From: Dmitry Osipenko <digetx@gmail.com>
To: Nicolin Chen <nicoleotsuka@gmail.com>
Cc: joro@8bytes.org, thierry.reding@gmail.com, will@kernel.org,
guillaume.tucker@collabora.com, vdumpa@nvidia.com,
jonathanh@nvidia.com, linux-tegra@vger.kernel.org,
iommu@lists.linux-foundation.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] iommu/tegra-smmu: Fix mc errors on tegra124-nyan
Date: Tue, 23 Feb 2021 08:10:41 +0300 [thread overview]
Message-ID: <79bb1248-497f-8adf-663b-74448bea3849@gmail.com> (raw)
In-Reply-To: <20210223021343.GA6539@Asurada-Nvidia>
23.02.2021 05:13, Nicolin Chen пишет:
> Hi Dmitry,
>
> On Sat, Feb 20, 2021 at 08:16:22AM +0300, Dmitry Osipenko wrote:
>> 19.02.2021 01:07, Nicolin Chen пишет:
>>> Commit 25938c73cd79 ("iommu/tegra-smmu: Rework tegra_smmu_probe_device()")
>>> removed certain hack in the tegra_smmu_probe() by relying on IOMMU core to
>>> of_xlate SMMU's SID per device, so as to get rid of tegra_smmu_find() and
>>> tegra_smmu_configure() that are typically done in the IOMMU core also.
>>>
>>> This approach works for both existing devices that have DT nodes and other
>>> devices (like PCI device) that don't exist in DT, on Tegra210 and Tegra3
>>> upon testing. However, Page Fault errors are reported on tegra124-Nyan:
>>>
>>> tegra-mc 70019000.memory-controller: display0a: read @0xfe056b40:
>>> EMEM address decode error (SMMU translation error [--S])
>>> tegra-mc 70019000.memory-controller: display0a: read @0xfe056b40:
>>> Page fault (SMMU translation error [--S])
>>>
>>> After debugging, I found that the mentioned commit changed some function
>>> callback sequence of tegra-smmu's, resulting in enabling SMMU for display
>>> client before display driver gets initialized. I couldn't reproduce exact
>>> same issue on Tegra210 as Tegra124 (arm-32) differs at arch-level code.
>>
>> Hello Nicolin,
>>
>> Could you please explain in a more details what exactly makes the
>> difference for the callback sequence?
>
> Here is a log with 5.11.0-rc6:
> https://lava.collabora.co.uk/scheduler/job/3187849
> [dump_stack was added in some tegra-smmu functions]
>
> And here is a corresponding log with reverting the original commit:
> https://lava.collabora.co.uk/scheduler/job/3187851
>
> Here is a log with 5.11.0-rc7-next-20210210:
> https://lava.collabora.co.uk/scheduler/job/3210245
>
> And here is a corresponding log with reverting the original commit:
> https://lava.collabora.co.uk/scheduler/job/3210596
>
> Both failing logs show that mc errors started right after client DC
> got enabled by ->attach_dev() callback that in the passing logs was
> not called until Host1x driver init. And note that two failing logs
> show that ->attach_dev() could be called from two different sources,
> of_dma_configure_id() or arch_setup_dma_ops().
>
> The reason why ->attach_dev() gets called is probably related to the
> following reasons (sorry, can't be 100% sure as I don't have Tegra124
> or other 32bit Tegra board to test):
> 1) With the commit reverted, all clients are probed in "arch" stage,
> which is even prior to iommu core initialization -- including it
> setting default domain type. This probably messed up the type of
> allocating domains against the default domain type. Also internal
> group is somehow affected. So some condition check in iommu core
> failed and then it bypassed ->attach_dev callback in really_probe
> stage, until Host1x driver does attach_dev again.
>
> 2) 32bit ARM has arch_setup_dma_ops() does an additional set of iommu
> domain allocation + attach_dev(), after of_dma_configure_id() did
> once. This isn't reproducible for me on Tegra210.
>
> As debugging online isn't very efficient, and given that Thierry has
> been working on the linear mapping of framebuffer carveout, I choose
> to partially revert as a quick fix.
The partially revert should be okay, but it's not clear to me what makes
difference for T124 since I don't see that problem on T30, which also
has active display at a boot time.
next prev parent reply other threads:[~2021-02-23 5:11 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-18 22:07 [PATCH] iommu/tegra-smmu: Fix mc errors on tegra124-nyan Nicolin Chen
2021-02-20 5:16 ` Dmitry Osipenko
2021-02-23 2:13 ` Nicolin Chen
2021-02-23 5:10 ` Dmitry Osipenko [this message]
2021-02-25 6:27 ` Nicolin Chen
2021-02-27 9:59 ` Dmitry Osipenko
2021-03-02 23:08 ` Nicolin Chen
2021-03-03 9:47 ` Dmitry Osipenko
2021-03-28 15:25 ` Dmitry Osipenko
2021-03-28 15:36 ` Dmitry Osipenko
2021-02-22 13:15 ` Guillaume Tucker
2021-03-03 18:26 ` Thierry Reding
2021-03-04 12:21 ` Joerg Roedel
2021-03-10 19:13 ` Dmitry Osipenko
2021-03-10 20:22 ` Dmitry Osipenko
2021-03-10 22:17 ` Nicolin Chen
2021-03-11 12:06 ` Dmitry Osipenko
2021-03-12 1:09 ` Nicolin Chen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=79bb1248-497f-8adf-663b-74448bea3849@gmail.com \
--to=digetx@gmail.com \
--cc=guillaume.tucker@collabora.com \
--cc=iommu@lists.linux-foundation.org \
--cc=jonathanh@nvidia.com \
--cc=joro@8bytes.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-tegra@vger.kernel.org \
--cc=nicoleotsuka@gmail.com \
--cc=thierry.reding@gmail.com \
--cc=vdumpa@nvidia.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).