linux-kernel - Re: [PATCH 3/4] ntb_perf: pass correct struct device to dma_alloc

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: <761e76e0-2e5e-6c71-3384-1ec10dcf8e88@deltatee.com>
Date:   Wed, 5 Feb 2020 11:44:05 -0700
From:   Logan Gunthorpe <logang@...tatee.com>
To:     Arindam Nath <arindam.nath@....com>, Jon Mason <jdmason@...zu.us>,
        Dave Jiang <dave.jiang@...el.com>,
        Allen Hubbe <allenbh@...il.com>,
        Sanjay R Mehta <sanju.mehta@....com>
Cc:     linux-ntb@...glegroups.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/4] ntb_perf: pass correct struct device to
 dma_alloc_coherent



On 2020-02-05 10:16 a.m., Arindam Nath wrote:
> From: Sanjay R Mehta <sanju.mehta@....com>
> 
> Currently, ntb->dev is passed to dma_alloc_coherent
> and dma_free_coherent calls. The returned dma_addr_t
> is the CPU physical address. This works fine as long
> as IOMMU is disabled. But when IOMMU is enabled, we
> need to make sure that IOVA is returned for dma_addr_t.
> So the correct way to achieve this is by changing the
> first parameter of dma_alloc_coherent() as ntb->pdev->dev
> instead.
> 
> Fixes: 5648e56 ("NTB: ntb_perf: Add full multi-port NTB API support")
> Signed-off-by: Sanjay R Mehta <sanju.mehta@....com>
> Signed-off-by: Arindam Nath <arindam.nath@....com>

Ugh, this has been fixed repeatedly and independently by a number of
people. I sent a fix[1] in more than a year ago and Sanjay repeated the
effort a couple months ago[2].

I have the same feed back for you that I had for him: once we fix the
bug we should also go in and remove the now unnecessary
dma_coerce_mask_and_coherent() calls in the drivers at the same time
seeing it no longer makes any sense. My patch did this already.

Thanks,

Logan

[1] https://lore.kernel.org/lkml/20190109192233.5752-3-logang@deltatee.com/
[2]
https://lore.kernel.org/lkml/1575983255-70377-1-git-send-email-Sanju.Mehta@amd.com/