Skip to contentSkip to conversation

DMA-BUF Heaps (destaging ION)

PATCH RESEND Superseded John Stultz 7 participants 42 messages Sep 6, 2019 – Oct 22, 2019

Subsystems (3)

v8 of 16 Superseded Latest is v16 →

lore
Share & tools
lore
#0 lore raw

v8 cover DMA-BUF Heaps (destaging ION)

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Thoughts and feedback would be greatly appreciated!

thanks
-john

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

 MAINTAINERS                                   |  18 ++
 drivers/dma-buf/Kconfig                       |  11 +
 drivers/dma-buf/Makefile                      |   2 +
 drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
 drivers/dma-buf/heaps/Kconfig                 |  14 +
 drivers/dma-buf/heaps/Makefile                |   4 +
 drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
 drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
 drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
 include/linux/dma-heap.h                      |  59 ++++
 include/uapi/linux/dma-heap.h                 |  55 ++++
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
 14 files changed, 1262 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c
Signature · 1 line
2.17.1
#1 lore raw

PATCH RESEND v8 1/5 dma-buf: Add dma-buf heaps framework

From: "Andrew F. Davis" <[email protected]>

This framework allows a unified userspace interface for dma-buf exporters, allowing userland to allocate specific types of memory for use in dma-buf sharing.

Each heap is given its own device node, which a user can allocate a dma-buf fd from using the DMA_HEAP_IOC_ALLOC.

This code is an evoluiton of the Android ION implementation, and a big thanks is due to its authors/maintainers over time for their effort:
Rebecca Schultz Zavin, Colin Cross, Benjamin Gaignard, Laura Abbott, and many other contributors!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Folded down fixes I had previously shared in implementing heaps
* Make flags a u64 (Suggested by Laura)
* Add PAGE_ALIGN() fix to the core alloc funciton
* IOCTL fixups suggested by Brian
* Added fixes suggested by Benjamin
* Removed core stats mgmt, as that should be implemented by per-heap code
* Changed alloc to return a dma-buf fd, rather than a buffer (as it simplifies error handling)
v3:
* Removed scare-quotes in MAINTAINERS email address
* Get rid of .release function as it didn't do anything (from Christoph)
* Renamed filp to file (suggested by Christoph)
* Split out ioctl handling to separate function (suggested by Christoph)
* Add comment documenting PAGE_ALIGN usage (suggested by Brian)
* Switch from idr to Xarray (suggested by Brian)
* Fixup cdev creation (suggested by Brian)
* Avoid EXPORT_SYMBOL until we finalize modules (suggested by Brian)
* Make struct dma_heap internal only (folded in from Andrew)
* Small cleanups suggested by GregKH
* Provide class->devnode callback to get consistent /dev/ subdirectory naming (Suggested by Bjorn)
v4:
* Folded down dma-heap.h change that was in a following patch
* Added fd_flags entry to allocation structure and pass it through to heap code for use on dma-buf fd creation (suggested by Benjamin)
v5:
* Minor cleanups
v6:
* Improved error path handling, minor whitespace fixes, both suggested by Brian
v7:
* Longer Kconfig description to quiet checkpatch warnings
* Re-add compat_ioctl bits (Hridya noticed 32bit userland wasn't working)
v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Checkpatch whitespace fixups

---
 MAINTAINERS                   |  18 +++
 drivers/dma-buf/Kconfig       |   9 ++
 drivers/dma-buf/Makefile      |   1 +
 drivers/dma-buf/dma-heap.c    | 250 ++++++++++++++++++++++++++++++++++
 include/linux/dma-heap.h      |  59 ++++++++
 include/uapi/linux/dma-heap.h |  55 ++++++++
 6 files changed, 392 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
MAINTAINERS +18 -0
index e7a47b5210fd..13e564e37161 100644
@@ -4904,6 +4904,24 @@ F:	include/linux/*fence.h
 F:	Documentation/driver-api/dma-buf.rst
 T:	git git://anongit.freedesktop.org/drm/drm-misc
 
+DMA-BUF HEAPS FRAMEWORK
+M:	Sumit Semwal <[email protected]>
+R:	Andrew F. Davis <[email protected]>
+R:	Benjamin Gaignard <[email protected]>
+R:	Liam Mark <[email protected]>
+R:	Laura Abbott <[email protected]>
+R:	Brian Starkey <[email protected]>
+R:	John Stultz <[email protected]>
+S:	Maintained
+L:	[email protected]
+L:	[email protected]
+L:	[email protected] (moderated for non-subscribers)
+F:	include/uapi/linux/dma-heap.h
+F:	include/linux/dma-heap.h
+F:	drivers/dma-buf/dma-heap.c
+F:	drivers/dma-buf/heaps/*
+T:	git git://anongit.freedesktop.org/drm/drm-misc
+
 DMA GENERIC OFFLOAD ENGINE SUBSYSTEM
 M:	Vinod Koul <[email protected]>
 L:	[email protected]
drivers/dma-buf/Kconfig +9 -0
index b6a9c2f1bc41..162e24e1e429 100644
@@ -39,4 +39,13 @@ config UDMABUF
 	  A driver to let userspace turn memfd regions into dma-bufs.
 	  Qemu can use this to create host dmabufs for guest framebuffers.
 
+menuconfig DMABUF_HEAPS
+	bool "DMA-BUF Userland Memory Heaps"
+	select DMA_SHARED_BUFFER
+	help
+	  Choose this option to enable the DMA-BUF userland memory heaps,
+	  this options creates per heap chardevs in /dev/dma_heap/ which
+	  allows userspace to use to allocate dma-bufs that can be shared
+	  between drivers.
+
 endmenu
drivers/dma-buf/Makefile +1 -0
index e8c7310cb800..1cb3dd104825 100644
@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 obj-y := dma-buf.o dma-fence.o dma-fence-array.o dma-fence-chain.o \
 	 reservation.o seqno-fence.o
+obj-$(CONFIG_DMABUF_HEAPS)	+= dma-heap.o
 obj-$(CONFIG_SYNC_FILE)		+= sync_file.o
 obj-$(CONFIG_SW_SYNC)		+= sw_sync.o sync_debug.o
 obj-$(CONFIG_UDMABUF)		+= udmabuf.o
drivers/dma-buf/dma-heap.c new file +250 -0

250 lines not on this page · raw message

include/linux/dma-heap.h new file +59 -0

59 lines not on this page · raw message

include/uapi/linux/dma-heap.h new file +55 -0

55 lines not on this page · raw message

Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
reviewer Reviewed-by on v8 1/5
#2 lore raw

Re: [RESEND][PATCH v8 1/5] dma-buf: Add dma-buf heaps framework

Hi John,

Quoted 24 lines · John Stultz

From: "Andrew F. Davis" <[email protected]>

This framework allows a unified userspace interface for dma-buf exporters, allowing userland to allocate specific types of memory for use in dma-buf sharing.

Each heap is given its own device node, which a user can allocate a dma-buf fd from using the DMA_HEAP_IOC_ALLOC.

This code is an evoluiton of the Android ION implementation, and a big thanks is due to its authors/maintainers over time for their effort:
Rebecca Schultz Zavin, Colin Cross, Benjamin Gaignard, Laura Abbott, and many other contributors!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Quoted 8 lines

One miniscule nit from me below, but whether you change it or not, you can add my r-b:

Thanks for pushing this through!

-Brian

Quoted 1 line

---

...

drivers/dma-buf/dma-heap.c line 213

Quoted 9 lines
drivers/dma-buf/dma-heap.c new file +9 -0
+
+	dev_ret = device_create(dma_heap_class,
+				NULL,
+				heap->heap_devt,
+				NULL,
+				heap->name);
+	if (IS_ERR(dev_ret)) {
+		pr_err("dma_heap: Unable to create device\n");
+		err_ret = (struct dma_heap *)dev_ret;

Tiny nit: ERR_CAST() would be more obvious for me here.

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#3 lore raw

Re: [RESEND][PATCH v8 1/5] dma-buf: Add dma-buf heaps framework

Quoted 6 lines · Brian Starkey

One miniscule nit from me below, but whether you change it or not, you can add my r-b:

Thanks for pushing this through!

Thanks again for the review! I'll address your issues and resubmit.

thanks
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#4 lore raw

PATCH RESEND v8 2/5 dma-buf: heaps: Add heap helpers

In reply to John Stultz

Add generic helper dmabuf ops for dma heaps, so we can reduce the amount of duplicative code for the exported dmabufs.

This code is an evolution of the Android ION implementation, so thanks to its original authors and maintainters:
Rebecca Schultz Zavin, Colin Cross, Laura Abbott, and others!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Removed cache management performance hack that I had accidentally folded in.
* Removed stats code that was in helpers
* Lots of checkpatch cleanups
v3:
* Uninline INIT_HEAP_HELPER_BUFFER (suggested by Christoph)
* Switch to WARN on buffer destroy failure (suggested by Brian)
* buffer->kmap_cnt decrementing cleanup (suggested by Christoph)
* Extra buffer->vaddr checking in dma_heap_dma_buf_kmap (suggested by Brian)
* Switch to_helper_buffer from macro to inline function (suggested by Benjamin)
* Rename kmap->vmap (folded in from Andrew)
* Use vmap for vmapping - not begin_cpu_access (folded in from Andrew)
* Drop kmap for now, as its optional (folded in from Andrew)
* Fold dma_heap_map_user into the single caller (foled in from Andrew)
* Folded in patch from Andrew to track page list per heap not sglist, which simplifies the tracking logic
v4:
* Moved dma-heap.h change out to previous patch
v6:
* Minor cleanups and typo fixes suggested by Brian
v7:
* Removed stray ;
* Make init_heap_helper_buffer lowercase, as suggested by Christoph
* Add dmabuf export helper to reduce boilerplate code v8:
* Remove unused private_flags value
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Checkpatch whitespace fixups

---
 drivers/dma-buf/Makefile             |   1 +
 drivers/dma-buf/heaps/Makefile       |   2 +
 drivers/dma-buf/heaps/heap-helpers.c | 269 +++++++++++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h |  55 ++++++
 4 files changed, 327 insertions(+)
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
drivers/dma-buf/Makefile +1 -0
index 1cb3dd104825..e3e3dca29e46 100644
@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 obj-y := dma-buf.o dma-fence.o dma-fence-array.o dma-fence-chain.o \
 	 reservation.o seqno-fence.o
+obj-$(CONFIG_DMABUF_HEAPS)	+= heaps/
 obj-$(CONFIG_DMABUF_HEAPS)	+= dma-heap.o
 obj-$(CONFIG_SYNC_FILE)		+= sync_file.o
 obj-$(CONFIG_SW_SYNC)		+= sw_sync.o sync_debug.o
drivers/dma-buf/heaps/Makefile new file +2 -0
new file mode 100644
index 000000000000..de49898112db
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-y					+= heap-helpers.o
drivers/dma-buf/heaps/heap-helpers.c new file +269 -0

269 lines not on this page · raw message

drivers/dma-buf/heaps/heap-helpers.h new file +55 -0

55 lines not on this page · raw message

Signature · 1 line
2.17.1
reviewer Reviewed-by on v8 2/5
#5 lore raw

Re: [RESEND][PATCH v8 2/5] dma-buf: heaps: Add heap helpers

Hi John,

Quoted 7 lines · John Stultz

Add generic helper dmabuf ops for dma heaps, so we can reduce the amount of duplicative code for the exported dmabufs.

This code is an evolution of the Android ION implementation, so thanks to its original authors and maintainters:
Rebecca Schultz Zavin, Colin Cross, Laura Abbott, and others!

Quoted 16 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Two minor things below.

Quoted 40 lines

---
v2:
* Removed cache management performance hack that I had accidentally folded in.
* Removed stats code that was in helpers
* Lots of checkpatch cleanups
v3:
* Uninline INIT_HEAP_HELPER_BUFFER (suggested by Christoph)
* Switch to WARN on buffer destroy failure (suggested by Brian)
* buffer->kmap_cnt decrementing cleanup (suggested by Christoph)
* Extra buffer->vaddr checking in dma_heap_dma_buf_kmap (suggested by Brian)
* Switch to_helper_buffer from macro to inline function (suggested by Benjamin)
* Rename kmap->vmap (folded in from Andrew)
* Use vmap for vmapping - not begin_cpu_access (folded in from Andrew)
* Drop kmap for now, as its optional (folded in from Andrew)
* Fold dma_heap_map_user into the single caller (foled in from Andrew)
* Folded in patch from Andrew to track page list per heap not sglist, which simplifies the tracking logic
v4:
* Moved dma-heap.h change out to previous patch
v6:
* Minor cleanups and typo fixes suggested by Brian
v7:
* Removed stray ;
* Make init_heap_helper_buffer lowercase, as suggested by Christoph
* Add dmabuf export helper to reduce boilerplate code v8:
* Remove unused private_flags value
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Checkpatch whitespace fixups
---

...

drivers/dma-buf/heaps/heap-helpers.c line 70

Quoted 12 lines
drivers/dma-buf/heaps/heap-helpers.c new file +12 -0
+
+static void *dma_heap_buffer_vmap_get(struct heap_helper_buffer *buffer)
+{
+	void *vaddr;
+
+	if (buffer->vmap_cnt) {
+		buffer->vmap_cnt++;
+		return buffer->vaddr;
+	}
+	vaddr = dma_heap_map_kernel(buffer);
+	if (WARN_ONCE(!vaddr,
+		      "heap->ops->map_kernel should return ERR_PTR on error"))

Looks like the message is out-of-date here.

...

Quoted 4 lines
drivers/dma-buf/heaps/heap-helpers.h new file +4 -0
+
+/**
+ * struct heap_helper_buffer - helper buffer metadata
+ * @heap:		back pointer to the heap the buffer came from

drivers/dma-buf/heaps/heap-helpers.h line 27

Quoted 10 lines
drivers/dma-buf/heaps/heap-helpers.h new file +10 -0
+ * @dmabuf:		backing dma-buf for this buffer
+ * @size:		size of the buffer
+ * @flags:		buffer specific flags
+ * @priv_virt		pointer to heap specific private value
+ * @lock		mutext to protect the data in this structure
+ * @vmap_cnt		count of vmap references on the buffer
+ * @vaddr		vmap'ed virtual address
+ * @pagecount		number of pages in the buffer
+ * @pages		list of page pointers
+ * @attachment		list of device attachments

s/attachment/attachments/

With those fixed, feel free to add:

Thanks,
-Brian

#6 lore raw

PATCH RESEND v8 3/5 dma-buf: heaps: Add system heap to dmabuf heaps

In reply to John Stultz

This patch adds system heap to the dma-buf heaps framework.

This allows applications to get a page-allocator backed dma-buf for non-contiguous memory.

This code is an evolution of the Android ION implementation, so thanks to its original authors and maintainters:
Rebecca Schultz Zavin, Colin Cross, Laura Abbott, and others!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Switch allocate to return dmabuf fd
* Simplify init code
* Checkpatch fixups
* Droped dead system-contig code
v3:
* Whitespace fixups from Benjamin
* Make sure we're zeroing the allocated pages (from Liam)
* Use PAGE_ALIGN() consistently (suggested by Brian)
* Fold in new registration style from Andrew
* Avoid needless dynamic allocation of sys_heap (suggested by Christoph)
* Minor cleanups
* Folded in changes from Andrew to use simplified page list from the heap helpers
v4:
* Optimization to allocate pages in chunks, similar to old pagepool code
* Use fd_flags when creating dmabuf fd (Suggested by Benjamin) v5:
* Back out large order page allocations (was leaking memory, as the page array didn't properly track order size)
v6:
* Minor whitespace change suggested by Brian
* Remove unused variable
v7:
* Use newly lower-cased init_heap_helper_buffer helper
* Add system heap DOS avoidance suggested by Laura from ION code
* Use new dmabuf export helper
v8:
* Make struct dma_heap_ops consts (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Add forgotten include file to fix build issue on x86

---
 drivers/dma-buf/Kconfig             |   2 +
 drivers/dma-buf/heaps/Kconfig       |   6 ++
 drivers/dma-buf/heaps/Makefile      |   1 +
 drivers/dma-buf/heaps/system_heap.c | 122 ++++++++++++++++++++++++++++
 4 files changed, 131 insertions(+)
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
drivers/dma-buf/Kconfig +2 -0
index 162e24e1e429..657ce743abda 100644
@@ -48,4 +48,6 @@ menuconfig DMABUF_HEAPS
 	  allows userspace to use to allocate dma-bufs that can be shared
 	  between drivers.
 
+source "drivers/dma-buf/heaps/Kconfig"
+
 endmenu
drivers/dma-buf/heaps/Kconfig new file +6 -0
new file mode 100644
index 000000000000..205052744169
@@ -0,0 +1,6 @@
+config DMABUF_HEAPS_SYSTEM
+	bool "DMA-BUF System Heap"
+	depends on DMABUF_HEAPS
+	help
+	  Choose this option to enable the system dmabuf heap. The system heap
+	  is backed by pages from the buddy allocator. If in doubt, say Y.
drivers/dma-buf/heaps/Makefile +1 -0
index de49898112db..d1808eca2581 100644
@@ -1,2 +1,3 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-y					+= heap-helpers.o
+obj-$(CONFIG_DMABUF_HEAPS_SYSTEM)	+= system_heap.o
drivers/dma-buf/heaps/system_heap.c new file +122 -0

122 lines not on this page · raw message

Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
reviewer Reviewed-by on v8 3/5
#7 lore raw

Re: [RESEND][PATCH v8 3/5] dma-buf: heaps: Add system heap to dmabuf heaps

Quoted 9 lines · John Stultz

This patch adds system heap to the dma-buf heaps framework.

This allows applications to get a page-allocator backed dma-buf for non-contiguous memory.

This code is an evolution of the Android ION implementation, so thanks to its original authors and maintainters:
Rebecca Schultz Zavin, Colin Cross, Laura Abbott, and others!

Quoted 16 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

LGTM:

#8 lore raw

PATCH RESEND v8 4/5 dma-buf: heaps: Add CMA heap to dmabuf heaps

In reply to John Stultz

This adds a CMA heap, which allows userspace to allocate a dma-buf of contiguous memory out of a CMA region.

This code is an evolution of the Android ION implementation, so thanks to its original author and maintainters:
Benjamin Gaignard, Laura Abbott, and others!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Switch allocate to return dmabuf fd
* Simplify init code
* Checkpatch fixups
v3:
* Switch to inline function for to_cma_heap()
* Minor cleanups suggested by Brian
* Fold in new registration style from Andrew
* Folded in changes from Andrew to use simplified page list from the heap helpers
v4:
* Use the fd_flags when creating dmabuf fd (Suggested by Benjamin)
* Use precalculated pagecount (Suggested by Andrew) v6:
* Changed variable names to improve clarity, as suggested by Brian
v7:
* Use newly lower-cased init_heap_helper_buffer helper
* Use new dmabuf export helper
v8:
* Make struct dma_heap_ops const (Suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Checkpatch whitespace fixups

---
 drivers/dma-buf/heaps/Kconfig    |   8 ++
 drivers/dma-buf/heaps/Makefile   |   1 +
 drivers/dma-buf/heaps/cma_heap.c | 164 +++++++++++++++++++++++++++++++
 3 files changed, 173 insertions(+)
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
drivers/dma-buf/heaps/Kconfig +8 -0
index 205052744169..a5eef06c4226 100644
@@ -4,3 +4,11 @@ config DMABUF_HEAPS_SYSTEM
 	help
 	  Choose this option to enable the system dmabuf heap. The system heap
 	  is backed by pages from the buddy allocator. If in doubt, say Y.
+
+config DMABUF_HEAPS_CMA
+	bool "DMA-BUF CMA Heap"
+	depends on DMABUF_HEAPS && DMA_CMA
+	help
+	  Choose this option to enable dma-buf CMA heap. This heap is backed
+	  by the Contiguous Memory Allocator (CMA). If your system has these
+	  regions, you should say Y here.
drivers/dma-buf/heaps/Makefile +1 -0
index d1808eca2581..6e54cdec3da0 100644
@@ -1,3 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-y					+= heap-helpers.o
 obj-$(CONFIG_DMABUF_HEAPS_SYSTEM)	+= system_heap.o
+obj-$(CONFIG_DMABUF_HEAPS_CMA)		+= cma_heap.o
drivers/dma-buf/heaps/cma_heap.c new file +164 -0

164 lines not on this page · raw message

Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
reviewer
#9 lore raw

Re: [RESEND][PATCH v8 4/5] dma-buf: heaps: Add CMA heap to dmabuf heaps

Hi John,

I spotted one thing below which might be harmless, but best to check.

Quoted 50 lines · John Stultz

This adds a CMA heap, which allows userspace to allocate a dma-buf of contiguous memory out of a CMA region.

This code is an evolution of the Android ION implementation, so thanks to its original author and maintainters:
Benjamin Gaignard, Laura Abbott, and others!

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Switch allocate to return dmabuf fd
* Simplify init code
* Checkpatch fixups
v3:
* Switch to inline function for to_cma_heap()
* Minor cleanups suggested by Brian
* Fold in new registration style from Andrew
* Folded in changes from Andrew to use simplified page list from the heap helpers
v4:
* Use the fd_flags when creating dmabuf fd (Suggested by Benjamin)
* Use precalculated pagecount (Suggested by Andrew) v6:
* Changed variable names to improve clarity, as suggested by Brian
v7:
* Use newly lower-cased init_heap_helper_buffer helper
* Use new dmabuf export helper
v8:
* Make struct dma_heap_ops const (Suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Checkpatch whitespace fixups
---

...

Quoted 51 lines
drivers/dma-buf/heaps/cma_heap.c new file +51 -0
+
+/* dmabuf heap CMA operations functions */
+static int cma_heap_allocate(struct dma_heap *heap,
+			     unsigned long len,
+			     unsigned long fd_flags,
+			     unsigned long heap_flags)
+{
+	struct cma_heap *cma_heap = dma_heap_get_data(heap);
+	struct heap_helper_buffer *helper_buffer;
+	struct page *cma_pages;
+	size_t size = PAGE_ALIGN(len);
+	unsigned long nr_pages = size >> PAGE_SHIFT;
+	unsigned long align = get_order(size);
+	struct dma_buf *dmabuf;
+	int ret = -ENOMEM;
+	pgoff_t pg;
+
+	if (align > CONFIG_CMA_ALIGNMENT)
+		align = CONFIG_CMA_ALIGNMENT;
+
+	helper_buffer = kzalloc(sizeof(*helper_buffer), GFP_KERNEL);
+	if (!helper_buffer)
+		return -ENOMEM;
+
+	init_heap_helper_buffer(helper_buffer, cma_heap_free);
+	helper_buffer->flags = heap_flags;
+	helper_buffer->heap = heap;
+	helper_buffer->size = len;
+
+	cma_pages = cma_alloc(cma_heap->cma, nr_pages, align, false);
+	if (!cma_pages)
+		goto free_buf;
+
+	if (PageHighMem(cma_pages)) {
+		unsigned long nr_clear_pages = nr_pages;
+		struct page *page = cma_pages;
+
+		while (nr_clear_pages > 0) {
+			void *vaddr = kmap_atomic(page);
+
+			memset(vaddr, 0, PAGE_SIZE);
+			kunmap_atomic(vaddr);
+			page++;
+			nr_clear_pages--;
+		}
+	} else {
+		memset(page_address(cma_pages), 0, size);
+	}
+
+	helper_buffer->pagecount = nr_pages;
+	helper_buffer->pages = kmalloc_array(helper_buffer->pagecount,

drivers/dma-buf/heaps/cma_heap.c line 98

Quoted 10 lines
drivers/dma-buf/heaps/cma_heap.c new file +10 -0
+					     sizeof(*helper_buffer->pages),
+					     GFP_KERNEL);
+	if (!helper_buffer->pages) {
+		ret = -ENOMEM;
+		goto free_cma;
+	}
+
+	for (pg = 0; pg < helper_buffer->pagecount; pg++) {
+		helper_buffer->pages[pg] = &cma_pages[pg];
+		if (!helper_buffer->pages[pg])

Is this ever really possible? If cma_pages is non-NULL (which you check earlier), then only if the pointer arithmetic overflows right?

If it's just redundant, then you could remove it (and in that case add my r-b). But maybe you meant to check something else?

Cheers,
-Brian
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#10 lore raw

PATCH RESEND v8 5/5 kselftests: Add dma-heap test

In reply to John Stultz

Add very trivial allocation and import test for dma-heaps, utilizing the vgem driver as a test importer.

A good chunk of this code taken from:
tools/testing/selftests/android/ion/ionmap_test.c Originally by Laura Abbott <[email protected]>

Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Switched to use reworked dma-heap apis
v3:
* Add simple mmap
* Utilize dma-buf testdev to test importing
v4:
* Rework to use vgem
* Pass in fd_flags to match interface changes
* Skip . and .. dirs
v6:
* Number of style/cleanups suggested by Brian
v7:
* Whitespace fixup for checkpatch
v8:
* More checkpatch whitespace fixups

---
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 ++++++++++++++++++
 2 files changed, 239 insertions(+)
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c
tools/testing/selftests/dmabuf-heaps/Makefile new file +9 -0
new file mode 100644
index 000000000000..8c4c36e2972d
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0
+CFLAGS += -static -O3 -Wl,-no-as-needed -Wall
+#LDLIBS += -lrt -lpthread -lm
+
+# these are all "safe" tests that don't modify
+# system time or require escalated privileges
+TEST_GEN_PROGS = dmabuf-heap
+
+include ../lib.mk
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +230 -0

230 lines not on this page · raw message

Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
reviewer
#11 lore raw

Re: [RESEND][PATCH v8 5/5] kselftests: Add dma-heap test

Hi John,

I didn't see any response about using the test harness. Did you decide against it?

Quoted 110 lines · John Stultz

Add very trivial allocation and import test for dma-heaps, utilizing the vgem driver as a test importer.

A good chunk of this code taken from:
tools/testing/selftests/android/ion/ionmap_test.c Originally by Laura Abbott <[email protected]>

Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

---
v2:
* Switched to use reworked dma-heap apis
v3:
* Add simple mmap
* Utilize dma-buf testdev to test importing
v4:
* Rework to use vgem
* Pass in fd_flags to match interface changes
* Skip . and .. dirs
v6:
* Number of style/cleanups suggested by Brian
v7:
* Whitespace fixup for checkpatch
v8:
* More checkpatch whitespace fixups

---
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 ++++++++++++++++++
 2 files changed, 239 insertions(+)
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c
tools/testing/selftests/dmabuf-heaps/Makefile new file +9 -0
new file mode 100644
index 000000000000..8c4c36e2972d
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0
+CFLAGS += -static -O3 -Wl,-no-as-needed -Wall
+#LDLIBS += -lrt -lpthread -lm
+
+# these are all "safe" tests that don't modify
+# system time or require escalated privileges
+TEST_GEN_PROGS = dmabuf-heap
+
+include ../lib.mk
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +44 -0
new file mode 100644
index 000000000000..e439d6cf3d81
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <dirent.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+
+#include <linux/dma-buf.h>
+#include <drm/drm.h>
+
+#include "../../../../include/uapi/linux/dma-heap.h"
+
+#define DEVPATH "/dev/dma_heap"
+
+static int check_vgem(int fd)
+{
+	drm_version_t version = { 0 };
+	char name[5];
+	int ret;
+
+	version.name_len = 4;
+	version.name = name;
+
+	ret = ioctl(fd, DRM_IOCTL_VERSION, &version);
+	if (ret)
+		return 0;
+
+	return !strcmp(name, "vgem");
+}
+
+static int open_vgem(void)
+{
+	int i, fd;
+	const char *drmstr = "/dev/dri/card";
+
+	fd = -1;
+	for (i = 0; i < 16; i++) {

tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c line 54

Quoted 10 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +10 -0
+		char name[80];
+
+		sprintf(name, "%s%u", drmstr, i);
+
+		fd = open(name, O_RDWR);
+		if (fd < 0)
+			continue;
+
+		if (!check_vgem(fd)) {
+			close(fd);

I didn't spot this last time, but there's an (unlikely) error scenario here if there's >= 16 DRM devices and none of them are vgem, then you'll return a stale fd.

Quoted 28 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +28 -0
+			continue;
+		} else {
+			break;
+		}
+	}
+	return fd;
+}
+
+static int import_vgem_fd(int vgem_fd, int dma_buf_fd, uint32_t *handle)
+{
+	struct drm_prime_handle import_handle = {
+		.fd = dma_buf_fd,
+		.flags = 0,
+		.handle = 0,
+	 };
+	int ret;
+
+	ret = ioctl(vgem_fd, DRM_IOCTL_PRIME_FD_TO_HANDLE, &import_handle);
+	if (ret == 0)
+		*handle = import_handle.handle;
+	return ret;
+}
+
+static void close_handle(int vgem_fd, uint32_t handle)
+{
+	struct drm_gem_close close = {
+		.handle = handle,
+	};

tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c line 92

Quoted 10 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +10 -0
+
+	ioctl(vgem_fd, DRM_IOCTL_GEM_CLOSE, &close);
+}
+
+static int dmabuf_heap_open(char *name)
+{
+	int ret, fd;
+	char buf[256];
+
+	ret = sprintf(buf, "%s/%s", DEVPATH, name);

snprintf(), just because why not?

Quoted 123 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +123 -0
+	if (ret < 0) {
+		printf("sprintf failed!\n");
+		return ret;
+	}
+
+	fd = open(buf, O_RDWR);
+	if (fd < 0)
+		printf("open %s failed!\n", buf);
+	return fd;
+}
+
+static int dmabuf_heap_alloc(int fd, size_t len, unsigned int flags,
+			     int *dmabuf_fd)
+{
+	struct dma_heap_allocation_data data = {
+		.len = len,
+		.fd_flags = O_RDWR | O_CLOEXEC,
+		.heap_flags = flags,
+	};
+	int ret;
+
+	if (!dmabuf_fd)
+		return -EINVAL;
+
+	ret = ioctl(fd, DMA_HEAP_IOC_ALLOC, &data);
+	if (ret < 0)
+		return ret;
+	*dmabuf_fd = (int)data.fd;
+	return ret;
+}
+
+static void dmabuf_sync(int fd, int start_stop)
+{
+	struct dma_buf_sync sync = {
+		.flags = start_stop | DMA_BUF_SYNC_RW,
+	};
+	int ret;
+
+	ret = ioctl(fd, DMA_BUF_IOCTL_SYNC, &sync);
+	if (ret)
+		printf("sync failed %d\n", errno);
+}
+
+#define ONE_MEG (1024 * 1024)
+
+static void do_test(char *heap_name)
+{
+	int heap_fd = -1, dmabuf_fd = -1, importer_fd = -1;
+	uint32_t handle = 0;
+	void *p = NULL;
+	int ret;
+
+	printf("Testing heap: %s\n", heap_name);
+
+	heap_fd = dmabuf_heap_open(heap_name);
+	if (heap_fd < 0)
+		return;
+
+	printf("Allocating 1 MEG\n");
+	ret = dmabuf_heap_alloc(heap_fd, ONE_MEG, 0, &dmabuf_fd);
+	if (ret) {
+		printf("Allocation Failed!\n");
+		goto out;
+	}
+	/* mmap and write a simple pattern */
+	p = mmap(NULL,
+		 ONE_MEG,
+		 PROT_READ | PROT_WRITE,
+		 MAP_SHARED,
+		 dmabuf_fd,
+		 0);
+	if (p == MAP_FAILED) {
+		printf("mmap() failed: %m\n");
+		goto out;
+	}
+	printf("mmap passed\n");
+
+	dmabuf_sync(dmabuf_fd, DMA_BUF_SYNC_START);
+
+	memset(p, 1, ONE_MEG / 2);
+	memset((char *)p + ONE_MEG / 2, 0, ONE_MEG / 2);
+	dmabuf_sync(dmabuf_fd, DMA_BUF_SYNC_END);
+
+	importer_fd = open_vgem();
+	if (importer_fd < 0) {
+		ret = importer_fd;
+		printf("Failed to open vgem\n");
+		goto out;
+	}
+
+	ret = import_vgem_fd(importer_fd, dmabuf_fd, &handle);
+	if (ret < 0) {
+		printf("Failed to import buffer\n");
+		goto out;
+	}
+	printf("import passed\n");
+
+	dmabuf_sync(dmabuf_fd, DMA_BUF_SYNC_START);
+	memset(p, 0xff, ONE_MEG);
+	dmabuf_sync(dmabuf_fd, DMA_BUF_SYNC_END);
+	printf("syncs passed\n");
+
+	close_handle(importer_fd, handle);
+
+out:
+	if (p)
+		munmap(p, ONE_MEG);
+	if (importer_fd >= 0)
+		close(importer_fd);
+	if (dmabuf_fd >= 0)
+		close(dmabuf_fd);
+	if (heap_fd >= 0)
+		close(heap_fd);
+}
+
+int main(void)
+{
+	DIR *d;
+	struct dirent *dir;
+
+	d = opendir(DEVPATH);
+	if (!d) {
+		printf("No %s directory?\n", DEVPATH);

tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c line 225

Quoted 10 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +10 -0
+		return -1;
+	}
+
+	while ((dir = readdir(d)) != NULL) {
+		if (!strncmp(dir->d_name, ".", 2))
+			continue;
+		if (!strncmp(dir->d_name, "..", 3))
+			continue;
+
+		do_test(dir->d_name);

As far as I understand it, if main() always returns zero, this test will always be indicated as a "pass" - shouldn't there be at least some failure scenarios?

Cheers,
-Brian

Quoted 8 lines
tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c new file +5 -0
+	}
+	closedir(d);
+
+	return 0;
+}
Signature · 1 line
2.17.1
#12 lore raw

Re: [RESEND][PATCH v8 5/5] kselftests: Add dma-heap test

Quoted 3 lines · Brian Starkey

I didn't see any response about using the test harness. Did you decide against it?

Hey! Spent a little time looking at this bit and just wanted to reply to this point. So first, apologies, I think I missed the suggestion earlier. That said, now that I've looked a little bit at the test harness, and at least at this point it feels like it makes it harder to reason with than standard c code. Maybe I need to spend a bit more time on it, but I'm a little hesitant to swap over just yet.

I'm not particularly passionate on this point, but are you? Or was this just a recommendation to check it out and consider it?

thanks
-john

reviewer
#13 lore raw

Re: [RESEND][PATCH v8 5/5] kselftests: Add dma-heap test

Hi John,

Quoted 12 lines · John Stultz
Quoted 3 lines · Brian Starkey

I didn't see any response about using the test harness. Did you decide against it?

Hey! Spent a little time looking at this bit and just wanted to reply to this point. So first, apologies, I think I missed the suggestion earlier. That said, now that I've looked a little bit at the test harness, and at least at this point it feels like it makes it harder to reason with than standard c code. Maybe I need to spend a bit more time on it, but I'm a little hesitant to swap over just yet.

Quoted 2 lines

I'm not particularly passionate on this point, but are you? Or was this just a recommendation to check it out and consider it?

No particular strong feelings. I was just poking around the kernel docs for testing and the only info there is about the test harness, so wanted to check if you'd seen/considered it.

A quick grep of tools/testing shows that it's obviously not that popular... so if you don't fancy it I won't complain.

Thanks,
-Brian

Quoted 3 lines

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

maintainer
#14 lore raw
In reply to John Stultz

Hello Christoph, everyone,

Quoted 28 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

Quoted 7 lines

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

It looks like most of the feedback has been taken care of. If there's no more objection to this series, I'd like to merge it in soon.

If there are any more review comments, may I request you to please provide them?

Quoted 14 lines

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Quoted 4 lines

Thoughts and feedback would be greatly appreciated!

thanks
-john

Best,
Sumit.

Quoted 56 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

 MAINTAINERS                                   |  18 ++
 drivers/dma-buf/Kconfig                       |  11 +
 drivers/dma-buf/Makefile                      |   2 +
 drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
 drivers/dma-buf/heaps/Kconfig                 |  14 +
 drivers/dma-buf/heaps/Makefile                |   4 +
 drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
 drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
 drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
 include/linux/dma-heap.h                      |  59 ++++
 include/uapi/linux/dma-heap.h                 |  55 ++++
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
 14 files changed, 1262 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c

--
2.17.1

Signature · 9 lines
Thanks and regards,

Sumit Semwal
Linaro Consumer Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Tested-by
#15 lore raw
Quoted 39 lines · Sumit Semwal

Hello Christoph, everyone,

Quoted 28 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

Quoted 7 lines

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

Quoted 4 lines

It looks like most of the feedback has been taken care of. If there's no more objection to this series, I'd like to merge it in soon.

If there are any more review comments, may I request you to please provide them?

I tested these patches using our internal test suite with Arm,komeda driver and the following node in dts

        reserved-memory {
                #address-cells = <0x2>;
                #size-cells = <0x2>;
                ranges;

                framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reg = <0x0 0x60000000 0x0 0x8000000>;
                };
        }

The tests went fine. Our tests allocates framebuffers of different sizes, posts them on screen and the driver writes back to one of the framebuffers. I havenot tested for any performance, latency or cache management related stuff. So, it that looks appropriate, feel free to add:-

Are you planning to write some igt tests for it ?

Quoted 89 lines
Quoted 15 lines

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Quoted 4 lines

Thoughts and feedback would be greatly appreciated!

thanks
-john

Best,
Sumit.

Quoted 56 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

 MAINTAINERS                                   |  18 ++
 drivers/dma-buf/Kconfig                       |  11 +
 drivers/dma-buf/Makefile                      |   2 +
 drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
 drivers/dma-buf/heaps/Kconfig                 |  14 +
 drivers/dma-buf/heaps/Makefile                |   4 +
 drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
 drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
 drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
 include/linux/dma-heap.h                      |  59 ++++
 include/uapi/linux/dma-heap.h                 |  55 ++++
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
 14 files changed, 1262 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c

--
2.17.1

Signature · 9 lines
Thanks and regards,

Sumit Semwal
Linaro Consumer Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#16 lore raw
Quoted 15 lines · Ayan Halder

I tested these patches using our internal test suite with Arm,komeda driver and the following node in dts

        reserved-memory {
                #address-cells = <0x2>;
                #size-cells = <0x2>;
                ranges;

                framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reg = <0x0 0x60000000 0x0 0x8000000>;
                };
        }
Quoted 6 lines

The tests went fine. Our tests allocates framebuffers of different sizes, posts them on screen and the driver writes back to one of the framebuffers. I havenot tested for any performance, latency or cache management related stuff. So, it that looks appropriate, feel free to add:-

Thanks so much for testing! I really appreciate it!

Quoted 1 line

Are you planning to write some igt tests for it ?

I've not personally as familiar with igt yet, which is why I started with kselftest, but it's a good idea. I'll take a look and try to take a swing at it after Sumit queues the patchset (I need to resubmit to address Brian's feedback).

thanks
-john

#17 lore raw
In reply to Ayan Halder
Quoted 48 lines · Ayan Halder
Quoted 39 lines · Sumit Semwal

Hello Christoph, everyone,

Quoted 28 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

Quoted 7 lines

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

Quoted 4 lines

It looks like most of the feedback has been taken care of. If there's no more objection to this series, I'd like to merge it in soon.

If there are any more review comments, may I request you to please provide them?

I tested these patches using our internal test suite with Arm,komeda driver and the following node in dts

Quoted 11 lines
        reserved-memory {
                #address-cells = <0x2>;
                #size-cells = <0x2>;
                ranges;

                framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reg = <0x0 0x60000000 0x0 0x8000000>;
                };
        }

Apologies for the confusion, this dts node is irrelevant as our tests were using the cma heap (via /dev/dma_heap/reserved).

That raises a question. How do we represent the reserved-memory nodes (as shown above) via the dma-buf heaps framework ?

Quoted 98 lines

The tests went fine. Our tests allocates framebuffers of different sizes, posts them on screen and the driver writes back to one of the framebuffers. I havenot tested for any performance, latency or cache management related stuff. So, it that looks appropriate, feel free to add:-

Are you planning to write some igt tests for it ?

Quoted 74 lines
Quoted 15 lines

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Quoted 4 lines

Thoughts and feedback would be greatly appreciated!

thanks
-john

Best,
Sumit.

Quoted 53 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

 MAINTAINERS                                   |  18 ++
 drivers/dma-buf/Kconfig                       |  11 +
 drivers/dma-buf/Makefile                      |   2 +
 drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
 drivers/dma-buf/heaps/Kconfig                 |  14 +
 drivers/dma-buf/heaps/Makefile                |   4 +
 drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
 drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
 drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
 include/linux/dma-heap.h                      |  59 ++++
 include/uapi/linux/dma-heap.h                 |  55 ++++
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
 14 files changed, 1262 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c
Quoted 15 lines
Quoted 3 lines

--
2.17.1

Signature · 9 lines
Thanks and regards,

Sumit Semwal
Linaro Consumer Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 4 lines

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#18 lore raw
Quoted 60 lines · Ayan Halder
Quoted 48 lines · Ayan Halder
Quoted 39 lines · Sumit Semwal

Hello Christoph, everyone,

Quoted 28 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

Quoted 7 lines

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

Quoted 4 lines

It looks like most of the feedback has been taken care of. If there's no more objection to this series, I'd like to merge it in soon.

If there are any more review comments, may I request you to please provide them?

I tested these patches using our internal test suite with Arm,komeda driver and the following node in dts

Quoted 11 lines
        reserved-memory {
                #address-cells = <0x2>;
                #size-cells = <0x2>;
                ranges;

                framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reg = <0x0 0x60000000 0x0 0x8000000>;
                };
        }
Quoted 5 lines

Apologies for the confusion, this dts node is irrelevant as our tests were using the cma heap (via /dev/dma_heap/reserved).

That raises a question. How do we represent the reserved-memory nodes (as shown above) via the dma-buf heaps framework ?

The CMA driver that registers these nodes will have to be expanded to export them using this framework as needed. We do something similar to export SRAM nodes:

https://lkml.org/lkml/2019/3/21/575

Unlike the system/default-cma driver which can be centralized in the tree, these extra exporters will probably live out in other subsystems and so are added in later steps.

Andrew

Quoted 102 lines
Quoted 102 lines

The tests went fine. Our tests allocates framebuffers of different sizes, posts them on screen and the driver writes back to one of the framebuffers. I havenot tested for any performance, latency or cache management related stuff. So, it that looks appropriate, feel free to add:-

Are you planning to write some igt tests for it ?

Quoted 74 lines
Quoted 15 lines

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Quoted 4 lines

Thoughts and feedback would be greatly appreciated!

thanks
-john

Best,
Sumit.

Quoted 53 lines
Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

 MAINTAINERS                                   |  18 ++
 drivers/dma-buf/Kconfig                       |  11 +
 drivers/dma-buf/Makefile                      |   2 +
 drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
 drivers/dma-buf/heaps/Kconfig                 |  14 +
 drivers/dma-buf/heaps/Makefile                |   4 +
 drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
 drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
 drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
 drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
 include/linux/dma-heap.h                      |  59 ++++
 include/uapi/linux/dma-heap.h                 |  55 ++++
 tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
 .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
 14 files changed, 1262 insertions(+)
 create mode 100644 drivers/dma-buf/dma-heap.c
 create mode 100644 drivers/dma-buf/heaps/Kconfig
 create mode 100644 drivers/dma-buf/heaps/Makefile
 create mode 100644 drivers/dma-buf/heaps/cma_heap.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.c
 create mode 100644 drivers/dma-buf/heaps/heap-helpers.h
 create mode 100644 drivers/dma-buf/heaps/system_heap.c
 create mode 100644 include/linux/dma-heap.h
 create mode 100644 include/uapi/linux/dma-heap.h
 create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile
 create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c
Quoted 15 lines
Quoted 3 lines

--
2.17.1

Signature · 9 lines
Thanks and regards,

Sumit Semwal
Linaro Consumer Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

reviewer
#19 lore raw

Hi Andrew,

Quoted 11 lines · Andrew F. Davis

The CMA driver that registers these nodes will have to be expanded to export them using this framework as needed. We do something similar to export SRAM nodes:

https://lkml.org/lkml/2019/3/21/575

Unlike the system/default-cma driver which can be centralized in the tree, these extra exporters will probably live out in other subsystems and so are added in later steps.

Andrew

I was under the impression that the "cma_for_each_area" loop in patch 4 would do that (add_cma_heaps). Is it not the case?

Thanks,
-Brian

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#20 lore raw
Quoted 9 lines · Brian Starkey

Hi Andrew,

Quoted 6 lines · Andrew F. Davis

The CMA driver that registers these nodes will have to be expanded to export them using this framework as needed. We do something similar to export SRAM nodes:

https://lkml.org/lkml/2019/3/21/575

Quoted 9 lines
Quoted 5 lines

Unlike the system/default-cma driver which can be centralized in the tree, these extra exporters will probably live out in other subsystems and so are added in later steps.

Andrew

I was under the impression that the "cma_for_each_area" loop in patch 4 would do that (add_cma_heaps). Is it not the case?

For these cma nodes yes, I thought you meant reserved memory areas in general.

Just as a side note, I'm not a huge fan of the cma_for_each_area() to begin with, it seems a bit out of place when they could be selectively added as heaps as needed. Not sure how that will work with cma nodes specifically assigned to devices, seems like we could just steal their memory space from userspace with this..

Andrew

Quoted 3 lines

Thanks,
-Brian

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#21 lore raw
Quoted 14 lines · Andrew F. Davis
Quoted 13 lines · Brian Starkey

Hi Andrew,

Quoted 10 lines · Andrew F. Davis

The CMA driver that registers these nodes will have to be expanded to export them using this framework as needed. We do something similar to export SRAM nodes:

https://lkml.org/lkml/2019/3/21/575

Unlike the system/default-cma driver which can be centralized in the tree, these extra exporters will probably live out in other subsystems and so are added in later steps.

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Andrew

I was under the impression that the "cma_for_each_area" loop in patch 4 would do that (add_cma_heaps). Is it not the case?

For these cma nodes yes, I thought you meant reserved memory areas in general.

Ok, sorry I didn't see this earlier, not only was I still dropped from the To list, but the copy I got from dri-devel ended up marked as spam.

Quoted 5 lines

Just as a side note, I'm not a huge fan of the cma_for_each_area() to begin with, it seems a bit out of place when they could be selectively added as heaps as needed. Not sure how that will work with cma nodes specifically assigned to devices, seems like we could just steal their memory space from userspace with this..

So this would be a concern with ION as well, since it does the same thing because being able to allocate from multiple CMA heaps for device specific purpose is really useful.
And at least with dmabuf heaps each heap can be given its own permissions so there's less likelihood for any abuse as you describe.

And it also allows various device cma nodes to still be allocated from using the same interface (rather then having to use a custom driver ioctl for each device).

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

thanks
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#22 lore raw
Quoted 34 lines · John Stultz
Quoted 14 lines · Andrew F. Davis
Quoted 13 lines · Brian Starkey

Hi Andrew,

Quoted 10 lines · Andrew F. Davis

The CMA driver that registers these nodes will have to be expanded to export them using this framework as needed. We do something similar to export SRAM nodes:

https://lkml.org/lkml/2019/3/21/575

Unlike the system/default-cma driver which can be centralized in the tree, these extra exporters will probably live out in other subsystems and so are added in later steps.

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Andrew

I was under the impression that the "cma_for_each_area" loop in patch 4 would do that (add_cma_heaps). Is it not the case?

For these cma nodes yes, I thought you meant reserved memory areas in general.

Ok, sorry I didn't see this earlier, not only was I still dropped from the To list, but the copy I got from dri-devel ended up marked as spam.

Quoted 5 lines

Just as a side note, I'm not a huge fan of the cma_for_each_area() to begin with, it seems a bit out of place when they could be selectively added as heaps as needed. Not sure how that will work with cma nodes specifically assigned to devices, seems like we could just steal their memory space from userspace with this..

Quoted 6 lines

So this would be a concern with ION as well, since it does the same thing because being able to allocate from multiple CMA heaps for device specific purpose is really useful.
And at least with dmabuf heaps each heap can be given its own permissions so there's less likelihood for any abuse as you describe.

Yes it was a problem with ION also, having individual files per heap does help with some permissions, but my issue is what if I don't want my CMA exported at all, cma_for_each_area() just grabs them all anyway.

Quoted 4 lines

And it also allows various device cma nodes to still be allocated from using the same interface (rather then having to use a custom driver ioctl for each device).

This is definitely the way to go, it's the implementation of how we get the CMAs to export in the first place that is a bit odd.

Quoted 3 lines

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Andrew

Quoted 3 lines

thanks
-john

#23 lore raw
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

So, <sigh>, I'll start on the rework for the CMA bits.

That said, I'm definitely wanting to make some progress on this patch series, so maybe we can still merge the core/helpers/system heap and just hold the cma heap for a rework on the enumeration bits. That way we can at least get other folks working on switching their vendor heaps from ION.

Sumit: Does that sound ok? Assuming no other objections, can you take the v11 set minus the CMA heap patch?

thanks
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

reviewer
#24 lore raw
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

Cheers,
-Brian

Quoted 9 lines

So, <sigh>, I'll start on the rework for the CMA bits.

That said, I'm definitely wanting to make some progress on this patch series, so maybe we can still merge the core/helpers/system heap and just hold the cma heap for a rework on the enumeration bits. That way we can at least get other folks working on switching their vendor heaps from ION.

Quoted 5 lines

Sumit: Does that sound ok? Assuming no other objections, can you take the v11 set minus the CMA heap patch?

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#25 lore raw
Quoted 26 lines · Brian Starkey
Quoted 25 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Quoted 6 lines
Quoted 3 lines

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

How about only auto-adding the system default CMA region (cma->name == "reserved")?

Great minds... :)

Quoted 2 lines

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

thanks
-john

#26 lore raw
In reply to Brian Starkey
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Quoting reserved-memory.txt :-
"The operating system can use the memory in this region with the limitation that
the device driver(s) owning the region need to be able to reclaim it back"

Thus as per my observation, without 'reusable', rmem_cma_setup() returns -EINVAL and the reserved-memory is not added as a cma region.

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c
[    0.458415] Modules linked in:                                                                                                             
[    0.461470] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.3.0-rc4-01377-g51dbcf03884c-dirty #15                                              
[    0.470017] Hardware name: ARM Juno development board (r0) (DT)                                                                            
[    0.475953] pstate: 80000005 (Nzcv daif -PAN -UAO)                                                                                         
[    0.480755] pc : cma_init_reserved_areas+0xec/0x22c  
[    0.485643] lr : cma_init_reserved_areas+0xe8/0x22c 

<----snip register dump --->

[    0.600646] Unable to handle kernel paging request at virtual address ffff7dffff800000
[    0.608591] Mem abort info:
[    0.611386]   ESR = 0x96000006
<---snip uninteresting bits --->
[    0.681069] pc : cma_init_reserved_areas+0x114/0x22c
[    0.686043] lr : cma_init_reserved_areas+0xe8/0x22c

I am looking into this now. My final objective is to get "/dev/dma_heap/framebuffer" (as a cma heap).
Any leads?

Quoted 17 lines

Cheers,
-Brian

Quoted 14 lines

So, <sigh>, I'll start on the rework for the CMA bits.

That said, I'm definitely wanting to make some progress on this patch series, so maybe we can still merge the core/helpers/system heap and just hold the cma heap for a rework on the enumeration bits. That way we can at least get other folks working on switching their vendor heaps from ION.

Sumit: Does that sound ok? Assuming no other objections, can you take the v11 set minus the CMA heap patch?

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#27 lore raw
Quoted 40 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

Quoted 12 lines
        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Right. It has to be a CMA region for us to expose it from the cma heap.

Quoted 3 lines

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

thanks
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#28 lore raw
Quoted 57 lines · John Stultz
Quoted 40 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

Quoted 12 lines
        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Right. It has to be a CMA region for us to expose it from the cma heap.

Quoted 7 lines
Quoted 3 lines

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

It is a valid memory region we use to allocate framebuffers.

Quoted 3 lines

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#29 lore raw
Quoted 10 lines · Ayan Halder
Quoted 8 lines · John Stultz
Quoted 3 lines · Ayan Halder

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

It is a valid memory region we use to allocate framebuffers.

Hrm. I guess I'd suggest digging to figure out why the kernel doesn't see it as such.

Does this only happen with my patches applied? I'm sort of assuming can you trip this even without them, but maybe I'm wrong?

thanks
-john

#30 lore raw
In reply to Ayan Halder
Quoted 58 lines · Ayan Halder
Quoted 57 lines · John Stultz
Quoted 40 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

Quoted 12 lines
        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Right. It has to be a CMA region for us to expose it from the cma heap.

Quoted 8 lines
Quoted 7 lines
Quoted 3 lines

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

It is a valid memory region we use to allocate framebuffers.

But does it have a valid kernel virtual mapping? Most ARM systems (just assuming you are working on ARM :)) that I'm familiar with have the DRAM space starting at 0x80000000 and so don't start having valid pfns until that point. Is this address you are reserving an SRAM?

Andrew

Quoted 3 lines
Quoted 3 lines

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

reviewer
#31 lore raw
Quoted 66 lines · Andrew F. Davis
Quoted 65 lines · Ayan Halder
Quoted 64 lines · John Stultz
Quoted 40 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

Quoted 12 lines
        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Right. It has to be a CMA region for us to expose it from the cma heap.

Quoted 3 lines

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

Quoted 8 lines
Quoted 1 line

It is a valid memory region we use to allocate framebuffers.

But does it have a valid kernel virtual mapping? Most ARM systems (just assuming you are working on ARM :)) that I'm familiar with have the DRAM space starting at 0x80000000 and so don't start having valid pfns until that point. Is this address you are reserving an SRAM?

Yeah, I think you've got it.

This region is DRAM on an FPGA expansion tile, but as you have noticed its "below" the start of main RAM, and I expect it's not in any of the declared /memory/ nodes.

When "reusable" isn't there, I think we'll end up going the coherent.c route, with dma_init_coherent_memory() setting up some pages.

If "reusable" is there, then I think we'll end up in contiguous.c and that expects us to already have pages.

So, @Ayan, you could perhaps try adding this region as a /memory/ node as-well, which should mean the kernel sets up some pages for it as normal memory. But, I have some ancient recollection that the arm64 kernel couldn't handle system RAM at addresses below 0x80000000 or something. That might be different now, I'm talking about several years ago.

Thanks,
-Brian

Quoted 6 lines

Andrew

Quoted 3 lines
Quoted 3 lines

thanks
-john

#32 lore raw
Quoted 88 lines · Brian Starkey
Quoted 66 lines · Andrew F. Davis
Quoted 65 lines · Ayan Halder
Quoted 64 lines · John Stultz
Quoted 40 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

Quoted 12 lines
        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Right. It has to be a CMA region for us to expose it from the cma heap.

Quoted 3 lines

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c

Is the value 0x60000000 you're using something you just guessed at? It seems like the warning here is saying the pfn calculated from the base address isn't valid.

Quoted 8 lines
Quoted 1 line

It is a valid memory region we use to allocate framebuffers.

But does it have a valid kernel virtual mapping? Most ARM systems (just assuming you are working on ARM :)) that I'm familiar with have the DRAM space starting at 0x80000000 and so don't start having valid pfns until that point. Is this address you are reserving an SRAM?

Yeah, I think you've got it.

This region is DRAM on an FPGA expansion tile, but as you have noticed its "below" the start of main RAM, and I expect it's not in any of the declared /memory/ nodes.

When "reusable" isn't there, I think we'll end up going the coherent.c route, with dma_init_coherent_memory() setting up some pages.

If "reusable" is there, then I think we'll end up in contiguous.c and that expects us to already have pages.

Quoted 7 lines

So, @Ayan, you could perhaps try adding this region as a /memory/ node as-well, which should mean the kernel sets up some pages for it as normal memory. But, I have some ancient recollection that the arm64 kernel couldn't handle system RAM at addresses below 0x80000000 or something. That might be different now, I'm talking about several years ago.

Thanks a lot for your suggestions.

I added the following node in the dts.

       memory@60000000 {
               device_type = "memory";
               reg = <0 0x60000000 0 0x08000000>;
       };

And kept the 'reusable' property in

        display_reserved:framebuffer@60000000 {...};

Now the kernel boots fine. I am able to get
/dev/dma_heap/framebuffer\@60000000 . :)

Quoted 9 lines

Thanks,
-Brian

Quoted 6 lines

Andrew

Quoted 3 lines
Quoted 3 lines

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#33 lore raw
In reply to Ayan Halder

++ [email protected] (Sorry, somehow I am missing your email while sending. :( )

Quoted 104 lines · Ayan Halder
Quoted 30 lines · Brian Starkey
Quoted 22 lines · John Stultz
Quoted 10 lines · Andrew F. Davis
Quoted 3 lines · John Stultz

But if the objection stands, do you have a proposal for an alternative way to enumerate a subset of CMA heaps?

When in staging ION had to reach into the CMA framework as the other direction would not be allowed, so cma_for_each_area() was added. If DMA-BUF heaps is not in staging then we can do the opposite, and have the CMA framework register heaps itself using our framework. That way the CMA system could decide what areas to export or not (maybe based on a DT property or similar).

Ok. Though the CMA core doesn't have much sense of DT details either, so it would probably have to be done in the reserved_mem logic, which doesn't feel right to me.

I'd probably guess we should have some sort of dt binding to describe a dmabuf cma heap and from that node link to a CMA node via a memory-region phandle. Along with maybe the default heap as well? Not eager to get into another binding review cycle, and I'm not sure what non-DT systems will do yet, but I'll take a shot at it and iterate.

Quoted 6 lines
Quoted 2 lines

The end result is the same so we can make this change later (it has to come after DMA-BUF heaps is in anyway).

Well, I'm hesitant to merge code that exposes all the CMA heaps and then add patches that becomes more selective, should anyone depend on the initial behavior. :/

Quoted 6 lines

How about only auto-adding the system default CMA region (cma->name == "reserved")?

And/or the CMA auto-add could be behind a config option? It seems a shame to further delay this, and the CMA heap itself really is useful.

A bit of a detour, comming back to the issue why the following node was not getting detected by the dma-buf heaps framework.

        reserved-memory {
                #address-cells = <2>;
                #size-cells = <2>;
                ranges;

                display_reserved: framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reusable; <<<<<<<<<<<<-----------This was missing in our
earlier node
                        reg = <0 0x60000000 0 0x08000000>;
                };

Quoting reserved-memory.txt :-
"The operating system can use the memory in this region with the limitation that
the device driver(s) owning the region need to be able to reclaim it back"

Thus as per my observation, without 'reusable', rmem_cma_setup() returns -EINVAL and the reserved-memory is not added as a cma region.

With 'reusable', rmem_cma_setup() succeeds , but the kernel crashes as follows :-

[    0.450562] WARNING: CPU: 2 PID: 1 at mm/cma.c:110 cma_init_reserved_areas+0xec/0x22c
[    0.458415] Modules linked in:                                                                                                             
[    0.461470] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.3.0-rc4-01377-g51dbcf03884c-dirty #15                                              
[    0.470017] Hardware name: ARM Juno development board (r0) (DT)                                                                            
[    0.475953] pstate: 80000005 (Nzcv daif -PAN -UAO)                                                                                         
[    0.480755] pc : cma_init_reserved_areas+0xec/0x22c  
[    0.485643] lr : cma_init_reserved_areas+0xe8/0x22c 

<----snip register dump --->

[    0.600646] Unable to handle kernel paging request at virtual address ffff7dffff800000
[    0.608591] Mem abort info:
[    0.611386]   ESR = 0x96000006
<---snip uninteresting bits --->
[    0.681069] pc : cma_init_reserved_areas+0x114/0x22c
[    0.686043] lr : cma_init_reserved_areas+0xe8/0x22c

I am looking into this now. My final objective is to get "/dev/dma_heap/framebuffer" (as a cma heap).
Any leads?

Quoted 17 lines

Cheers,
-Brian

Quoted 14 lines

So, <sigh>, I'll start on the rework for the CMA bits.

That said, I'm definitely wanting to make some progress on this patch series, so maybe we can still merge the core/helpers/system heap and just hold the cma heap for a rework on the enumeration bits. That way we can at least get other folks working on switching their vendor heaps from ION.

Sumit: Does that sound ok? Assuming no other objections, can you take the v11 set minus the CMA heap patch?

thanks
-john

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#34 lore raw
In reply to Ayan Halder
Quoted 61 lines · Ayan Halder
Quoted 48 lines · Ayan Halder
Quoted 39 lines · Sumit Semwal

Hello Christoph, everyone,

Quoted 28 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:
https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

Quoted 7 lines

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

Quoted 4 lines

It looks like most of the feedback has been taken care of. If there's no more objection to this series, I'd like to merge it in soon.

If there are any more review comments, may I request you to please provide them?

I tested these patches using our internal test suite with Arm,komeda driver and the following node in dts

Quoted 11 lines
        reserved-memory {
                #address-cells = <0x2>;
                #size-cells = <0x2>;
                ranges;

                framebuffer@60000000 {
                        compatible = "shared-dma-pool";
                        linux,cma-default;
                        reg = <0x0 0x60000000 0x0 0x8000000>;
                };
        }
Quoted 5 lines

Apologies for the confusion, this dts node is irrelevant as our tests were using the cma heap (via /dev/dma_heap/reserved).

That raises a question. How do we represent the reserved-memory nodes (as shown above) via the dma-buf heaps framework ?

(Apologies I didn't initially see this as you somehow left me off the reply list)

So yea, as Brian mentioned, we'll generate a heap for each cma area. So the above should generate a heap named "framebuffer".

For example, on HiKey960 the following patch adds (originally for ION, but the same dt node will work for dmabuf heaps) the cma heap "linux,cma"
https://git.linaro.org/people/john.stultz/android-dev.git/commit/?h=dev/dma-buf-heap&id=00538fe70e17acf07fdcbc441816d91cdd227207

thanks
-john

#35 lore raw

Re: [RESEND][PATCH v8 1/5] dma-buf: Add dma-buf heaps framework

In reply to John Stultz

Review

Quoted 9 lines · On Fri, 6 Sep 2019 18:47:08 +0000 John Stultz wrote:
(quoted diff) +9 -0
+/**
+ * dma_heap_get_data() - get per-heap driver data
+ * @heap: DMA-Heap to retrieve private data for
+ *
+ * Returns:
+ * The per-heap data for the heap.
+ */
+void *dma_heap_get_data(struct dma_heap *heap);
+

It will help readers more than thought understand this framework if s/get_data/get_drvdata/

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#36 lore raw

Re: [RESEND][PATCH v8 1/5] dma-buf: Add dma-buf heaps framework

Quoted 13 lines · Hillf Danton

Review

Quoted 9 lines · On Fri, 6 Sep 2019 18:47:08 +0000 John Stultz wrote:
(quoted diff) +9 -0
+/**
+ * dma_heap_get_data() - get per-heap driver data
+ * @heap: DMA-Heap to retrieve private data for
+ *
+ * Returns:
+ * The per-heap data for the heap.
+ */
+void *dma_heap_get_data(struct dma_heap *heap);
+

It will help readers more than thought understand this framework if s/get_data/get_drvdata/

Sounds good!

Thanks for the review and suggestion!
-john

#37 lore raw

Re: [RESEND][PATCH v8 3/5] dma-buf: heaps: Add system heap to dmabuf heaps

In reply to John Stultz
Quoted 10 lines · On Fri, 6 Sep 2019 18:47:09 +0000 John Stultz wrote:
(quoted diff) +9 -0
+static int system_heap_allocate(struct dma_heap *heap,
+				unsigned long len,
+				unsigned long fd_flags,
+				unsigned long heap_flags)
+{
+	struct heap_helper_buffer *helper_buffer;
+	struct dma_buf *dmabuf;
+	int ret = -ENOMEM;
+	pgoff_t pg;

Review

Quoted 10 lines
(quoted diff) +10 -0
+
+	helper_buffer = kzalloc(sizeof(*helper_buffer), GFP_KERNEL);
+	if (!helper_buffer)
+		return -ENOMEM;
+
+	init_heap_helper_buffer(helper_buffer, system_heap_free);
+	helper_buffer->flags = heap_flags;
+	helper_buffer->heap = heap;
+	helper_buffer->size = len;
+

A couple of lines looks needed to handle len if it is not PAGE_SIZE aligned.

Review

Quoted 4 lines
(quoted diff) +4 -0
+	helper_buffer->pagecount = len / PAGE_SIZE;
+	helper_buffer->pages = kmalloc_array(helper_buffer->pagecount,
+					     sizeof(*helper_buffer->pages),
+					     GFP_KERNEL);

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#38 lore raw

Re: [RESEND][PATCH v8 3/5] dma-buf: heaps: Add system heap to dmabuf heaps

Quoted 17 lines · Hillf Danton
Quoted 14 lines · On Fri, 6 Sep 2019 18:47:09 +0000 John Stultz wrote:
(quoted diff) +13 -0
+static int system_heap_allocate(struct dma_heap *heap,
+                             unsigned long len,
+                             unsigned long fd_flags,
+                             unsigned long heap_flags)
+{
+     struct heap_helper_buffer *helper_buffer;
+     struct dma_buf *dmabuf;
+     int ret = -ENOMEM;
+     pgoff_t pg;
+
+     helper_buffer = kzalloc(sizeof(*helper_buffer), GFP_KERNEL);
+     if (!helper_buffer)
+             return -ENOMEM;
Quoted 8 lines

Review

Quoted 6 lines
(quoted diff) +6 -0
+
+     init_heap_helper_buffer(helper_buffer, system_heap_free);
+     helper_buffer->flags = heap_flags;
+     helper_buffer->heap = heap;
+     helper_buffer->size = len;
+

A couple of lines looks needed to handle len if it is not PAGE_SIZE aligned.

Hey! Thanks so much for the review!

dma_heap_buffer_alloc() sets "len = PAGE_ALIGN(len);" before calling into the heap allocation hook.
So hopefully this isn't a concern, or am I missing something?

thanks
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#39 lore raw

Re: [RESEND][PATCH v8 4/5] dma-buf: heaps: Add CMA heap to dmabuf heaps

In reply to John Stultz
Quoted 10 lines · On Fri, 6 Sep 2019 18:47:09 +0000 John Stultz wrote:
(quoted diff) +9 -0
+	cma_pages = cma_alloc(cma_heap->cma, nr_pages, align, false);
+	if (!cma_pages)
+		goto free_buf;
+
+	if (PageHighMem(cma_pages)) {
+		unsigned long nr_clear_pages = nr_pages;
+		struct page *page = cma_pages;
+
+		while (nr_clear_pages > 0) {

Review

Quoted 10 lines
(quoted diff) +10 -0
+			void *vaddr = kmap_atomic(page);
+
+			memset(vaddr, 0, PAGE_SIZE);
+			kunmap_atomic(vaddr);
+			page++;
+			nr_clear_pages--;
+		}
+	} else {
+		memset(page_address(cma_pages), 0, size);
+	}

Take a breath after zeroing a page, and a peep at pending signal.

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#40 lore raw

Re: [RESEND][PATCH v8 4/5] dma-buf: heaps: Add CMA heap to dmabuf heaps

Quoted 15 lines · Hillf Danton
Quoted 14 lines · On Fri, 6 Sep 2019 18:47:09 +0000 John Stultz wrote:
(quoted diff) +13 -0
+     cma_pages = cma_alloc(cma_heap->cma, nr_pages, align, false);
+     if (!cma_pages)
+             goto free_buf;
+
+     if (PageHighMem(cma_pages)) {
+             unsigned long nr_clear_pages = nr_pages;
+             struct page *page = cma_pages;
+
+             while (nr_clear_pages > 0) {
+                     void *vaddr = kmap_atomic(page);
+
+                     memset(vaddr, 0, PAGE_SIZE);
+                     kunmap_atomic(vaddr);
Quoted 8 lines

Review

Quoted 6 lines
(quoted diff) +6 -0
+                     page++;
+                     nr_clear_pages--;
+             }
+     } else {
+             memset(page_address(cma_pages), 0, size);
+     }

Take a breath after zeroing a page, and a peep at pending signal.

Ok. Took a swing at this. It will be in the next revision.

Thanks again for the review!
-john
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Acked-by on the v8 series
#41 lore raw
In reply to John Stultz
Quoted 93 lines · John Stultz

Here is yet another pass at the dma-buf heaps patchset Andrew and I have been working on which tries to destage a fair chunk of ION functionality.

The patchset implements per-heap devices which can be opened directly and then an ioctl is used to allocate a dmabuf from the heap.

The interface is similar, but much simpler then IONs, only providing an ALLOC ioctl.

Also, I've provided relatively simple system and cma heaps.

I've booted and tested these patches with AOSP on the HiKey960 using the kernel tree here:
https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/dma-buf-heap

And the userspace changes here:

   https://android-review.googlesource.com/c/device/linaro/hikey/+/909436

Compared to ION, this patchset is missing the system-contig, carveout and chunk heaps, as I don't have a device that uses those, so I'm unable to do much useful validation there. Additionally we have no upstream users of chunk or carveout, and the system-contig has been deprecated in the common/andoid-* kernels, so this should be ok.

I've also removed the stats accounting, since any such accounting should be implemented by dma-buf core or the heaps themselves.

Most of the changes in this revision are adddressing the more concrete feedback from Christoph (many thanks!). Though I'm not sure if some of the less specific feedback was completely resolved in discussion last time around. Please let me know!

New in v8:
* Make struct dma_heap_ops consts (Suggested by Christoph)
* Add flush_kernel_vmap_range/invalidate_kernel_vmap_range calls (suggested by Christoph)
* Condense dma_heap_buffer and heap_helper_buffer (suggested by Christoph)
* Get rid of needless struct system_heap (suggested by Christoph)
* Fix indentation by using shorter argument names (suggested by Christoph)
* Remove unused private_flags value
* Add forgotten include file to fix build issue on x86
* Checkpatch whitespace fixups

Thoughts and feedback would be greatly appreciated!

thanks
-john

Cc Laura Abbott <[email protected]>
Cc Benjamin Gaignard <[email protected]>
Cc Sumit Semwal <[email protected]>
Cc Liam Mark <[email protected]>
Cc Pratik Patel <[email protected]>
Cc Brian Starkey <[email protected]>
Cc Vincent Donnefort <[email protected]>
Cc Sudipto Paul <[email protected]>
Cc Andrew F. Davis <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Chenbo Feng <[email protected]>
Cc Alistair Strachan <[email protected]>
Cc Hridya Valsaraju <[email protected]>

Andrew F. Davis (1):
dma-buf: Add dma-buf heaps framework

John Stultz (4):
dma-buf: heaps: Add heap helpers
dma-buf: heaps: Add system heap to dmabuf heaps
dma-buf: heaps: Add CMA heap to dmabuf heaps
kselftests: Add dma-heap test

  MAINTAINERS                                   |  18 ++
  drivers/dma-buf/Kconfig                       |  11 +
  drivers/dma-buf/Makefile                      |   2 +
  drivers/dma-buf/dma-heap.c                    | 250 ++++++++++++++++
  drivers/dma-buf/heaps/Kconfig                 |  14 +
  drivers/dma-buf/heaps/Makefile                |   4 +
  drivers/dma-buf/heaps/cma_heap.c              | 164 +++++++++++
  drivers/dma-buf/heaps/heap-helpers.c          | 269 ++++++++++++++++++
  drivers/dma-buf/heaps/heap-helpers.h          |  55 ++++
  drivers/dma-buf/heaps/system_heap.c           | 122 ++++++++
  include/linux/dma-heap.h                      |  59 ++++
  include/uapi/linux/dma-heap.h                 |  55 ++++
  tools/testing/selftests/dmabuf-heaps/Makefile |   9 +
  .../selftests/dmabuf-heaps/dmabuf-heap.c      | 230 +++++++++++++++
  14 files changed, 1262 insertions(+)
Quoted 12 lines

create mode 100644 drivers/dma-buf/dma-heap.c
create mode 100644 drivers/dma-buf/heaps/Kconfig
create mode 100644 drivers/dma-buf/heaps/Makefile
create mode 100644 drivers/dma-buf/heaps/cma_heap.c
create mode 100644 drivers/dma-buf/heaps/heap-helpers.c create mode 100644 drivers/dma-buf/heaps/heap-helpers.h create mode 100644 drivers/dma-buf/heaps/system_heap.c create mode 100644 include/linux/dma-heap.h
create mode 100644 include/uapi/linux/dma-heap.h
create mode 100644 tools/testing/selftests/dmabuf-heaps/Makefile create mode 100644 tools/testing/selftests/dmabuf-heaps/dmabuf-heap.c

I've seen a couple of details that need to be fixed and can be fixed fairly easily but as far as the overall design goes it looks good. Once those are fixed up, you can add