linux-kernel - Re: [PATCH v5 13/14] iommu/amd: Refactor logic to program the host page table in DTE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20251119002015.GJ120075@nvidia.com>
Date: Tue, 18 Nov 2025 20:20:15 -0400
From: Jason Gunthorpe <jgg@...dia.com>
To: Nicolin Chen <nicolinc@...dia.com>
Cc: Suravee Suthikulpanit <suravee.suthikulpanit@....com>,
	linux-kernel@...r.kernel.org, robin.murphy@....com, will@...nel.org,
	joro@...tes.org, kevin.tian@...el.com, jsnitsel@...hat.com,
	vasant.hegde@....com, iommu@...ts.linux.dev, santosh.shukla@....com,
	sairaj.arunkodilkar@....com, jon.grimm@....com,
	prashanthpra@...gle.com, wvw@...gle.com, wnliu@...gle.com,
	gptran@...gle.com, kpsingh@...gle.com, joao.m.martins@...cle.com,
	alejandro.j.jimenez@...cle.com
Subject: Re: [PATCH v5 13/14] iommu/amd: Refactor logic to program the host
 page table in DTE

On Thu, Nov 13, 2025 at 01:19:13PM -0800, Nicolin Chen wrote:

> >  	else {
> > @@ -2097,35 +2115,29 @@ static void set_dte_entry(struct amd_iommu *iommu,
> >  						       &pt_info);
> >  			}
> >  
> > -			new.data[0] |= __sme_set(pt_info.host_pt_root) |
> > -				       (pt_info.mode & DEV_ENTRY_MODE_MASK)
> > -					       << DEV_ENTRY_MODE_SHIFT;
> > +			pt_info.host_pt_root = __sme_set(pt_info.host_pt_root);
> >  		}
> >  	}
> 
> And this __IOMMU_DOMAIN_PAGING path seems to be used by v1 only.
> So, it could be squashed into amd_iommu_set_dte_v1(). This could
> tidy set_dte_entry() further.

Oh yes, definately should be done, the whole pt info grab for v1
belongs inside the set v1 function.

> >  	/*
> >  	 * When SNP is enabled, we can only support TV=1 with non-zero domain ID.
> >  	 * This is prevented by the SNP-enable and IOMMU_DOMAIN_IDENTITY check in
> >  	 * do_iommu_domain_alloc().
> >  	 */
> >  	WARN_ON(amd_iommu_snp_en && (domid == 0));
> 
> This, if it's very necessary, can go into individual functions.
>
> Though I am not sure if it is accurate as it seems to imply that
> IOMMU_DOMAIN_IDENTITY has domid == 0?

I never quite understood this comment myself either :\

> But an IOMMU_DOMAIN_IDENTITY does:
> 	identity_domain.id = amd_iommu_pdom_id_alloc();
> doing:
> 	ida_alloc_range(&pdom_ids, 1, MAX_DOMAIN_ID - 1, GFP_ATOMIC);
> which means it never returns 0.

Hmm!

Jason