Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 22 Mar 2023 09:50:03 -0300
From:   Jason Gunthorpe <jgg@ziepe.ca>
To:     Steven Price <steven.price@arm.com>
Cc:     Heiko Stuebner <heiko@sntech.de>, Joerg Roedel <joro@8bytes.org>,
        Will Deacon <will@kernel.org>,
        Robin Murphy <robin.murphy@arm.com>, iommu@lists.linux.dev,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        linux-rockchip@lists.infradead.org,
        Lu Baolu <baolu.lu@linux.intel.com>
Subject: Re: [PATCH] iommu/rockchip: Add missing set_platform_dma_ops callback
Message-ID: <ZBr5e6tn1i7EE/16@ziepe.ca>
References: <20230315164152.333251-1-steven.price@arm.com>
 <ZBnBU9OU4iV6CV0W@ziepe.ca>
 <85607806-b888-2d5e-67a4-e9d63ebd1976@arm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <85607806-b888-2d5e-67a4-e9d63ebd1976@arm.com>
Precedence: bulk

On Wed, Mar 22, 2023 at 09:02:41AM +0000, Steven Price wrote:
> On 21/03/2023 14:38, Jason Gunthorpe wrote:
> > On Wed, Mar 15, 2023 at 04:41:52PM +0000, Steven Price wrote:
> >> Similar to exynos, we need a set_platform_dma_ops() callback for proper
> >> operation on ARM 32 bit after recent changes in the IOMMU framework
> >> (detach ops removal).
> >>
> >> Fixes: c1fe9119ee70 ("iommu: Add set_platform_dma_ops callbacks")
> >> Signed-off-by: Steven Price <steven.price@arm.com>
> >> ---
> >> This fixes a splat I was seeing on a Firefly-RK3288, more details here:
> >> https://lore.kernel.org/all/26a5d1b8-40b3-b1e4-bc85-740409c26838@arm.com/
> > 
> > Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
> 
> Thanks for the review.
> 
> > Do you know what state the iommu is left in after
> > rk_iommu_detach_device()? Ie is it blocking DMA or doing identity or
> > something else?
> 
> To be honest I really don't know for sure. But from my small
> understanding of the code: rk_iommu_detach_device() ends up in
> rk_iommu_disable_paging() which appears to switch to identity mode
> ("Disable memory translation").

Can you consider writing this patch like this instead:

https://lore.kernel.org/linux-iommu/ZBnef7g7GCxogPNz@ziepe.ca/

?

I'd much rather we move toward clearly documenting what is going on
with the HW and remove this undefined "detach" language.

Jason