[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20251206194225.GG1219718@nvidia.com>
Date: Sat, 6 Dec 2025 15:42:25 -0400
From: Jason Gunthorpe <jgg@...dia.com>
To: Shuai Xue <xueshuai@...ux.alibaba.com>
Cc: Nicolin Chen <nicolinc@...dia.com>, will@...nel.org,
robin.murphy@....com, joro@...tes.org,
linux-arm-kernel@...ts.infradead.org, iommu@...ts.linux.dev,
linux-kernel@...r.kernel.org, skolothumtho@...dia.com,
praan@...gle.com
Subject: Re: [PATCH rc v1 4/4] iommu/arm-smmu-v3-test: Add nested s1bypass
coverage
On Sat, Dec 06, 2025 at 08:34:09PM +0800, Shuai Xue wrote:
>
> + arm_smmu_test_make_s2_ste(&s2_ste, 0);
> + arm_smmu_v3_test_ste_expect_hitless_transition(test, &s1_ste, &s2_ste,
> + NUM_EXPECTED_SYNCS(3));
>
> With get_ignored(), a nested s1dssbypass STE to a nested s1bypass STE
> will be hitless, a.k.a, NUM_EXPECTED_SYNCS(1).
hitless is tested by the ste_expect_hitless in the function name. The
expected SYNCS have to do with how many updates are required to fix
the STE, 3 is still fine for a hitless update. One to set the unused
bits, one to set the critical qword, one to clear unused bits.
Jason
Powered by blists - more mailing lists