From: Robin Murphy <robin.murphy@arm.com>
To: John Garry <john.g.garry@oracle.com>, joro@8bytes.org
Cc: will@kernel.org, iommu@lists.linux.dev,
linux-kernel@vger.kernel.org,
Linus Torvalds <torvalds@linux-foundation.org>,
kuba@kernel.org
Subject: Re: [PATCH v3] iommu: Optimise PCI SAC address trick
Date: Thu, 13 Apr 2023 11:44:56 +0100 [thread overview]
Message-ID: <08c9f2a2-b2fc-2574-e10a-b6dbd0caedb0@arm.com> (raw)
In-Reply-To: <02eebcda-f60e-f8ed-7057-cf293d15a173@oracle.com>
On 2023-04-13 11:16, John Garry wrote:
> On 20/01/2023 11:33, John Garry wrote:
>> On 18/01/2023 17:26, Robin Murphy wrote:
>>> Per the reasoning in commit 4bf7fda4dce2 ("iommu/dma: Add config for
>>> PCI SAC address trick") and its subsequent revert, this mechanism no
>>> longer serves its original purpose, but now only works around broken
>>> hardware/drivers in a way that is unfortunately too impactful to remove.
>>>
>>> This does not, however prevent us from solving the performance impact
>>> which the workaround imposes on large-scale systems that don't need it.
>>> That is felt once the 32-bit IOVA space fills up and we keep
>>> unsuccessfully trying to allocate from it. However, if we get to that
>>> point then in fact it's already the endgame. The nature of the allocator
>>> is such that the first IOVA we give to a device after the 32-bit space
>>> runs out will be the highest possible address for that device, ever.
>>> If that works, then great, we can be pretty sure it's safe to optimise
>>> for speed by always allocating from the full range. And if it doesn't,
>>> then the worst has already happened and any brokenness is now showing,
>>> so there's no point continuing to try to hide it.
>>>
>>> To that end, implement a flag to refine this into a per-device policy
>>> that can automatically get itself out of the way if and when it stops
>>> being useful.
>>>
>>> CC: John Garry <john.garry@huawei.com>
>>> CC: Linus Torvalds <torvalds@linux-foundation.org>
>>> Signed-off-by: Robin Murphy <robin.murphy@arm.com>
>>> ---
>>>
>>> v3: Expand the flag name, add a print with inline commentary for good
>>> measure, and refactor the code flow even more (too many ifs and
>>> indents...) such that I didn't presume to carry forward John's R-b.
>>
>> I like the new changes, so feel free to add:
>>
>> Reviewed-by: John Garry <john.g.garry@oracle.com>
>
> Is there any chance that this can be picked up?
>
> I also saw that it fixed an issue for Jakub (cc'ed) recently.
Oh, thanks for the reminder - IIRC this wants a minor rebase now, let me
double-check and send a v4 that applies cleanly...
Cheers,
Robin.
prev parent reply other threads:[~2023-04-13 10:45 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-18 17:26 [PATCH v3] iommu: Optimise PCI SAC address trick Robin Murphy
2023-01-20 11:33 ` John Garry
2023-04-13 10:16 ` John Garry
2023-04-13 10:44 ` Robin Murphy [this message]
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=08c9f2a2-b2fc-2574-e10a-b6dbd0caedb0@arm.com \
--to=robin.murphy@arm.com \
--cc=iommu@lists.linux.dev \
--cc=john.g.garry@oracle.com \
--cc=joro@8bytes.org \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=torvalds@linux-foundation.org \
--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 read-only IMAP folder(s) and NNTP newsgroup(s).