Linux s390 Architecture development
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Aneesh Kumar K.V (Arm)" <aneesh.kumar@kernel.org>
Cc: linux-s390@vger.kernel.org, "Vasily Gorbik" <gor@linux.ibm.com>,
	"Alexander Gordeev" <agordeev@linux.ibm.com>,
	"Christian Borntraeger" <borntraeger@linux.ibm.com>,
	"Heiko Carstens" <hca@linux.ibm.com>
Subject: Re: [PATCH v3 3/4] dma: swiotlb: Centralize memory-encryption pool sizing
Date: Fri, 04 Sep 2026 10:36:44 +0000	[thread overview]
Message-ID: <20260904103644.CD15E1F00A3D@smtp.kernel.org> (raw)
In-Reply-To: <20260904101810.1193505-4-aneesh.kumar@kernel.org>

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Skipping SWIOTLB_POOL_CC_GUEST when a restricted DMA pool is present can result in SWIOTLB_POOL_NONE or SWIOTLB_POOL_MINIMAL being selected, breaking DMA for all other devices.
--

commit d8ab20a637a8736036e923dfa3f89a8da55215c3
Author: Aneesh Kumar K.V (Arm) <aneesh.kumar@kernel.org>

dma: swiotlb: Centralize memory-encryption pool sizing

This patch centralizes SWIOTLB memory-encryption pool sizing for guests into
the SWIOTLB core. It sizes the confidential guest pool to 6% of guest memory
by default and records the presence of restricted DMA pools to avoid selecting
the confidential-guest policy solely based on guest memory encryption.

> diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
> index 90e0bfd0715cf..c4611b2c45403 100644
> --- a/kernel/dma/swiotlb.c
> +++ b/kernel/dma/swiotlb.c

[ ... ]

> @@ -509,7 +553,8 @@ swiotlb_select_pool_policy(unsigned int flags)
>  	if (swiotlb_force_disable)
>  		return SWIOTLB_POOL_NONE;
>  
> -	if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT))
> +	if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT) &&
> +	    !restricted_dma_pool_present)
>  		return SWIOTLB_POOL_CC_GUEST;

[Severity: High]
Does skipping SWIOTLB_POOL_CC_GUEST here leave the system without a proper
fallback for non-restricted devices?

Looking at swiotlb_select_pool_policy() in kernel/dma/swiotlb.c, if a
confidential guest boots with a restricted DMA pool present in the Device
Tree, and system RAM is small enough that SWIOTLB_INIT_ADDRESSING_LIMIT is
not set (e.g., <= 4GB), this check will skip SWIOTLB_POOL_CC_GUEST. 

The function then falls through all the way to SWIOTLB_POOL_MINIMAL (a tiny
1MB/GB pool) or SWIOTLB_POOL_NONE. If it returns SWIOTLB_POOL_NONE,
swiotlb_init_remap() will exit without allocating a default SWIOTLB pool.
As a result, unassigned devices that still rely on the default pool for
bounce buffers will fail to map DMA memory, potentially leading to I/O
errors or boot panics.

Should this path safely fall back to SWIOTLB_POOL_DEFAULT when a restricted
DMA pool is present, to ensure a standard default pool is still allocated?

>  
>  	if (cc_platform_has(CC_ATTR_HOST_MEM_ENCRYPT))
>  		return SWIOTLB_POOL_DEFAULT;

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260904101810.1193505-1-aneesh.kumar@kernel.org?part=3

  reply	other threads:[~2026-09-04 10:36 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-04 10:18 [PATCH v3 0/4] dma: swiotlb: Centralize default pool policy and sizing Aneesh Kumar K.V (Arm)
2026-09-04 10:18 ` [PATCH v3 1/4] dma: swiotlb: Centralize default pool policy selection Aneesh Kumar K.V (Arm)
2026-09-04 10:40   ` sashiko-bot
2026-09-04 10:18 ` [PATCH v3 2/4] dma: swiotlb: Centralize minimal pool sizing Aneesh Kumar K.V (Arm)
2026-09-04 10:31   ` sashiko-bot
2026-09-04 10:18 ` [PATCH v3 3/4] dma: swiotlb: Centralize memory-encryption " Aneesh Kumar K.V (Arm)
2026-09-04 10:36   ` sashiko-bot [this message]
2026-09-04 10:18 ` [PATCH v3 4/4] dma: swiotlb: Remove SWIOTLB_ANY Aneesh Kumar K.V (Arm)
2026-09-04 10:35   ` sashiko-bot

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=20260904103644.CD15E1F00A3D@smtp.kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=agordeev@linux.ibm.com \
    --cc=aneesh.kumar@kernel.org \
    --cc=borntraeger@linux.ibm.com \
    --cc=gor@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    --cc=linux-s390@vger.kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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