linux-kernel - Re: [PATCH v9 6/7] iommu/arm-smmu-v3: Add arm_smmu_invs based arm_smmu_domain_inv

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] [day] [month] [year] [list]

Message-ID: <20260123135600.GX1134360@nvidia.com>
Date: Fri, 23 Jan 2026 09:56:00 -0400
From: Jason Gunthorpe <jgg@...dia.com>
To: Pranjal Shrivastava <praan@...gle.com>
Cc: Nicolin Chen <nicolinc@...dia.com>, will@...nel.org,
	jean-philippe@...aro.org, robin.murphy@....com, joro@...tes.org,
	balbirs@...dia.com, miko.lenczewski@....com, peterz@...radead.org,
	kevin.tian@...el.com, linux-arm-kernel@...ts.infradead.org,
	iommu@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v9 6/7] iommu/arm-smmu-v3: Add arm_smmu_invs based
 arm_smmu_domain_inv_range()

On Fri, Jan 23, 2026 at 09:48:37AM +0000, Pranjal Shrivastava wrote:

> For INV_TYPE_S1_ASID, the new code loops and checks size_too_big 
> via arm_smmu_inv_to_cmdq_batch. However, for INV_TYPE_ATS, it issues
> a single command for the entire range. While this matches the current
> driver, are we confident arm_smmu_atc_inv_to_cmd handles all massive
> sizes correctly without needing a similar loop or "too big" fallback?

At the PCI level ATS invalidations accept an arbitary byte range, so
all ranges should be converted into a single CMDQ command and pushed
as a single TLP.

The issue the "too big" stuff is solving is for the IOTLB itself which
can require an invalidation rage to be exploded into many smaller
commands.

Jason