Re: [RFC PATCH 4/4] dma-buf: heaps: add Linaro restricted dmabuf heap support

[Jens Wiklander]

On Wed, Sep 4, 2024 at 11:42 PM T.J. Mercier <tjmercier@xxxxxxxxxx> wrote:
>
> On Wed, Sep 4, 2024 at 2:44 AM Jens Wiklander <jens.wiklander@xxxxxxxxxx> wrote:
> >
> > On Tue, Sep 3, 2024 at 7:50 PM T.J. Mercier <tjmercier@xxxxxxxxxx> wrote:
> > >
> > > On Fri, Aug 30, 2024 at 12:04 AM Jens Wiklander
> > > <jens.wiklander@xxxxxxxxxx> wrote:
> > > >
> > > > Add a Linaro restricted heap using the linaro,restricted-heap bindings
> > > > implemented based on the generic restricted heap.
> > > >
> > > > The bindings defines a range of physical restricted memory. The heap
> > > > manages this address range using genalloc. The allocated dma-buf file
> > > > descriptor can later be registered with the TEE subsystem for later use
> > > > via Trusted Applications in the secure world.
> > > >
> > > > Co-developed-by: Olivier Masse <olivier.masse@xxxxxxx>
> > > > Signed-off-by: Olivier Masse <olivier.masse@xxxxxxx>
> > > > Signed-off-by: Jens Wiklander <jens.wiklander@xxxxxxxxxx>
> > > > ---
> > > >  drivers/dma-buf/heaps/Kconfig                 |  10 ++
> > > >  drivers/dma-buf/heaps/Makefile                |   1 +
> > > >  .../dma-buf/heaps/restricted_heap_linaro.c    | 165 ++++++++++++++++++
> > > >  3 files changed, 176 insertions(+)
> > > >  create mode 100644 drivers/dma-buf/heaps/restricted_heap_linaro.c
> > > >
> > > > diff --git a/drivers/dma-buf/heaps/Kconfig b/drivers/dma-buf/heaps/Kconfig
> > > > index 58903bc62ac8..82e2c5d09242 100644
> > > > --- a/drivers/dma-buf/heaps/Kconfig
> > > > +++ b/drivers/dma-buf/heaps/Kconfig
> > > > @@ -28,3 +28,13 @@ config DMABUF_HEAPS_RESTRICTED_MTK
> > > >         help
> > > >           Enable restricted dma-buf heaps for MediaTek platform. This heap is backed by
> > > >           TEE client interfaces. If in doubt, say N.
> > > > +
> > > > +config DMABUF_HEAPS_RESTRICTED_LINARO
> > > > +       bool "Linaro DMA-BUF Restricted Heap"
> > > > +       depends on DMABUF_HEAPS_RESTRICTED
> > > > +       help
> > > > +         Choose this option to enable the Linaro restricted dma-buf heap.
> > > > +         The restricted heap pools are defined according to the DT. Heaps
> > > > +         are allocated in the pools using gen allocater.
> > > > +         If in doubt, say N.
> > > > +
> > > > diff --git a/drivers/dma-buf/heaps/Makefile b/drivers/dma-buf/heaps/Makefile
> > > > index 0028aa9d875f..66b2f67c47b5 100644
> > > > --- a/drivers/dma-buf/heaps/Makefile
> > > > +++ b/drivers/dma-buf/heaps/Makefile
> > > > @@ -2,4 +2,5 @@
> > > >  obj-$(CONFIG_DMABUF_HEAPS_CMA)         += cma_heap.o
> > > >  obj-$(CONFIG_DMABUF_HEAPS_RESTRICTED)  += restricted_heap.o
> > > >  obj-$(CONFIG_DMABUF_HEAPS_RESTRICTED_MTK)      += restricted_heap_mtk.o
> > > > +obj-$(CONFIG_DMABUF_HEAPS_RESTRICTED_LINARO)   += restricted_heap_linaro.o
> > > >  obj-$(CONFIG_DMABUF_HEAPS_SYSTEM)      += system_heap.o
> > > > diff --git a/drivers/dma-buf/heaps/restricted_heap_linaro.c b/drivers/dma-buf/heaps/restricted_heap_linaro.c
> > > > new file mode 100644
> > > > index 000000000000..4b08ed514023
> > > > --- /dev/null
> > > > +++ b/drivers/dma-buf/heaps/restricted_heap_linaro.c
> > > > @@ -0,0 +1,165 @@
> > > > +// SPDX-License-Identifier: GPL-2.0
> > > > +/*
> > > > + * DMABUF secure heap exporter
> > > > + *
> > > > + * Copyright 2021 NXP.
> > > > + * Copyright 2024 Linaro Limited.
> > > > + */
> > > > +
> > > > +#define pr_fmt(fmt)     "rheap_linaro: " fmt
> > > > +
> > > > +#include <linux/dma-buf.h>
> > > > +#include <linux/err.h>
> > > > +#include <linux/genalloc.h>
> > > > +#include <linux/module.h>
> > > > +#include <linux/of.h>
> > > > +#include <linux/of_fdt.h>
> > > > +#include <linux/of_reserved_mem.h>
> > > > +#include <linux/scatterlist.h>
> > > > +#include <linux/slab.h>
> > > > +
> > > > +#include "restricted_heap.h"
> > > > +
> > > > +#define MAX_HEAP_COUNT 2
> > >
> > > Are multiple supported because of what Cyrille mentioned here about permissions?
> > > https://lore.kernel.org/lkml/DBBPR04MB7514E006455AEA407041E4F788709@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> >
> > Yes, I kept that as is.
>
> Ok thanks.
>
> > > So this is just some arbitrary limit? I'd prefer to have some sort of
> > > documentation about this.
> >
> > How about removing the limit and using dynamic allocation instead?
>
> That works too!

It turns out that was easier said than done. The limit is hardcoded
because dynamic memory allocation isn't available at that stage during
boot. We have a short description of this heap in Kconfig. I'll add
something about the limit there if that makes sense.

Thanks,
Jens