diff mbox series

[v3,21/43] drm/fbdev-dma: Implement damage handling and deferred I/O

Message ID 20240419083331.7761-22-tzimmermann@suse.de (mailing list archive)
State Handled Elsewhere
Headers show
Series drm: Provide fbdev emulation per memory manager | expand

Commit Message

Thomas Zimmermann April 19, 2024, 8:29 a.m. UTC 
Add support for damage handling and deferred I/O to fbdev-dma. This
enables fbdev-dma to support all DMA-memory-based DRM drivers, even
such with a dirty callback in their framebuffers.

The patch adds the code for deferred I/O and also sets a dedicated
helper for struct fb_ops.fb_mmap that support coherent mappings.

v3:
- init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
---
 drivers/gpu/drm/drm_fbdev_dma.c | 65 ++++++++++++++++++++++++++-------
 1 file changed, 51 insertions(+), 14 deletions(-)

Comments

Linus Walleij Sept. 3, 2024, 10:53 p.m. UTC | #1 
On Fri, Apr 19, 2024 at 10:35 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:

> Add support for damage handling and deferred I/O to fbdev-dma. This
> enables fbdev-dma to support all DMA-memory-based DRM drivers, even
> such with a dirty callback in their framebuffers.
>
> The patch adds the code for deferred I/O and also sets a dedicated
> helper for struct fb_ops.fb_mmap that support coherent mappings.
>
> v3:
> - init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)
>
> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>

For some reason this makes the Versatile Express crash in QEMU
(I can retest on real hardware if necessary):

8<--- cut here ---
Unable to handle kernel paging request at virtual address 18095000 when write
[18095000] *pgd=00000000
Internal error: Oops: 805 [#1] SMP ARM
CPU: 0 PID: 75 Comm: S08_ledflash Not tainted 6.9.0-rc6+ #44
Hardware name: ARM-Versatile Express
PC is at mmioset+0x34/0xac
LR is at 0x0
pc : [<809c2a54>]    lr : [<00000000>]    psr: 20000013
sp : f8dddc38  ip : 18095000  fp : 00000000
r10: 81109a18  r9 : 81238b04  r8 : 00000000
r7 : 00440dc0  r6 : ed4f32a0  r5 : ed4f32a0  r4 : 00000001
r3 : 00000000  r2 : 00000fc0  r1 : 00000000  r0 : 18095000
Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
Control: 10c5387d  Table: 81d6406a  DAC: 00000051
Register r0 information: non-paged memory
Register r1 information: NULL pointer
Register r2 information: non-paged memory
Register r3 information: NULL pointer
Register r4 information: non-paged memory
Register r5 information: non-slab/vmalloc memory
Register r6 information: non-slab/vmalloc memory
Register r7 information: non-paged memory
Register r8 information: NULL pointer
Register r9 information: non-slab/vmalloc memory
Register r10 information: non-slab/vmalloc memory
Register r11 information: NULL pointer
Register r12 information: non-paged memory
Process S08_ledflash (pid: 75, stack limit = 0x(ptrval))
Stack: (0xf8dddc38 to 0xf8dde000)
dc20:                                                       00000801 802af888
dc40: 80a0fb80 00440dc0 00000801 811fad80 00000000 f8dddd28 ed4f32a0 811fa680
dc60: 00000000 802b0c30 ee1d2140 ee1d215c 95ff4385 00000001 00000001 00000001
dc80: 00000000 00000000 ffffffff 8196a70b 000000b0 00000064 0000006c 00000000
dca0: 00000001 00000002 00000001 811fa798 00000801 811fa940 80a0fb80 00000cc0
dcc0: 00000001 00000001 00000000 f8dddd28 811fa93c 811fa680 00000002 802b0c30
dce0: ee1d2140 00440dc0 00000000 00000000 00000001 00000000 811fad80 00440dc0
dd00: 00000001 00000000 00000000 83201100 00440dc0 00000000 00000000 802afa18
dd20: 00000bdc 00000bd8 811fad80 00000000 811fad80 00000000 00000000 00000000
dd40: adf00ec5 816d5fa8 816d5fa8 821c3c00 00000000 7ebc2000 7ebe3000 00000000
dd60: f8ddde44 8028cfa0 816d5fa8 8288df50 8169cf00 821c3c00 7ebc2000 7ebe3000
dd80: f8ddde44 8028e1e8 00000001 80bbe932 f8ddde44 f8dddde8 f8ddde44 8288df50
dda0: 83201100 809cf474 f8ddde44 8169cf00 00000020 8169cf94 821c3c94 7ebe3000
ddc0: 7ebe2fff fffffff4 00000000 816d5fa8 821c3c00 8169cf00 81d65fa8 809cdbb0
dde0: 8288df50 81ffa910 7ebe3000 82890b00 00000000 00000000 00000000 00000000
de00: 00000000 00000000 adf00ec5 00000000 00000000 8169cf00 8288df50 83201100
de20: 821c3c00 81ffa910 f8ddde44 8011f6e4 00000000 00000000 821c3c7c 8169cf7c
de40: 83200880 821c3c40 7ebc2000 7ebe2fff 82890b0c 76ea4000 ffffffff 00000000
de60: 00000000 0f000a01 f8ddde68 f8ddde68 adf00ec5 00000000 00000000 00000000
de80: 00000000 01200000 83201100 00000000 00000000 8011e2ac 00000000 83200d7c
dea0: 83201020 83200e70 83200e80 00000000 f8dddf30 83200880 00000000 00000000
dec0: 00000000 00000000 00000000 82182f38 adf00ec5 00000000 01200000 f8dddf30
dee0: 00000000 00000000 00000000 00000078 000cb21c 8011fe24 00000020 00000000
df00: 828a5d80 80a10140 adf00ec5 01200011 00000000 00000000 00000000 f8dddf30
df20: 00000000 00000078 000cb21c 801203d4 01200000 00000000 00000000 76f46468
df40: 00000000 00000000 00000011 00000000 00000000 00000000 00000000 00000000
df60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
df80: 00000003 adf00ec5 76f46468 00000000 76f468c0 00000078 80100290 83201100
dfa0: 000cb21c 80100060 76f46468 00000000 01200011 00000000 00000000 00000000
dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
dfe0: 00000078 7ebe2938 76e0854d 76dae7e6 20000030 01200011 00000000 00000000
Call trace:
 mmioset from prep_new_page+0x160/0x194
 prep_new_page from get_page_from_freelist+0x10f4/0x110c
 get_page_from_freelist from __alloc_pages+0x15c/0x280
 __alloc_pages from __pte_alloc+0x28/0x1bc
 __pte_alloc from copy_page_range+0xc10/0xd28
 copy_page_range from copy_mm+0x5cc/0x850
 copy_mm from copy_process+0x508/0xd8c
 copy_process from kernel_clone+0x94/0x338
 kernel_clone from sys_clone+0x94/0xb0
 sys_clone from ret_fast_syscall+0x0/0x1c
Exception stack(0xf8dddfa8 to 0xf8dddff0)
dfa0:                   76f46468 00000000 01200011 00000000 00000000 00000000
dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
dfe0: 00000078 7ebe2938 76e0854d 76dae7e6
Code: e92d4100 e1a08001 e1a0e003 e2522040 (a8ac410a)
---[ end trace 0000000000000000 ]---

I bisected down to this commit and reverting the commit solves the issue.

What is special about the Versatile Express graphics is that it uses a special
dedicated video RAM, looks like this in the device tree


        reserved-memory {
                #address-cells = <1>;
                #size-cells = <1>;
                ranges;

                /* Chipselect 3 is physically at 0x4c000000 */
                vram: vram@4c000000 {
                        /* 8 MB of designated video RAM */
                        compatible = "shared-dma-pool";
                        reg = <0x4c000000 0x00800000>;
                        no-map;
                };
        };
(...)
         clcd@1f000 {
                      compatible = "arm,pl111", "arm,primecell";
(...)
                      memory-region = <&vram>;

This gets picked up in the driver
drivers/gpu/drm/pl111/pl111_drv.c
like this:

        ret = of_reserved_mem_device_init(dev);
        if (!ret) {
                dev_info(dev, "using device-specific reserved memory\n");
                priv->use_device_memory = true;
        }

Yours,
Linus Walleij
Alexander Stein Sept. 4, 2024, 6:09 a.m. UTC | #2 
Hi Linus,

Am Mittwoch, 4. September 2024, 00:53:42 CEST schrieb Linus Walleij:
> On Fri, Apr 19, 2024 at 10:35 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:
> 
> > Add support for damage handling and deferred I/O to fbdev-dma. This
> > enables fbdev-dma to support all DMA-memory-based DRM drivers, even
> > such with a dirty callback in their framebuffers.
> >
> > The patch adds the code for deferred I/O and also sets a dedicated
> > helper for struct fb_ops.fb_mmap that support coherent mappings.
> >
> > v3:
> > - init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)
> >
> > Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
> > Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
> 
> For some reason this makes the Versatile Express crash in QEMU
> (I can retest on real hardware if necessary):

This commit also breaks i.MX6 on actual hardware, see [1].

[1] https://lore.kernel.org/all/23636953.6Emhk5qWAg@steina-w/

> 8<--- cut here ---
> Unable to handle kernel paging request at virtual address 18095000 when write
> [18095000] *pgd=00000000
> Internal error: Oops: 805 [#1] SMP ARM
> CPU: 0 PID: 75 Comm: S08_ledflash Not tainted 6.9.0-rc6+ #44
> Hardware name: ARM-Versatile Express
> PC is at mmioset+0x34/0xac
> LR is at 0x0
> pc : [<809c2a54>]    lr : [<00000000>]    psr: 20000013
> sp : f8dddc38  ip : 18095000  fp : 00000000
> r10: 81109a18  r9 : 81238b04  r8 : 00000000
> r7 : 00440dc0  r6 : ed4f32a0  r5 : ed4f32a0  r4 : 00000001
> r3 : 00000000  r2 : 00000fc0  r1 : 00000000  r0 : 18095000
> Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
> Control: 10c5387d  Table: 81d6406a  DAC: 00000051
> Register r0 information: non-paged memory
> Register r1 information: NULL pointer
> Register r2 information: non-paged memory
> Register r3 information: NULL pointer
> Register r4 information: non-paged memory
> Register r5 information: non-slab/vmalloc memory
> Register r6 information: non-slab/vmalloc memory
> Register r7 information: non-paged memory
> Register r8 information: NULL pointer
> Register r9 information: non-slab/vmalloc memory
> Register r10 information: non-slab/vmalloc memory
> Register r11 information: NULL pointer
> Register r12 information: non-paged memory
> Process S08_ledflash (pid: 75, stack limit = 0x(ptrval))
> Stack: (0xf8dddc38 to 0xf8dde000)
> dc20:                                                       00000801 802af888
> dc40: 80a0fb80 00440dc0 00000801 811fad80 00000000 f8dddd28 ed4f32a0 811fa680
> dc60: 00000000 802b0c30 ee1d2140 ee1d215c 95ff4385 00000001 00000001 00000001
> dc80: 00000000 00000000 ffffffff 8196a70b 000000b0 00000064 0000006c 00000000
> dca0: 00000001 00000002 00000001 811fa798 00000801 811fa940 80a0fb80 00000cc0
> dcc0: 00000001 00000001 00000000 f8dddd28 811fa93c 811fa680 00000002 802b0c30
> dce0: ee1d2140 00440dc0 00000000 00000000 00000001 00000000 811fad80 00440dc0
> dd00: 00000001 00000000 00000000 83201100 00440dc0 00000000 00000000 802afa18
> dd20: 00000bdc 00000bd8 811fad80 00000000 811fad80 00000000 00000000 00000000
> dd40: adf00ec5 816d5fa8 816d5fa8 821c3c00 00000000 7ebc2000 7ebe3000 00000000
> dd60: f8ddde44 8028cfa0 816d5fa8 8288df50 8169cf00 821c3c00 7ebc2000 7ebe3000
> dd80: f8ddde44 8028e1e8 00000001 80bbe932 f8ddde44 f8dddde8 f8ddde44 8288df50
> dda0: 83201100 809cf474 f8ddde44 8169cf00 00000020 8169cf94 821c3c94 7ebe3000
> ddc0: 7ebe2fff fffffff4 00000000 816d5fa8 821c3c00 8169cf00 81d65fa8 809cdbb0
> dde0: 8288df50 81ffa910 7ebe3000 82890b00 00000000 00000000 00000000 00000000
> de00: 00000000 00000000 adf00ec5 00000000 00000000 8169cf00 8288df50 83201100
> de20: 821c3c00 81ffa910 f8ddde44 8011f6e4 00000000 00000000 821c3c7c 8169cf7c
> de40: 83200880 821c3c40 7ebc2000 7ebe2fff 82890b0c 76ea4000 ffffffff 00000000
> de60: 00000000 0f000a01 f8ddde68 f8ddde68 adf00ec5 00000000 00000000 00000000
> de80: 00000000 01200000 83201100 00000000 00000000 8011e2ac 00000000 83200d7c
> dea0: 83201020 83200e70 83200e80 00000000 f8dddf30 83200880 00000000 00000000
> dec0: 00000000 00000000 00000000 82182f38 adf00ec5 00000000 01200000 f8dddf30
> dee0: 00000000 00000000 00000000 00000078 000cb21c 8011fe24 00000020 00000000
> df00: 828a5d80 80a10140 adf00ec5 01200011 00000000 00000000 00000000 f8dddf30
> df20: 00000000 00000078 000cb21c 801203d4 01200000 00000000 00000000 76f46468
> df40: 00000000 00000000 00000011 00000000 00000000 00000000 00000000 00000000
> df60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
> df80: 00000003 adf00ec5 76f46468 00000000 76f468c0 00000078 80100290 83201100
> dfa0: 000cb21c 80100060 76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6 20000030 01200011 00000000 00000000
> Call trace:
>  mmioset from prep_new_page+0x160/0x194
>  prep_new_page from get_page_from_freelist+0x10f4/0x110c
>  get_page_from_freelist from __alloc_pages+0x15c/0x280
>  __alloc_pages from __pte_alloc+0x28/0x1bc
>  __pte_alloc from copy_page_range+0xc10/0xd28
>  copy_page_range from copy_mm+0x5cc/0x850
>  copy_mm from copy_process+0x508/0xd8c
>  copy_process from kernel_clone+0x94/0x338
>  kernel_clone from sys_clone+0x94/0xb0
>  sys_clone from ret_fast_syscall+0x0/0x1c
> Exception stack(0xf8dddfa8 to 0xf8dddff0)
> dfa0:                   76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6
> Code: e92d4100 e1a08001 e1a0e003 e2522040 (a8ac410a)
> ---[ end trace 0000000000000000 ]---
> 
> I bisected down to this commit and reverting the commit solves the issue.
> 
> What is special about the Versatile Express graphics is that it uses a special
> dedicated video RAM, looks like this in the device tree
> 
> 
>         reserved-memory {
>                 #address-cells = <1>;
>                 #size-cells = <1>;
>                 ranges;
> 
>                 /* Chipselect 3 is physically at 0x4c000000 */
>                 vram: vram@4c000000 {
>                         /* 8 MB of designated video RAM */
>                         compatible = "shared-dma-pool";
>                         reg = <0x4c000000 0x00800000>;

Can you please check in which memory zone this VRAM is located. In my case
it's important CMA is located in Normal zone to trigger this problem.

Best regards,
Alexander

>                         no-map;
>                 };
>         };
> (...)
>          clcd@1f000 {
>                       compatible = "arm,pl111", "arm,primecell";
> (...)
>                       memory-region = <&vram>;
> 
> This gets picked up in the driver
> drivers/gpu/drm/pl111/pl111_drv.c
> like this:
> 
>         ret = of_reserved_mem_device_init(dev);
>         if (!ret) {
>                 dev_info(dev, "using device-specific reserved memory\n");
>                 priv->use_device_memory = true;
>         }
> 
> Yours,
> Linus Walleij
>
Linus Walleij Sept. 4, 2024, 6:36 a.m. UTC | #3 
On Wed, Sep 4, 2024 at 8:09 AM Alexander Stein
<alexander.stein@ew.tq-group.com> wrote:

> Can you please check in which memory zone this VRAM is located. In my case
> it's important CMA is located in Normal zone to trigger this problem.

dmesg says this:

Reserved memory: created DMA memory pool at 0x18000000, size 8 MiB
OF: reserved mem: initialized node vram@18000000, compatible id shared-dma-pool
OF: reserved mem: 0x18000000..0x187fffff (8192 KiB) nomap non-reusable
vram@18000000
cma: Reserved 16 MiB at 0xef000000 on node -1
Zone ranges:
  Normal   [mem 0x0000000080000000-0x00000000efffffff]
Movable zone start for each node
Early memory node ranges
  node   0: [mem 0x0000000080000000-0x00000000efffffff]
Initmem setup node 0 [mem 0x0000000080000000-0x00000000efffffff]
percpu: Embedded 15 pages/cpu s39052 r0 d22388 u61440

So at the top of the normal zone at 0xef000000 as I read it.

Yours,
Linus Walleij
Alexander Stein Sept. 4, 2024, 6:43 a.m. UTC | #4 
Hi Linus,

Am Mittwoch, 4. September 2024, 08:36:46 CEST schrieb Linus Walleij:
> On Wed, Sep 4, 2024 at 8:09 AM Alexander Stein
> <alexander.stein@ew.tq-group.com> wrote:
> 
> > Can you please check in which memory zone this VRAM is located. In my case
> > it's important CMA is located in Normal zone to trigger this problem.
> 
> dmesg says this:
> 
> Reserved memory: created DMA memory pool at 0x18000000, size 8 MiB
> OF: reserved mem: initialized node vram@18000000, compatible id shared-dma-pool
> OF: reserved mem: 0x18000000..0x187fffff (8192 KiB) nomap non-reusable
> vram@18000000
> cma: Reserved 16 MiB at 0xef000000 on node -1
> Zone ranges:
>   Normal   [mem 0x0000000080000000-0x00000000efffffff]
> Movable zone start for each node
> Early memory node ranges
>   node   0: [mem 0x0000000080000000-0x00000000efffffff]
> Initmem setup node 0 [mem 0x0000000080000000-0x00000000efffffff]
> percpu: Embedded 15 pages/cpu s39052 r0 d22388 u61440
> 
> So at the top of the normal zone at 0xef000000 as I read it.

Okay, so it is the same as for me. It seems you don't have any HighMem.
On a different model having more memory there is also HighMem available:

[    0.000000] Zone ranges:
[    0.000000]   Normal   [mem 0x0000000010000000-0x000000003fffffff]
[    0.000000]   HighMem  [mem 0x0000000040000000-0x000000004fffffff]

If I explicitly configure CMA in HighMem I don't have any issues.
But the 512MiB models don't have enough memory for fixed allocation in HighMem.

I was about to send a revert myself as I didn't get any feedback yet.

Best regards,
Alexander
Thomas Zimmermann Sept. 4, 2024, 7:04 a.m. UTC | #5 
Hi Linus

Am 04.09.24 um 00:53 schrieb Linus Walleij:
> On Fri, Apr 19, 2024 at 10:35 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:
>
>> Add support for damage handling and deferred I/O to fbdev-dma. This
>> enables fbdev-dma to support all DMA-memory-based DRM drivers, even
>> such with a dirty callback in their framebuffers.
>>
>> The patch adds the code for deferred I/O and also sets a dedicated
>> helper for struct fb_ops.fb_mmap that support coherent mappings.
>>
>> v3:
>> - init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)
>>
>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
>> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
> For some reason this makes the Versatile Express crash in QEMU
> (I can retest on real hardware if necessary):
>
> 8<--- cut here ---
> Unable to handle kernel paging request at virtual address 18095000 when write
> [18095000] *pgd=00000000
> Internal error: Oops: 805 [#1] SMP ARM
> CPU: 0 PID: 75 Comm: S08_ledflash Not tainted 6.9.0-rc6+ #44
> Hardware name: ARM-Versatile Express
> PC is at mmioset+0x34/0xac
> LR is at 0x0
> pc : [<809c2a54>]    lr : [<00000000>]    psr: 20000013
> sp : f8dddc38  ip : 18095000  fp : 00000000
> r10: 81109a18  r9 : 81238b04  r8 : 00000000
> r7 : 00440dc0  r6 : ed4f32a0  r5 : ed4f32a0  r4 : 00000001
> r3 : 00000000  r2 : 00000fc0  r1 : 00000000  r0 : 18095000
> Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
> Control: 10c5387d  Table: 81d6406a  DAC: 00000051
> Register r0 information: non-paged memory
> Register r1 information: NULL pointer
> Register r2 information: non-paged memory
> Register r3 information: NULL pointer
> Register r4 information: non-paged memory
> Register r5 information: non-slab/vmalloc memory
> Register r6 information: non-slab/vmalloc memory
> Register r7 information: non-paged memory
> Register r8 information: NULL pointer
> Register r9 information: non-slab/vmalloc memory
> Register r10 information: non-slab/vmalloc memory
> Register r11 information: NULL pointer
> Register r12 information: non-paged memory
> Process S08_ledflash (pid: 75, stack limit = 0x(ptrval))
> Stack: (0xf8dddc38 to 0xf8dde000)
> dc20:                                                       00000801 802af888
> dc40: 80a0fb80 00440dc0 00000801 811fad80 00000000 f8dddd28 ed4f32a0 811fa680
> dc60: 00000000 802b0c30 ee1d2140 ee1d215c 95ff4385 00000001 00000001 00000001
> dc80: 00000000 00000000 ffffffff 8196a70b 000000b0 00000064 0000006c 00000000
> dca0: 00000001 00000002 00000001 811fa798 00000801 811fa940 80a0fb80 00000cc0
> dcc0: 00000001 00000001 00000000 f8dddd28 811fa93c 811fa680 00000002 802b0c30
> dce0: ee1d2140 00440dc0 00000000 00000000 00000001 00000000 811fad80 00440dc0
> dd00: 00000001 00000000 00000000 83201100 00440dc0 00000000 00000000 802afa18
> dd20: 00000bdc 00000bd8 811fad80 00000000 811fad80 00000000 00000000 00000000
> dd40: adf00ec5 816d5fa8 816d5fa8 821c3c00 00000000 7ebc2000 7ebe3000 00000000
> dd60: f8ddde44 8028cfa0 816d5fa8 8288df50 8169cf00 821c3c00 7ebc2000 7ebe3000
> dd80: f8ddde44 8028e1e8 00000001 80bbe932 f8ddde44 f8dddde8 f8ddde44 8288df50
> dda0: 83201100 809cf474 f8ddde44 8169cf00 00000020 8169cf94 821c3c94 7ebe3000
> ddc0: 7ebe2fff fffffff4 00000000 816d5fa8 821c3c00 8169cf00 81d65fa8 809cdbb0
> dde0: 8288df50 81ffa910 7ebe3000 82890b00 00000000 00000000 00000000 00000000
> de00: 00000000 00000000 adf00ec5 00000000 00000000 8169cf00 8288df50 83201100
> de20: 821c3c00 81ffa910 f8ddde44 8011f6e4 00000000 00000000 821c3c7c 8169cf7c
> de40: 83200880 821c3c40 7ebc2000 7ebe2fff 82890b0c 76ea4000 ffffffff 00000000
> de60: 00000000 0f000a01 f8ddde68 f8ddde68 adf00ec5 00000000 00000000 00000000
> de80: 00000000 01200000 83201100 00000000 00000000 8011e2ac 00000000 83200d7c
> dea0: 83201020 83200e70 83200e80 00000000 f8dddf30 83200880 00000000 00000000
> dec0: 00000000 00000000 00000000 82182f38 adf00ec5 00000000 01200000 f8dddf30
> dee0: 00000000 00000000 00000000 00000078 000cb21c 8011fe24 00000020 00000000
> df00: 828a5d80 80a10140 adf00ec5 01200011 00000000 00000000 00000000 f8dddf30
> df20: 00000000 00000078 000cb21c 801203d4 01200000 00000000 00000000 76f46468
> df40: 00000000 00000000 00000011 00000000 00000000 00000000 00000000 00000000
> df60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
> df80: 00000003 adf00ec5 76f46468 00000000 76f468c0 00000078 80100290 83201100
> dfa0: 000cb21c 80100060 76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6 20000030 01200011 00000000 00000000
> Call trace:
>   mmioset from prep_new_page+0x160/0x194
>   prep_new_page from get_page_from_freelist+0x10f4/0x110c
>   get_page_from_freelist from __alloc_pages+0x15c/0x280
>   __alloc_pages from __pte_alloc+0x28/0x1bc
>   __pte_alloc from copy_page_range+0xc10/0xd28
>   copy_page_range from copy_mm+0x5cc/0x850
>   copy_mm from copy_process+0x508/0xd8c
>   copy_process from kernel_clone+0x94/0x338
>   kernel_clone from sys_clone+0x94/0xb0
>   sys_clone from ret_fast_syscall+0x0/0x1c
> Exception stack(0xf8dddfa8 to 0xf8dddff0)
> dfa0:                   76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6
> Code: e92d4100 e1a08001 e1a0e003 e2522040 (a8ac410a)
> ---[ end trace 0000000000000000 ]---
>
> I bisected down to this commit and reverting the commit solves the issue.

Thanks for bisecting. Simply reverting isn't an option as that would 
break many other drivers.

Is it only the pl111 driver? If nothing else works, we could give it a 
dedicated fbdev emulation.

Best regards
Thomas

>
> What is special about the Versatile Express graphics is that it uses a special
> dedicated video RAM, looks like this in the device tree
>
>
>          reserved-memory {
>                  #address-cells = <1>;
>                  #size-cells = <1>;
>                  ranges;
>
>                  /* Chipselect 3 is physically at 0x4c000000 */
>                  vram: vram@4c000000 {
>                          /* 8 MB of designated video RAM */
>                          compatible = "shared-dma-pool";
>                          reg = <0x4c000000 0x00800000>;
>                          no-map;
>                  };
>          };
> (...)
>           clcd@1f000 {
>                        compatible = "arm,pl111", "arm,primecell";
> (...)
>                        memory-region = <&vram>;
>
> This gets picked up in the driver
> drivers/gpu/drm/pl111/pl111_drv.c
> like this:
>
>          ret = of_reserved_mem_device_init(dev);
>          if (!ret) {
>                  dev_info(dev, "using device-specific reserved memory\n");
>                  priv->use_device_memory = true;
>          }
>
> Yours,
> Linus Walleij
Thomas Zimmermann Sept. 4, 2024, 7:08 a.m. UTC | #6 
Hi

Am 04.09.24 um 08:43 schrieb Alexander Stein:
> Hi Linus,
>
> Am Mittwoch, 4. September 2024, 08:36:46 CEST schrieb Linus Walleij:
>> On Wed, Sep 4, 2024 at 8:09 AM Alexander Stein
>> <alexander.stein@ew.tq-group.com> wrote:
>>
>>> Can you please check in which memory zone this VRAM is located. In my case
>>> it's important CMA is located in Normal zone to trigger this problem.
>> dmesg says this:
>>
>> Reserved memory: created DMA memory pool at 0x18000000, size 8 MiB
>> OF: reserved mem: initialized node vram@18000000, compatible id shared-dma-pool
>> OF: reserved mem: 0x18000000..0x187fffff (8192 KiB) nomap non-reusable
>> vram@18000000
>> cma: Reserved 16 MiB at 0xef000000 on node -1
>> Zone ranges:
>>    Normal   [mem 0x0000000080000000-0x00000000efffffff]
>> Movable zone start for each node
>> Early memory node ranges
>>    node   0: [mem 0x0000000080000000-0x00000000efffffff]
>> Initmem setup node 0 [mem 0x0000000080000000-0x00000000efffffff]
>> percpu: Embedded 15 pages/cpu s39052 r0 d22388 u61440
>>
>> So at the top of the normal zone at 0xef000000 as I read it.
> Okay, so it is the same as for me. It seems you don't have any HighMem.
> On a different model having more memory there is also HighMem available:
>
> [    0.000000] Zone ranges:
> [    0.000000]   Normal   [mem 0x0000000010000000-0x000000003fffffff]
> [    0.000000]   HighMem  [mem 0x0000000040000000-0x000000004fffffff]

IIRC vram needs to be located in HighMem for the code to work. What is 
the difference between these zones?

Best regards
Thomas

>
> If I explicitly configure CMA in HighMem I don't have any issues.
> But the 512MiB models don't have enough memory for fixed allocation in HighMem.
>
> I was about to send a revert myself as I didn't get any feedback yet.
>
> Best regards,
> Alexander
Thomas Zimmermann Sept. 4, 2024, 7:41 a.m. UTC | #7 
Hi Linus, Alexander,

Attached is a patch to mitigate the problem and go back to the old 
behavior for pl111. Can you please test it on affected and unaffected 
systems and report the results?

Best regards
Thomas

Am 04.09.24 um 00:53 schrieb Linus Walleij:
> On Fri, Apr 19, 2024 at 10:35 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:
>
>> Add support for damage handling and deferred I/O to fbdev-dma. This
>> enables fbdev-dma to support all DMA-memory-based DRM drivers, even
>> such with a dirty callback in their framebuffers.
>>
>> The patch adds the code for deferred I/O and also sets a dedicated
>> helper for struct fb_ops.fb_mmap that support coherent mappings.
>>
>> v3:
>> - init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)
>>
>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
>> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
> For some reason this makes the Versatile Express crash in QEMU
> (I can retest on real hardware if necessary):
>
> 8<--- cut here ---
> Unable to handle kernel paging request at virtual address 18095000 when write
> [18095000] *pgd=00000000
> Internal error: Oops: 805 [#1] SMP ARM
> CPU: 0 PID: 75 Comm: S08_ledflash Not tainted 6.9.0-rc6+ #44
> Hardware name: ARM-Versatile Express
> PC is at mmioset+0x34/0xac
> LR is at 0x0
> pc : [<809c2a54>]    lr : [<00000000>]    psr: 20000013
> sp : f8dddc38  ip : 18095000  fp : 00000000
> r10: 81109a18  r9 : 81238b04  r8 : 00000000
> r7 : 00440dc0  r6 : ed4f32a0  r5 : ed4f32a0  r4 : 00000001
> r3 : 00000000  r2 : 00000fc0  r1 : 00000000  r0 : 18095000
> Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
> Control: 10c5387d  Table: 81d6406a  DAC: 00000051
> Register r0 information: non-paged memory
> Register r1 information: NULL pointer
> Register r2 information: non-paged memory
> Register r3 information: NULL pointer
> Register r4 information: non-paged memory
> Register r5 information: non-slab/vmalloc memory
> Register r6 information: non-slab/vmalloc memory
> Register r7 information: non-paged memory
> Register r8 information: NULL pointer
> Register r9 information: non-slab/vmalloc memory
> Register r10 information: non-slab/vmalloc memory
> Register r11 information: NULL pointer
> Register r12 information: non-paged memory
> Process S08_ledflash (pid: 75, stack limit = 0x(ptrval))
> Stack: (0xf8dddc38 to 0xf8dde000)
> dc20:                                                       00000801 802af888
> dc40: 80a0fb80 00440dc0 00000801 811fad80 00000000 f8dddd28 ed4f32a0 811fa680
> dc60: 00000000 802b0c30 ee1d2140 ee1d215c 95ff4385 00000001 00000001 00000001
> dc80: 00000000 00000000 ffffffff 8196a70b 000000b0 00000064 0000006c 00000000
> dca0: 00000001 00000002 00000001 811fa798 00000801 811fa940 80a0fb80 00000cc0
> dcc0: 00000001 00000001 00000000 f8dddd28 811fa93c 811fa680 00000002 802b0c30
> dce0: ee1d2140 00440dc0 00000000 00000000 00000001 00000000 811fad80 00440dc0
> dd00: 00000001 00000000 00000000 83201100 00440dc0 00000000 00000000 802afa18
> dd20: 00000bdc 00000bd8 811fad80 00000000 811fad80 00000000 00000000 00000000
> dd40: adf00ec5 816d5fa8 816d5fa8 821c3c00 00000000 7ebc2000 7ebe3000 00000000
> dd60: f8ddde44 8028cfa0 816d5fa8 8288df50 8169cf00 821c3c00 7ebc2000 7ebe3000
> dd80: f8ddde44 8028e1e8 00000001 80bbe932 f8ddde44 f8dddde8 f8ddde44 8288df50
> dda0: 83201100 809cf474 f8ddde44 8169cf00 00000020 8169cf94 821c3c94 7ebe3000
> ddc0: 7ebe2fff fffffff4 00000000 816d5fa8 821c3c00 8169cf00 81d65fa8 809cdbb0
> dde0: 8288df50 81ffa910 7ebe3000 82890b00 00000000 00000000 00000000 00000000
> de00: 00000000 00000000 adf00ec5 00000000 00000000 8169cf00 8288df50 83201100
> de20: 821c3c00 81ffa910 f8ddde44 8011f6e4 00000000 00000000 821c3c7c 8169cf7c
> de40: 83200880 821c3c40 7ebc2000 7ebe2fff 82890b0c 76ea4000 ffffffff 00000000
> de60: 00000000 0f000a01 f8ddde68 f8ddde68 adf00ec5 00000000 00000000 00000000
> de80: 00000000 01200000 83201100 00000000 00000000 8011e2ac 00000000 83200d7c
> dea0: 83201020 83200e70 83200e80 00000000 f8dddf30 83200880 00000000 00000000
> dec0: 00000000 00000000 00000000 82182f38 adf00ec5 00000000 01200000 f8dddf30
> dee0: 00000000 00000000 00000000 00000078 000cb21c 8011fe24 00000020 00000000
> df00: 828a5d80 80a10140 adf00ec5 01200011 00000000 00000000 00000000 f8dddf30
> df20: 00000000 00000078 000cb21c 801203d4 01200000 00000000 00000000 76f46468
> df40: 00000000 00000000 00000011 00000000 00000000 00000000 00000000 00000000
> df60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
> df80: 00000003 adf00ec5 76f46468 00000000 76f468c0 00000078 80100290 83201100
> dfa0: 000cb21c 80100060 76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6 20000030 01200011 00000000 00000000
> Call trace:
>   mmioset from prep_new_page+0x160/0x194
>   prep_new_page from get_page_from_freelist+0x10f4/0x110c
>   get_page_from_freelist from __alloc_pages+0x15c/0x280
>   __alloc_pages from __pte_alloc+0x28/0x1bc
>   __pte_alloc from copy_page_range+0xc10/0xd28
>   copy_page_range from copy_mm+0x5cc/0x850
>   copy_mm from copy_process+0x508/0xd8c
>   copy_process from kernel_clone+0x94/0x338
>   kernel_clone from sys_clone+0x94/0xb0
>   sys_clone from ret_fast_syscall+0x0/0x1c
> Exception stack(0xf8dddfa8 to 0xf8dddff0)
> dfa0:                   76f46468 00000000 01200011 00000000 00000000 00000000
> dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> dfe0: 00000078 7ebe2938 76e0854d 76dae7e6
> Code: e92d4100 e1a08001 e1a0e003 e2522040 (a8ac410a)
> ---[ end trace 0000000000000000 ]---
>
> I bisected down to this commit and reverting the commit solves the issue.
>
> What is special about the Versatile Express graphics is that it uses a special
> dedicated video RAM, looks like this in the device tree
>
>
>          reserved-memory {
>                  #address-cells = <1>;
>                  #size-cells = <1>;
>                  ranges;
>
>                  /* Chipselect 3 is physically at 0x4c000000 */
>                  vram: vram@4c000000 {
>                          /* 8 MB of designated video RAM */
>                          compatible = "shared-dma-pool";
>                          reg = <0x4c000000 0x00800000>;
>                          no-map;
>                  };
>          };
> (...)
>           clcd@1f000 {
>                        compatible = "arm,pl111", "arm,primecell";
> (...)
>                        memory-region = <&vram>;
>
> This gets picked up in the driver
> drivers/gpu/drm/pl111/pl111_drv.c
> like this:
>
>          ret = of_reserved_mem_device_init(dev);
>          if (!ret) {
>                  dev_info(dev, "using device-specific reserved memory\n");
>                  priv->use_device_memory = true;
>          }
>
> Yours,
> Linus Walleij
Alexander Stein Sept. 4, 2024, 8:29 a.m. UTC | #8 
Hi Thomas,

Am Mittwoch, 4. September 2024, 09:41:12 CEST schrieb Thomas Zimmermann:
> Hi Linus, Alexander,
> 
> Attached is a patch to mitigate the problem and go back to the old 
> behavior for pl111. Can you please test it on affected and unaffected 
> systems and report the results?

Thank you for the patch. This indeed fixes the mmap problem on my i.MX6
platform which only has Normal memory:

[    0.000000] Zone ranges:
[    0.000000]   Normal   [mem 0x0000000010000000-0x000000002fffffff]
[    0.000000]   HighMem  empty

But please note this is not pl111 based but IPUv3 (drivers/gpu/ipu-v3).
You might want to adjust the commit message.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>

Best regards,
Alexander

> Am 04.09.24 um 00:53 schrieb Linus Walleij:
> > On Fri, Apr 19, 2024 at 10:35 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:
> >
> >> Add support for damage handling and deferred I/O to fbdev-dma. This
> >> enables fbdev-dma to support all DMA-memory-based DRM drivers, even
> >> such with a dirty callback in their framebuffers.
> >>
> >> The patch adds the code for deferred I/O and also sets a dedicated
> >> helper for struct fb_ops.fb_mmap that support coherent mappings.
> >>
> >> v3:
> >> - init fb_ops with FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS() (Javier)
> >>
> >> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
> >> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
> > For some reason this makes the Versatile Express crash in QEMU
> > (I can retest on real hardware if necessary):
> >
> > 8<--- cut here ---
> > Unable to handle kernel paging request at virtual address 18095000 when write
> > [18095000] *pgd=00000000
> > Internal error: Oops: 805 [#1] SMP ARM
> > CPU: 0 PID: 75 Comm: S08_ledflash Not tainted 6.9.0-rc6+ #44
> > Hardware name: ARM-Versatile Express
> > PC is at mmioset+0x34/0xac
> > LR is at 0x0
> > pc : [<809c2a54>]    lr : [<00000000>]    psr: 20000013
> > sp : f8dddc38  ip : 18095000  fp : 00000000
> > r10: 81109a18  r9 : 81238b04  r8 : 00000000
> > r7 : 00440dc0  r6 : ed4f32a0  r5 : ed4f32a0  r4 : 00000001
> > r3 : 00000000  r2 : 00000fc0  r1 : 00000000  r0 : 18095000
> > Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
> > Control: 10c5387d  Table: 81d6406a  DAC: 00000051
> > Register r0 information: non-paged memory
> > Register r1 information: NULL pointer
> > Register r2 information: non-paged memory
> > Register r3 information: NULL pointer
> > Register r4 information: non-paged memory
> > Register r5 information: non-slab/vmalloc memory
> > Register r6 information: non-slab/vmalloc memory
> > Register r7 information: non-paged memory
> > Register r8 information: NULL pointer
> > Register r9 information: non-slab/vmalloc memory
> > Register r10 information: non-slab/vmalloc memory
> > Register r11 information: NULL pointer
> > Register r12 information: non-paged memory
> > Process S08_ledflash (pid: 75, stack limit = 0x(ptrval))
> > Stack: (0xf8dddc38 to 0xf8dde000)
> > dc20:                                                       00000801 802af888
> > dc40: 80a0fb80 00440dc0 00000801 811fad80 00000000 f8dddd28 ed4f32a0 811fa680
> > dc60: 00000000 802b0c30 ee1d2140 ee1d215c 95ff4385 00000001 00000001 00000001
> > dc80: 00000000 00000000 ffffffff 8196a70b 000000b0 00000064 0000006c 00000000
> > dca0: 00000001 00000002 00000001 811fa798 00000801 811fa940 80a0fb80 00000cc0
> > dcc0: 00000001 00000001 00000000 f8dddd28 811fa93c 811fa680 00000002 802b0c30
> > dce0: ee1d2140 00440dc0 00000000 00000000 00000001 00000000 811fad80 00440dc0
> > dd00: 00000001 00000000 00000000 83201100 00440dc0 00000000 00000000 802afa18
> > dd20: 00000bdc 00000bd8 811fad80 00000000 811fad80 00000000 00000000 00000000
> > dd40: adf00ec5 816d5fa8 816d5fa8 821c3c00 00000000 7ebc2000 7ebe3000 00000000
> > dd60: f8ddde44 8028cfa0 816d5fa8 8288df50 8169cf00 821c3c00 7ebc2000 7ebe3000
> > dd80: f8ddde44 8028e1e8 00000001 80bbe932 f8ddde44 f8dddde8 f8ddde44 8288df50
> > dda0: 83201100 809cf474 f8ddde44 8169cf00 00000020 8169cf94 821c3c94 7ebe3000
> > ddc0: 7ebe2fff fffffff4 00000000 816d5fa8 821c3c00 8169cf00 81d65fa8 809cdbb0
> > dde0: 8288df50 81ffa910 7ebe3000 82890b00 00000000 00000000 00000000 00000000
> > de00: 00000000 00000000 adf00ec5 00000000 00000000 8169cf00 8288df50 83201100
> > de20: 821c3c00 81ffa910 f8ddde44 8011f6e4 00000000 00000000 821c3c7c 8169cf7c
> > de40: 83200880 821c3c40 7ebc2000 7ebe2fff 82890b0c 76ea4000 ffffffff 00000000
> > de60: 00000000 0f000a01 f8ddde68 f8ddde68 adf00ec5 00000000 00000000 00000000
> > de80: 00000000 01200000 83201100 00000000 00000000 8011e2ac 00000000 83200d7c
> > dea0: 83201020 83200e70 83200e80 00000000 f8dddf30 83200880 00000000 00000000
> > dec0: 00000000 00000000 00000000 82182f38 adf00ec5 00000000 01200000 f8dddf30
> > dee0: 00000000 00000000 00000000 00000078 000cb21c 8011fe24 00000020 00000000
> > df00: 828a5d80 80a10140 adf00ec5 01200011 00000000 00000000 00000000 f8dddf30
> > df20: 00000000 00000078 000cb21c 801203d4 01200000 00000000 00000000 76f46468
> > df40: 00000000 00000000 00000011 00000000 00000000 00000000 00000000 00000000
> > df60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
> > df80: 00000003 adf00ec5 76f46468 00000000 76f468c0 00000078 80100290 83201100
> > dfa0: 000cb21c 80100060 76f46468 00000000 01200011 00000000 00000000 00000000
> > dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> > dfe0: 00000078 7ebe2938 76e0854d 76dae7e6 20000030 01200011 00000000 00000000
> > Call trace:
> >   mmioset from prep_new_page+0x160/0x194
> >   prep_new_page from get_page_from_freelist+0x10f4/0x110c
> >   get_page_from_freelist from __alloc_pages+0x15c/0x280
> >   __alloc_pages from __pte_alloc+0x28/0x1bc
> >   __pte_alloc from copy_page_range+0xc10/0xd28
> >   copy_page_range from copy_mm+0x5cc/0x850
> >   copy_mm from copy_process+0x508/0xd8c
> >   copy_process from kernel_clone+0x94/0x338
> >   kernel_clone from sys_clone+0x94/0xb0
> >   sys_clone from ret_fast_syscall+0x0/0x1c
> > Exception stack(0xf8dddfa8 to 0xf8dddff0)
> > dfa0:                   76f46468 00000000 01200011 00000000 00000000 00000000
> > dfc0: 76f46468 00000000 76f468c0 00000078 00000000 76e81000 76f46400 000cb21c
> > dfe0: 00000078 7ebe2938 76e0854d 76dae7e6
> > Code: e92d4100 e1a08001 e1a0e003 e2522040 (a8ac410a)
> > ---[ end trace 0000000000000000 ]---
> >
> > I bisected down to this commit and reverting the commit solves the issue.
> >
> > What is special about the Versatile Express graphics is that it uses a special
> > dedicated video RAM, looks like this in the device tree
> >
> >
> >          reserved-memory {
> >                  #address-cells = <1>;
> >                  #size-cells = <1>;
> >                  ranges;
> >
> >                  /* Chipselect 3 is physically at 0x4c000000 */
> >                  vram: vram@4c000000 {
> >                          /* 8 MB of designated video RAM */
> >                          compatible = "shared-dma-pool";
> >                          reg = <0x4c000000 0x00800000>;
> >                          no-map;
> >                  };
> >          };
> > (...)
> >           clcd@1f000 {
> >                        compatible = "arm,pl111", "arm,primecell";
> > (...)
> >                        memory-region = <&vram>;
> >
> > This gets picked up in the driver
> > drivers/gpu/drm/pl111/pl111_drv.c
> > like this:
> >
> >          ret = of_reserved_mem_device_init(dev);
> >          if (!ret) {
> >                  dev_info(dev, "using device-specific reserved memory\n");
> >                  priv->use_device_memory = true;
> >          }
> >
> > Yours,
> > Linus Walleij
> 
>
diff mbox series

Patch

diff --git a/drivers/gpu/drm/drm_fbdev_dma.c b/drivers/gpu/drm/drm_fbdev_dma.c
index 6c9427bb4053b..5eeb5164e9e2b 100644
--- a/drivers/gpu/drm/drm_fbdev_dma.c
+++ b/drivers/gpu/drm/drm_fbdev_dma.c
@@ -4,6 +4,7 @@ 
 
 #include <drm/drm_crtc_helper.h>
 #include <drm/drm_drv.h>
+#include <drm/drm_fb_dma_helper.h>
 #include <drm/drm_fb_helper.h>
 #include <drm/drm_framebuffer.h>
 #include <drm/drm_gem_dma_helper.h>
@@ -35,6 +36,22 @@  static int drm_fbdev_dma_fb_release(struct fb_info *info, int user)
 	return 0;
 }
 
+FB_GEN_DEFAULT_DEFERRED_DMAMEM_OPS(drm_fbdev_dma,
+				   drm_fb_helper_damage_range,
+				   drm_fb_helper_damage_area);
+
+static int drm_fbdev_dma_fb_mmap(struct fb_info *info, struct vm_area_struct *vma)
+{
+	struct drm_fb_helper *fb_helper = info->par;
+	struct drm_framebuffer *fb = fb_helper->fb;
+	struct drm_gem_dma_object *dma = drm_fb_dma_get_gem_obj(fb, 0);
+
+	if (!dma->map_noncoherent)
+		vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
+
+	return fb_deferred_io_mmap(info, vma);
+}
+
 static void drm_fbdev_dma_fb_destroy(struct fb_info *info)
 {
 	struct drm_fb_helper *fb_helper = info->par;
@@ -51,20 +68,13 @@  static void drm_fbdev_dma_fb_destroy(struct fb_info *info)
 	kfree(fb_helper);
 }
 
-static int drm_fbdev_dma_fb_mmap(struct fb_info *info, struct vm_area_struct *vma)
-{
-	struct drm_fb_helper *fb_helper = info->par;
-
-	return drm_gem_prime_mmap(fb_helper->buffer->gem, vma);
-}
-
 static const struct fb_ops drm_fbdev_dma_fb_ops = {
 	.owner = THIS_MODULE,
 	.fb_open = drm_fbdev_dma_fb_open,
 	.fb_release = drm_fbdev_dma_fb_release,
-	__FB_DEFAULT_DMAMEM_OPS_RDWR,
+	__FB_DEFAULT_DEFERRED_OPS_RDWR(drm_fbdev_dma),
 	DRM_FB_HELPER_DEFAULT_OPS,
-	__FB_DEFAULT_DMAMEM_OPS_DRAW,
+	__FB_DEFAULT_DEFERRED_OPS_DRAW(drm_fbdev_dma),
 	.fb_mmap = drm_fbdev_dma_fb_mmap,
 	.fb_destroy = drm_fbdev_dma_fb_destroy,
 };
@@ -98,10 +108,6 @@  static int drm_fbdev_dma_helper_fb_probe(struct drm_fb_helper *fb_helper,
 	dma_obj = to_drm_gem_dma_obj(buffer->gem);
 
 	fb = buffer->fb;
-	if (drm_WARN_ON(dev, fb->funcs->dirty)) {
-		ret = -ENODEV; /* damage handling not supported; use generic emulation */
-		goto err_drm_client_buffer_delete;
-	}
 
 	ret = drm_client_buffer_vmap(buffer, &map);
 	if (ret) {
@@ -112,7 +118,7 @@  static int drm_fbdev_dma_helper_fb_probe(struct drm_fb_helper *fb_helper,
 	}
 
 	fb_helper->buffer = buffer;
-	fb_helper->fb = buffer->fb;
+	fb_helper->fb = fb;
 
 	info = drm_fb_helper_alloc_info(fb_helper);
 	if (IS_ERR(info)) {
@@ -133,8 +139,19 @@  static int drm_fbdev_dma_helper_fb_probe(struct drm_fb_helper *fb_helper,
 	info->fix.smem_start = page_to_phys(virt_to_page(info->screen_buffer));
 	info->fix.smem_len = info->screen_size;
 
+	/* deferred I/O */
+	fb_helper->fbdefio.delay = HZ / 20;
+	fb_helper->fbdefio.deferred_io = drm_fb_helper_deferred_io;
+
+	info->fbdefio = &fb_helper->fbdefio;
+	ret = fb_deferred_io_init(info);
+	if (ret)
+		goto err_drm_fb_helper_release_info;
+
 	return 0;
 
+err_drm_fb_helper_release_info:
+	drm_fb_helper_release_info(fb_helper);
 err_drm_client_buffer_vunmap:
 	fb_helper->fb = NULL;
 	fb_helper->buffer = NULL;
@@ -144,8 +161,28 @@  static int drm_fbdev_dma_helper_fb_probe(struct drm_fb_helper *fb_helper,
 	return ret;
 }
 
+static int drm_fbdev_dma_helper_fb_dirty(struct drm_fb_helper *helper,
+					 struct drm_clip_rect *clip)
+{
+	struct drm_device *dev = helper->dev;
+	int ret;
+
+	/* Call damage handlers only if necessary */
+	if (!(clip->x1 < clip->x2 && clip->y1 < clip->y2))
+		return 0;
+
+	if (helper->fb->funcs->dirty) {
+		ret = helper->fb->funcs->dirty(helper->fb, NULL, 0, 0, clip, 1);
+		if (drm_WARN_ONCE(dev, ret, "Dirty helper failed: ret=%d\n", ret))
+			return ret;
+	}
+
+	return 0;
+}
+
 static const struct drm_fb_helper_funcs drm_fbdev_dma_helper_funcs = {
 	.fb_probe = drm_fbdev_dma_helper_fb_probe,
+	.fb_dirty = drm_fbdev_dma_helper_fb_dirty,
 };
 
 /*