| 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
| ||
|
Message-ID: <CAHUa44HEPF3w=b2O-4DcRUBgF4rJUUze8zy48_g_FCQ5Y34U6A@mail.gmail.com>
Date: Wed, 7 May 2025 18:32:08 +0200
From: Jens Wiklander <jens.wiklander@...aro.org>
To: Robin Murphy <robin.murphy@....com>
Cc: linux-kernel@...r.kernel.org, linux-media@...r.kernel.org,
dri-devel@...ts.freedesktop.org, linaro-mm-sig@...ts.linaro.org,
op-tee@...ts.trustedfirmware.org, linux-arm-kernel@...ts.infradead.org,
Olivier Masse <olivier.masse@....com>, Thierry Reding <thierry.reding@...il.com>,
Yong Wu <yong.wu@...iatek.com>, Sumit Semwal <sumit.semwal@...aro.org>,
Benjamin Gaignard <benjamin.gaignard@...labora.com>, Brian Starkey <Brian.Starkey@....com>,
John Stultz <jstultz@...gle.com>, "T . J . Mercier" <tjmercier@...gle.com>,
Christian König <christian.koenig@....com>,
Sumit Garg <sumit.garg@...nel.org>, Matthias Brugger <matthias.bgg@...il.com>,
AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>, azarrabi@....qualcomm.com,
Simona Vetter <simona.vetter@...ll.ch>, Daniel Stone <daniel@...ishbar.org>,
Rouven Czerwinski <rouven.czerwinski@...aro.org>
Subject: Re: [PATCH v8 11/14] tee: add tee_shm_alloc_cma_phys_mem()
Hi,
On Fri, May 2, 2025 at 5:11 PM Robin Murphy <robin.murphy@....com> wrote:
>
> On 02/05/2025 10:59 am, Jens Wiklander wrote:
> > Add tee_shm_alloc_cma_phys_mem() to allocate a physical memory using
> > from the default CMA pool. The memory is represented by a tee_shm object
> > using the new flag TEE_SHM_CMA_BUF to identify it as physical memory
> > from CMA.
>
> If and when it's possible to dynamically delegate any old kernel memory
> to the TEE, it's far from clear why that should involve poking around in
> CMA internals...
>
> > Signed-off-by: Jens Wiklander <jens.wiklander@...aro.org>
> > ---
> > drivers/tee/tee_shm.c | 55 ++++++++++++++++++++++++++++++++++++++--
> > include/linux/tee_core.h | 4 +++
> > 2 files changed, 57 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/tee/tee_shm.c b/drivers/tee/tee_shm.c
> > index e1ed52ee0a16..faaa0a87bb18 100644
> > --- a/drivers/tee/tee_shm.c
> > +++ b/drivers/tee/tee_shm.c
> > @@ -3,8 +3,11 @@
> > * Copyright (c) 2015-2017, 2019-2021 Linaro Limited
> > */
> > #include <linux/anon_inodes.h>
> > +#include <linux/cma.h>
> > #include <linux/device.h>
> > #include <linux/dma-buf.h>
> > +#include <linux/dma-map-ops.h>
> > +#include <linux/highmem.h>
> > #include <linux/idr.h>
> > #include <linux/io.h>
> > #include <linux/mm.h>
> > @@ -13,7 +16,6 @@
> > #include <linux/tee_core.h>
> > #include <linux/uaccess.h>
> > #include <linux/uio.h>
> > -#include <linux/highmem.h>
> > #include "tee_private.h"
> >
> > static void shm_put_kernel_pages(struct page **pages, size_t page_count)
> > @@ -49,7 +51,14 @@ static void tee_shm_release(struct tee_device *teedev, struct tee_shm *shm)
> > struct tee_shm *parent_shm = NULL;
> > void *p = shm;
> >
> > - if (shm->flags & TEE_SHM_DMA_BUF) {
> > + if (shm->flags & TEE_SHM_CMA_BUF) {
> > +#if IS_ENABLED(CONFIG_CMA)
> > + struct page *page = phys_to_page(shm->paddr);
> > + struct cma *cma = dev_get_cma_area(&shm->ctx->teedev->dev);
>
> If you want dma_contiguous_default_area as the commit message implies,
> use dma_contiguous_default_area. Appearing to support per-device CMA
> pools but relying on the device not having one is pretty yucky.
>
> But again, why? If you want page-backed DMA-able memory, with all the
> other assumptions being made here, you may as well just rely on
> dma_alloc_pages(DMA_ATTR_SKIP_CPU_SYNC) doing what you want, while also
> being potentially more flexible for !CMA and non-invasive. Or at the
> very least, could the TEE delegation not be composed on top of the
> existing CMA heap allocator?
Thanks for suggesting dma_alloc_pages(). I'll use that in the next
version of the patch set.
Cheers,
Jens
>
> Thanks,
> Robin.
>
> > +
> > + cma_release(cma, page, shm->size / PAGE_SIZE);
> > +#endif
> > + } else if (shm->flags & TEE_SHM_DMA_BUF) {
> > struct tee_shm_dmabuf_ref *ref;
> >
> > ref = container_of(shm, struct tee_shm_dmabuf_ref, shm);
> > @@ -306,6 +315,48 @@ struct tee_shm *tee_shm_alloc_priv_buf(struct tee_context *ctx, size_t size)
> > }
> > EXPORT_SYMBOL_GPL(tee_shm_alloc_priv_buf);
> >
> > +struct tee_shm *tee_shm_alloc_cma_phys_mem(struct tee_context *ctx,
> > + size_t page_count, size_t align)
> > +{
> > +#if IS_ENABLED(CONFIG_CMA)
> > + struct tee_device *teedev = ctx->teedev;
> > + struct cma *cma = dev_get_cma_area(&teedev->dev);
> > + struct tee_shm *shm;
> > + struct page *page;
> > +
> > + if (!tee_device_get(teedev))
> > + return ERR_PTR(-EINVAL);
> > +
> > + page = cma_alloc(cma, page_count, align, true/*no_warn*/);
> > + if (!page)
> > + goto err_put_teedev;
> > +
> > + shm = kzalloc(sizeof(*shm), GFP_KERNEL);
> > + if (!shm)
> > + goto err_cma_crelease;
> > +
> > + refcount_set(&shm->refcount, 1);
> > + shm->ctx = ctx;
> > + shm->paddr = page_to_phys(page);
> > + shm->size = page_count * PAGE_SIZE;
> > + shm->flags = TEE_SHM_CMA_BUF;
> > +
> > + teedev_ctx_get(ctx);
> > +
> > + return shm;
> > +
> > +err_cma_crelease:
> > + cma_release(cma, page, page_count);
> > +err_put_teedev:
> > + tee_device_put(teedev);
> > +
> > + return ERR_PTR(-ENOMEM);
> > +#else
> > + return ERR_PTR(-EINVAL);
> > +#endif
> > +}
> > +EXPORT_SYMBOL_GPL(tee_shm_alloc_cma_phys_mem);
> > +
> > int tee_dyn_shm_alloc_helper(struct tee_shm *shm, size_t size, size_t align,
> > int (*shm_register)(struct tee_context *ctx,
> > struct tee_shm *shm,
> > diff --git a/include/linux/tee_core.h b/include/linux/tee_core.h
> > index 02c07f661349..3a4e1b00fcc7 100644
> > --- a/include/linux/tee_core.h
> > +++ b/include/linux/tee_core.h
> > @@ -29,6 +29,7 @@
> > #define TEE_SHM_POOL BIT(2) /* Memory allocated from pool */
> > #define TEE_SHM_PRIV BIT(3) /* Memory private to TEE driver */
> > #define TEE_SHM_DMA_BUF BIT(4) /* Memory with dma-buf handle */
> > +#define TEE_SHM_CMA_BUF BIT(5) /* CMA allocated memory */
> >
> > #define TEE_DEVICE_FLAG_REGISTERED 0x1
> > #define TEE_MAX_DEV_NAME_LEN 32
> > @@ -310,6 +311,9 @@ void *tee_get_drvdata(struct tee_device *teedev);
> > */
> > struct tee_shm *tee_shm_alloc_priv_buf(struct tee_context *ctx, size_t size);
> >
> > +struct tee_shm *tee_shm_alloc_cma_phys_mem(struct tee_context *ctx,
> > + size_t page_count, size_t align);
> > +
> > int tee_dyn_shm_alloc_helper(struct tee_shm *shm, size_t size, size_t align,
> > int (*shm_register)(struct tee_context *ctx,
> > struct tee_shm *shm,
Powered by blists - more mailing lists