linux-kernel - Re: [PATCH 1/3] iommu/io-pgtable-arm: Add nents_per_pgtable in struct io_pgtable

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: <ZPIsJl832r68bUaU@Asurada-Nvidia>
Date:   Fri, 1 Sep 2023 11:23:34 -0700
From:   Nicolin Chen <nicolinc@...dia.com>
To:     Robin Murphy <robin.murphy@....com>
CC:     Will Deacon <will@...nel.org>, <jgg@...dia.com>, <joro@...tes.org>,
        <jean-philippe@...aro.org>, <apopple@...dia.com>,
        <linux-kernel@...r.kernel.org>,
        <linux-arm-kernel@...ts.infradead.org>, <iommu@...ts.linux.dev>
Subject: Re: [PATCH 1/3] iommu/io-pgtable-arm: Add nents_per_pgtable in
 struct io_pgtable_cfg

On Fri, Sep 01, 2023 at 07:02:19PM +0100, Robin Murphy wrote:

> > Our test team encountered a soft lockup in this path today:
> > --------------------------------------------------------------------
> > watchdog: BUG: soft lockup - CPU#244 stuck for 26s!
> 
> That's a lot of TLBIs!

Well, imagining the use cases of NVIDIA Grace, I'd expect it soon
to be no longer a surprise that we see more stressful cases :-/

> > I think it is the same problem that we fixed in tlbflush.h using
> > MAX_TLBI_OPS. So, I plan to send a cleaner bug fix (cc stable)

> > What do you think about it?
> 
> Looks reasonable to me - I think it's the right shape to foreshadow the
> bigger refactoring we discussed, 

Cool! Thanks for the quick ack. And I could move the MAX_TLBI_OPS
to the SMMU header so a later change in smmu.c can just use it.

> and I can't object to using
> PAGE_{SIZE,SHIFT} for the calculation when it's specifically in the
> context of SVA.

Yea, the current SVA code passes in PAGE_SIZE as the TLBI granule.

Thanks
Nic