says 2MB and 1TB. I would make both consistent.
Yea, different authors for some of this text.
>
> > + */
> > +static int init_pamt_refcounts(void)
> > +{
> > + size_t size = DIV_ROUND_UP(max_pfn, PTRS_PER_PTE) * sizeof(*pamt_refcounts);
> > +
> > + if (!tdx_supports_dynamic_pamt(&tdx_sysinfo))
> > + return 0;
> > +
> > + pamt_refcounts = __vmalloc(size, GFP_KERNEL | __GFP_ZERO);
>
> vzalloc()?
Totally.
>
> > + if (!pamt_refcounts)
> > + return -ENOMEM;
> > +
> > + return 0;
> > +}
> > +
> > +static void free_pamt_refcounts(void)
> > +{
> > + if (!tdx_supports_dynamic_pamt(&tdx_sysinfo))
> > + return;
> > +
> > + vfree(pamt_refcounts);
> > + pamt_refcounts = NULL;
> > +}
> > +
> > +/* Find PAMT refcount for a given physical address */
>
> This comment is probably not that useful. The function name is
> descriptive by itself.
I'll delete it.
>
> > +static atomic_t * __maybe_unused tdx_find_pamt_refcount(unsigned long pfn)
> > +{
> > + /* Find which PMD a PFN is in. */
> > + unsigned long index = pfn >> (PMD_SHIFT - PAGE_SHIFT);
> > +
> > + return &pamt_refcounts[index];
> > +}
> > +
next prev parent reply other threads:[~2026-07-08 23:21 UTC|newest]
Thread overview: 94+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-26 2:35 [PATCH v6 00/11] Dynamic PAMT Rick Edgecombe
2026-05-26 2:35 ` [PATCH v6 01/11] x86/virt/tdx: Simplify tdmr_get_pamt_sz() Rick Edgecombe
2026-06-04 16:05 ` Kiryl Shutsemau
2026-07-01 0:08 ` Edgecombe, Rick P
2026-06-11 18:25 ` Vishal Annapurve
2026-07-03 5:48 ` Chao Gao
2026-07-06 20:18 ` Edgecombe, Rick P
2026-07-07 3:24 ` Yan Zhao
2026-07-07 19:08 ` Edgecombe, Rick P
2026-07-07 23:22 ` Sohil Mehta
2026-07-08 1:23 ` Edgecombe, Rick P
2026-05-26 2:35 ` [PATCH v6 02/11] x86/virt/tdx: Allocate page bitmap for Dynamic PAMT Rick Edgecombe
2026-06-04 16:14 ` Kiryl Shutsemau
2026-07-01 0:14 ` Edgecombe, Rick P
2026-06-11 18:47 ` Vishal Annapurve
2026-07-03 8:26 ` Chao Gao
2026-07-07 3:59 ` Yan Zhao
2026-07-07 21:25 ` Edgecombe, Rick P
2026-07-08 0:49 ` Sohil Mehta
2026-07-08 2:07 ` Edgecombe, Rick P
2026-07-08 2:10 ` Edgecombe, Rick P
2026-07-08 3:35 ` Sohil Mehta
2026-07-08 20:50 ` Edgecombe, Rick P
2026-07-08 21:20 ` Sohil Mehta
2026-05-26 2:35 ` [PATCH v6 03/11] x86/virt/tdx: Add tdx_alloc/free_control_page() helpers Rick Edgecombe
2026-06-08 2:11 ` Binbin Wu
2026-06-08 2:18 ` Yan Zhao
2026-07-01 0:15 ` Edgecombe, Rick P
2026-07-06 12:31 ` Chao Gao
2026-07-06 20:23 ` Edgecombe, Rick P
2026-07-08 3:50 ` Sohil Mehta
2026-07-08 22:54 ` Edgecombe, Rick P
2026-05-26 2:35 ` [PATCH v6 04/11] x86/virt/tdx: Allocate ref counts for Dynamic PAMT memory Rick Edgecombe
2026-07-02 7:20 ` Binbin Wu
2026-07-06 20:26 ` Edgecombe, Rick P
2026-07-06 13:22 ` Chao Gao
2026-07-06 20:26 ` Edgecombe, Rick P
2026-07-07 4:53 ` Yan Zhao
2026-07-07 22:32 ` Edgecombe, Rick P
2026-07-08 20:49 ` Sohil Mehta
2026-07-08 23:21 ` Edgecombe, Rick P [this message]
2026-05-26 2:35 ` [PATCH v6 05/11] x86/virt/tdx: Handle concurrent callers in tdx_pamt_get/put() Rick Edgecombe
2026-07-02 7:39 ` Binbin Wu
2026-07-06 20:27 ` Edgecombe, Rick P
2026-07-07 5:54 ` Chao Gao
2026-07-08 6:46 ` Yan Zhao
2026-07-08 19:01 ` Dave Hansen
2026-07-08 21:03 ` Edgecombe, Rick P
2026-07-08 21:11 ` Sohil Mehta
2026-07-08 21:13 ` Dave Hansen
2026-05-26 2:35 ` [PATCH v6 06/11] x86/virt/tdx: Optimize tdx_pamt_get/put() Rick Edgecombe
2026-05-26 8:57 ` Chao Gao
2026-05-26 16:42 ` Edgecombe, Rick P
2026-06-04 16:59 ` Kiryl Shutsemau
2026-06-05 5:40 ` Chao Gao
2026-06-05 11:42 ` Kiryl Shutsemau
2026-06-05 16:23 ` Dave Hansen
2026-06-08 9:14 ` Kiryl Shutsemau
2026-06-08 9:50 ` Yan Zhao
2026-07-01 1:45 ` Edgecombe, Rick P
2026-07-01 5:37 ` Yan Zhao
2026-07-01 1:05 ` Edgecombe, Rick P
2026-07-07 6:45 ` Chao Gao
2026-07-08 8:46 ` Yan Zhao
2026-05-26 2:35 ` [PATCH v6 07/11] KVM: TDX: Allocate PAMT memory for TD and vCPU control structures Rick Edgecombe
2026-07-02 8:55 ` Binbin Wu
2026-07-06 23:47 ` Sean Christopherson
2026-07-06 23:54 ` Edgecombe, Rick P
2026-07-07 0:25 ` Sean Christopherson
2026-07-08 18:30 ` Dave Hansen
2026-07-07 6:54 ` Chao Gao
2026-07-08 9:13 ` Yan Zhao
2026-05-26 2:35 ` [PATCH v6 08/11] x86/tdx: Add APIs to support Dynamic PAMT ops from KVM's fault path Rick Edgecombe
2026-06-04 17:11 ` Kiryl Shutsemau
2026-07-02 9:32 ` Binbin Wu
2026-07-07 7:25 ` Chao Gao
2026-05-26 2:35 ` [PATCH v6 09/11] KVM: TDX: Get/put PAMT pages when (un)mapping private memory Rick Edgecombe
2026-07-03 3:15 ` Binbin Wu
2026-07-06 20:47 ` Edgecombe, Rick P
2026-07-06 21:02 ` Sean Christopherson
2026-07-06 21:52 ` Edgecombe, Rick P
2026-07-08 8:35 ` Chao Gao
2026-05-26 2:35 ` [PATCH v6 10/11] x86/virt/tdx: Enable Dynamic PAMT Rick Edgecombe
2026-06-04 17:14 ` Kiryl Shutsemau
2026-06-05 5:25 ` Chao Gao
2026-07-01 1:20 ` Edgecombe, Rick P
2026-07-06 20:48 ` Edgecombe, Rick P
2026-07-03 4:35 ` Binbin Wu
2026-05-26 2:35 ` [PATCH v6 11/11] Documentation/x86: Add documentation for TDX's " Rick Edgecombe
2026-07-03 4:54 ` Binbin Wu
2026-06-08 5:45 ` [PATCH v6 00/11] " Tony Lindgren
2026-07-06 21:01 ` Edgecombe, Rick P
2026-07-07 16:17 ` Sean Christopherson
2026-07-07 17:25 ` Edgecombe, Rick P
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=a3a878fe02496be580443e338855e6aeda65d710.camel@intel.com \
--to=rick.p.edgecombe@intel.com \
--cc=bp@alien8.de \
--cc=chao.gao@intel.com \
--cc=dave.hansen@intel.com \
--cc=hpa@zytor.com \
--cc=kai.huang@intel.com \
--cc=kas@kernel.org \
--cc=kirill.shutemov@linux.intel.com \
--cc=kvm@vger.kernel.org \
--cc=linux-coco@lists.linux.dev \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=nik.borisov@suse.com \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.com \
--cc=sohil.mehta@intel.com \
--cc=tglx@kernel.org \
--cc=vannapurve@google.com \
--cc=x86@kernel.org \
--cc=yan.y.zhao@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox