From: Aneesh Kumar K.V <aneesh.kumar@kernel.org>
To: Jason Gunthorpe <jgg@ziepe.ca>
Cc: iommu@lists.linux.dev, linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, linux-coco@lists.linux.dev,
Robin Murphy <robin.murphy@arm.com>,
Marek Szyprowski <m.szyprowski@samsung.com>,
Will Deacon <will@kernel.org>, Marc Zyngier <maz@kernel.org>,
Steven Price <steven.price@arm.com>,
Suzuki K Poulose <Suzuki.Poulose@arm.com>,
Catalin Marinas <catalin.marinas@arm.com>,
Jiri Pirko <jiri@resnulli.us>,
Mostafa Saleh <smostafa@google.com>,
Petr Tesarik <ptesarik@suse.com>,
Alexey Kardashevskiy <aik@amd.com>,
Dan Williams <dan.j.williams@intel.com>,
Xu Yilun <yilun.xu@linux.intel.com>,
linuxppc-dev@lists.ozlabs.org, linux-s390@vger.kernel.org,
Madhavan Srinivasan <maddy@linux.ibm.com>,
Michael Ellerman <mpe@ellerman.id.au>,
Nicholas Piggin <npiggin@gmail.com>,
"Christophe Leroy (CS GROUP)" <chleroy@kernel.org>,
Alexander Gordeev <agordeev@linux.ibm.com>,
Gerald Schaefer <gerald.schaefer@linux.ibm.com>,
Heiko Carstens <hca@linux.ibm.com>,
Vasily Gorbik <gor@linux.ibm.com>,
Christian Borntraeger <borntraeger@linux.ibm.com>,
Sven Schnelle <svens@linux.ibm.com>,
x86@kernel.org, Jiri Pirko <jiri@nvidia.com>,
Michael Kelley <mhklinux@outlook.com>
Subject: Re: [PATCH v6 04/20] dma-pool: track decrypted atomic pools and select them via attrs
Date: Thu, 11 Jun 2026 10:21:50 +0530 [thread overview]
Message-ID: <yq5acxxxn0gp.fsf@kernel.org> (raw)
In-Reply-To: <20260610164153.GQ2764304@ziepe.ca>
Jason Gunthorpe <jgg@ziepe.ca> writes:
> On Wed, Jun 10, 2026 at 01:37:26PM +0530, Aneesh Kumar K.V wrote:
>> Jason Gunthorpe <jgg@ziepe.ca> writes:
>>
>> > On Thu, Jun 04, 2026 at 02:09:43PM +0530, Aneesh Kumar K.V (Arm) wrote:
>> >> struct page *dma_alloc_from_pool(struct device *dev, size_t size,
>> >> - void **cpu_addr, gfp_t gfp,
>> >> + void **cpu_addr, gfp_t gfp, unsigned long attrs,
>> >> bool (*phys_addr_ok)(struct device *, phys_addr_t, size_t))
>> >> {
>> >> - struct gen_pool *pool = NULL;
>> >> + struct dma_gen_pool *dma_pool = NULL;
>> >> struct page *page;
>> >> bool pool_found = false;
>> >>
>> >> - while ((pool = dma_guess_pool(pool, gfp))) {
>> >> + while ((dma_pool = dma_guess_pool(dma_pool, gfp))) {
>> >> +
>> >> + if (dma_pool->unencrypted != !!(attrs & DMA_ATTR_CC_SHARED))
>> >> + continue;
>> >
>> > I don't think you should be overloading DMA_ATTR_CC_SHARED like this.
>> >
>> > /*
>> > * DMA_ATTR_CC_SHARED is not a caller-visible dma_alloc_*()
>> > * attribute. The direct allocator uses it internally after it has
>> > * decided that the backing pages must be shared/decrypted, so the
>> > * rest of the allocation path can consistently select DMA addresses,
>> > * choose compatible pools and restore encryption on free.
>> > */
>> > if (attrs & DMA_ATTR_CC_SHARED)
>> > return NULL;
>> >
>> > if (force_dma_unencrypted(dev)) {
>> > attrs |= DMA_ATTR_CC_SHARED;
>> > mark_mem_decrypt = true;
>> > }
>> >
>> > It is fine to have a bit inside the attrs that is only used by the
>> > internal logic, but it needs to have a clearer name
>> > __DMA_ATTR_REQUIRE_CC_SHARED perhaps.
>> >
>>
>> Are you suggesting adding another attribute in addition to
>> DMA_ATTR_CC_SHARED?
>>
>> Is the idea that __DMA_ATTR_REQUIRE_CC_SHARED would be used in the
>> allocation path to request a CC_SHARED allocation, while
>> DMA_ATTR_CC_SHARED would be used in the mapping path to describe the
>> attribute of the address?
>
> Yeah, it is a thought at least
>
> Maybe a comment is good enough.
>
> I just find it hard to follow when we have this dual usage. Like the
> code above for dma_pool->unencrypted is completely wrong if it is an
> "attribute of an address". Easy to cut & paste that into the wrong
> context.
>
> Especially if you move things up higher.. having the alloc set both
> CC_SHARED and REQUIRE_CC_SHARED or maybe ALLOC_CC_SHARED would make it
> clearer that the alloc code lives under that callchain
>
> Jason
>
If we are adding DMA_ATTR_ALLOC_SHARED, should we also allow
dma_alloc_attrs() to take that attribute value?
Does this look okay?
(Note: Parts of the documentation text were updated using Codex.)
modified Documentation/core-api/dma-attributes.rst
@@ -179,3 +179,32 @@ interface when building their uAPIs, when possible.
It must never be used in an in-kernel driver that only works with
kernel memory.
+
+DMA_ATTR_CC_SHARED
+------------------
+
+This attribute indicates that a DMA mapping is shared, or decrypted, for
+confidential computing guests. For normal system memory, the caller must
+already have marked the memory decrypted with set_memory_decrypted(). CPU
+PTEs for the mapping must use pgprot_decrypted(), and the same shared
+semantic may be passed to a vIOMMU when it sets up the IOPTE.
+
+This attribute describes an existing mapping. It does not allocate shared
+backing pages and must not be passed to dma_alloc_attrs(). For MMIO, use
+this together with DMA_ATTR_MMIO to indicate shared MMIO. Unless
+DMA_ATTR_MMIO is provided, the mapping requires a struct page.
+
+DMA_ATTR_ALLOC_CC_SHARED
+------------------------
+
+This attribute indicates that a dma_alloc_attrs() allocation must use
+shared, or decrypted, backing pages for confidential computing guests.
+Allocation paths use this request when they select shared DMA pools,
+decrypt newly allocated pages or restore encryption on free.
+
+DMA_ATTR_ALLOC_CC_SHARED differs from DMA_ATTR_CC_SHARED in that it
+requests shared backing memory from the allocation path. DMA_ATTR_CC_SHARED
+describes an already-shared mapping and requires the caller to have
+prepared normal system memory before mapping it. Callers that need shared
+memory from dma_alloc_attrs() should request DMA_ATTR_ALLOC_CC_SHARED
+instead of DMA_ATTR_CC_SHARED.
modified include/linux/dma-mapping.h
@@ -103,6 +103,13 @@
*/
#define DMA_ATTR_CC_SHARED (1UL << 13)
+/*
+ * DMA_ATTR_ALLOC_CC_SHARED: Allocates DMA memory as shared (decrypted) for
+ * confidential computing guests. Unlike DMA_ATTR_CC_SHARED, this attribute
+ * is used by dma_alloc_attrs() paths that create shared backing pages;
+ * DMA_ATTR_CC_SHARED describes an already-shared mapping.
+ */
+#define DMA_ATTR_ALLOC_CC_SHARED (1UL << 14)
/*
* A dma_addr_t can hold any valid DMA or bus address for the platform. It can
* be given to a device to use as a DMA source or target. It is specific to a
next prev parent reply other threads:[~2026-06-11 4:52 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-04 8:39 [PATCH v6 00/20] dma-mapping: Use DMA_ATTR_CC_SHARED through direct, pool and swiotlb paths Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 01/20] s390: Expose protected virtualization through cc_platform_has() Aneesh Kumar K.V (Arm)
2026-06-06 0:34 ` JAEHOON KIM
2026-06-09 13:44 ` Catalin Marinas
2026-06-04 8:39 ` [PATCH v6 02/20] dma-direct: swiotlb: handle swiotlb alloc/free outside __dma_direct_alloc_pages Aneesh Kumar K.V (Arm)
2026-06-09 12:15 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 03/20] dma-direct: use DMA_ATTR_CC_SHARED in alloc/free paths Aneesh Kumar K.V (Arm)
2026-06-09 12:18 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 04/20] dma-pool: track decrypted atomic pools and select them via attrs Aneesh Kumar K.V (Arm)
2026-06-09 12:23 ` Petr Tesarik
2026-06-09 14:32 ` Jason Gunthorpe
2026-06-10 8:07 ` Aneesh Kumar K.V
2026-06-10 16:41 ` Jason Gunthorpe
2026-06-11 4:51 ` Aneesh Kumar K.V [this message]
2026-06-11 11:30 ` Jason Gunthorpe
2026-06-11 5:25 ` Aneesh Kumar K.V
2026-06-11 11:37 ` Jason Gunthorpe
2026-06-11 11:50 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 05/20] dma: swiotlb: pass mapping attributes by reference Aneesh Kumar K.V (Arm)
2026-06-09 12:21 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 06/20] dma: swiotlb: track pool encryption state and honor DMA_ATTR_CC_SHARED Aneesh Kumar K.V (Arm)
2026-06-09 12:48 ` Petr Tesarik
2026-06-10 8:46 ` Aneesh Kumar K.V
2026-06-04 8:39 ` [PATCH v6 07/20] dma-mapping: make dma_pgprot() " Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 08/20] dma-direct: pass attrs to dma_capable() for DMA_ATTR_CC_SHARED checks Aneesh Kumar K.V (Arm)
2026-06-09 12:50 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 09/20] dma-direct: make dma_direct_map_phys() honor DMA_ATTR_CC_SHARED Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 10/20] dma-direct: set decrypted flag for remapped DMA allocations Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 11/20] dma-direct: select DMA address encoding from DMA_ATTR_CC_SHARED Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 12/20] dma-pool: fix page leak in atomic_pool_expand() cleanup Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 13/20] dma-direct: rename ret to cpu_addr in alloc helpers Aneesh Kumar K.V (Arm)
2026-06-09 12:54 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 14/20] dma-direct: return struct page from dma_direct_alloc_from_pool() Aneesh Kumar K.V (Arm)
2026-06-09 13:12 ` Petr Tesarik
2026-06-09 13:45 ` Catalin Marinas
2026-06-09 14:15 ` Jason Gunthorpe
2026-06-04 8:39 ` [PATCH v6 15/20] iommu/dma: Check atomic pool allocation result directly Aneesh Kumar K.V (Arm)
2026-06-09 13:13 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 16/20] dma: swiotlb: free dynamic pools from process context Aneesh Kumar K.V (Arm)
2026-06-09 13:23 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 17/20] dma: swiotlb: handle set_memory_decrypted() failures Aneesh Kumar K.V (Arm)
2026-06-09 13:32 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 18/20] dma: free atomic pool pages by physical address Aneesh Kumar K.V (Arm)
2026-06-04 8:39 ` [PATCH v6 19/20] swiotlb: Preserve allocation virtual address for dynamic pools Aneesh Kumar K.V (Arm)
2026-06-09 13:40 ` Petr Tesarik
2026-06-04 8:39 ` [PATCH v6 20/20] swiotlb: remove unused SWIOTLB_FORCE flag Aneesh Kumar K.V (Arm)
2026-06-09 13:44 ` Petr Tesarik
2026-06-09 13:43 ` [PATCH v6 00/20] dma-mapping: Use DMA_ATTR_CC_SHARED through direct, pool and swiotlb paths Catalin Marinas
2026-06-09 14:47 ` Jason Gunthorpe
2026-06-11 5:52 ` Aneesh Kumar K.V
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=yq5acxxxn0gp.fsf@kernel.org \
--to=aneesh.kumar@kernel.org \
--cc=Suzuki.Poulose@arm.com \
--cc=agordeev@linux.ibm.com \
--cc=aik@amd.com \
--cc=borntraeger@linux.ibm.com \
--cc=catalin.marinas@arm.com \
--cc=chleroy@kernel.org \
--cc=dan.j.williams@intel.com \
--cc=gerald.schaefer@linux.ibm.com \
--cc=gor@linux.ibm.com \
--cc=hca@linux.ibm.com \
--cc=iommu@lists.linux.dev \
--cc=jgg@ziepe.ca \
--cc=jiri@nvidia.com \
--cc=jiri@resnulli.us \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-coco@lists.linux.dev \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-s390@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=m.szyprowski@samsung.com \
--cc=maddy@linux.ibm.com \
--cc=maz@kernel.org \
--cc=mhklinux@outlook.com \
--cc=mpe@ellerman.id.au \
--cc=npiggin@gmail.com \
--cc=ptesarik@suse.com \
--cc=robin.murphy@arm.com \
--cc=smostafa@google.com \
--cc=steven.price@arm.com \
--cc=svens@linux.ibm.com \
--cc=will@kernel.org \
--cc=x86@kernel.org \
--cc=yilun.xu@linux.intel.com \
/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