-----Original Message-----
From: Nicolas Dufresne [mailto:nicolas@ndufresne.ca]
Sent: Thursday, September 23, 2021 10:40 PM
To: Ming Qian <ming.qian@nxp.com>; mchehab@kernel.org;
shawnguo@kernel.org; robh+dt@kernel.org; s.hauer@pengutronix.de
Cc: hverkuil-cisco@xs4all.nl; kernel@pengutronix.de; festevam@gmail.com;
dl-linux-imx [off-list ref]; Aisheng Dong [off-list ref];
linux-media@vger.kernel.org; linux-kernel@vger.kernel.org;
devicetree@vger.kernel.org; linux-arm-kernel@lists.infradead.org
Subject: [EXT] Re: [PATCH v9 06/13] media: amphion: add vpu v4l2 m2m
support

Caution: EXT Email

Le lundi 13 septembre 2021 à 17:11 +0800, Ming Qian a écrit :

quoted

vpu_v4l2.c implements the v4l2 m2m driver methods.
vpu_helpers.c implements the common helper functions
vpu_color.c converts the v4l2 colorspace with iso

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Shijie Qin <redacted>
Signed-off-by: Zhou Peng <eagle.zhou@nxp.com>
Reported-by: kernel test robot <redacted>
---
 drivers/media/platform/amphion/vpu_color.c   | 192 ++++++
 drivers/media/platform/amphion/vpu_helpers.c | 453 ++++++++++++++
 drivers/media/platform/amphion/vpu_helpers.h |  72 +++
 drivers/media/platform/amphion/vpu_v4l2.c    | 625

+++++++++++++++++++

quoted

 drivers/media/platform/amphion/vpu_v4l2.h    |  53 ++
 5 files changed, 1395 insertions(+)
 create mode 100644 drivers/media/platform/amphion/vpu_color.c
 create mode 100644 drivers/media/platform/amphion/vpu_helpers.c
 create mode 100644 drivers/media/platform/amphion/vpu_helpers.h
 create mode 100644 drivers/media/platform/amphion/vpu_v4l2.c
 create mode 100644 drivers/media/platform/amphion/vpu_v4l2.h

+dma_addr_t vpu_get_vb_phy_addr(struct vb2_buffer *vb, u32 plane_no)
+{
+     return vb2_dma_contig_plane_dma_addr(vb, plane_no) +
+                     vb->planes[plane_no].data_offset;

I've change the num_planes to 1 (tried 2, but failed differently), and it
crashed while running:

GST_DEBUG="v4l2*:7" gst-launch-1.0 videotestsrc num-buffers=100 !
v4l2h264enc ! fakesink

So perhaps there is a overall review of the buffer allocation vs S_FMT
implementation to be done here.

Hi Nicolas,
   What's the failure of setting num_planes to 2?
   When num_planes is set to 1, driver still try to access the planes[1], but got NULL pointer. Then led to this panic.
   I'll try to support V4L2_PIX_FMT_NV12M, and check the plane number carefully.

[   81.520167] Unable to handle kernel NULL pointer dereference at virtual
address 0000000000000000
[   81.529168] Mem abort info:
[   81.532467]   ESR = 0x96000004
[   81.539550]   EC = 0x25: DABT (current EL), IL = 32 bits
[   81.546053]   SET = 0, FnV = 0
[   81.549771]   EA = 0, S1PTW = 0
[   81.553186]   FSC = 0x04: level 0 translation fault
[   81.558280] Data abort info:
[   81.561437]   ISV = 0, ISS = 0x00000004
[   81.565527]   CM = 0, WnR = 0
[   81.568581] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000882741000
[   81.575840] [0000000000000000] pgd=0000000000000000,
p4d=0000000000000000
[   81.586637] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[   81.592247] Modules linked in: amphion_vpu_core v4l2_mem2mem
videobuf2_vmalloc videobuf2_dma_contig videobuf2_memops videobuf2_v4l2
videobuf2_common imx_sc_key imx_sc_thermal crct10dif_ce rtc_imx_sc
imx_sc_wdt
amphion_vpu_dev fsl_imx8_ddr_perf isl29018 mpl3115
industrialio_triggered_buffer
kfifo_buf videodev mc fuse drm ip_tables x_tables ipv6
[   81.622805] CPU: 0 PID: 348 Comm: videotestsrc0:s Not tainted
5.15.0-rc2-
imx8qxp+ #2
[   81.630562] Hardware name: Freescale i.MX8QXP MEK (DT)
[   81.635711] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS
BTYPE=--)
[   81.642687] pc : vpu_get_vb_phy_addr+0x28/0x50 [amphion_vpu_core]
[   81.648838] lr : vpu_get_vb_phy_addr+0x20/0x50 [amphion_vpu_core]
[   81.654971] sp : ffff800012fcba30
[   81.658287] x29: ffff800012fcba30 x28: ffff800012bd1040 x27:
0000000000000000
[   81.665449] x26: 0000000000000000 x25: ffff800012fcbd28 x24:
ffff800012bd1010
[   81.672607] x23: ffff800012bd3200 x22: 0000000000000000 x21:
ffff800015807044
[   81.679766] x20: ffff000806a1ec00 x19: 0000000000000003 x18:
0000000000000000
[   81.686924] x17: 0000000000000000 x16: 0000000000000000 x15:
0000a95ba29c327a
[   81.694083] x14: 000000000000039f x13: 0000000000000000 x12:
0000000000000000
[   81.701241] x11: 0000000000000000 x10: 0000000000000990 x9 :
ffff800012fcb900
[   81.708400] x8 : ffff0008068f97f0 x7 : ffff00083f98b180 x6 :
ffff000802070af8
[   81.715558] x5 : ffff000802070af8 x4 : ffff800009296000 x3 :
ffff000802073480
[   81.722717] x2 : ffff000806a1ec00 x1 : 0000000000000002 x0 :
0000000000000000
[   81.729877] Call trace:
[   81.732332]  vpu_get_vb_phy_addr+0x28/0x50 [amphion_vpu_core]
[   81.738119]  vpu_windsor_input_frame+0x84/0xbc [amphion_vpu_core]
[   81.744252]  venc_process_output+0x88/0x10c [amphion_vpu_core]
[   81.750125]  vpu_process_output_buffer+0xac/0x100
[amphion_vpu_core]
[   81.756519]  venc_start_session+0x264/0x2cc [amphion_vpu_core]
[   81.762392]  vpu_vb2_start_streaming+0x88/0xf4 [amphion_vpu_core]
[   81.768525]  vb2_start_streaming+0x68/0x15c [videobuf2_common]
[   81.774398]  vb2_core_streamon+0x94/0x19c [videobuf2_common]
[   81.780088]  vb2_streamon+0x20/0x70 [videobuf2_v4l2]
[   81.785083]  v4l2_m2m_ioctl_streamon+0x3c/0xa0 [v4l2_mem2mem]
[   81.790869]  v4l_streamon+0x28/0x34 [videodev]
[   81.795412]  __video_do_ioctl+0x17c/0x3e0 [videodev]
[   81.800460]  video_usercopy+0x368/0x7e0 [videodev]
[   81.805333]  video_ioctl2+0x1c/0x3c [videodev]
[   81.809859]  v4l2_ioctl+0x44/0x64 [videodev]
[   81.814211]  __arm64_sys_ioctl+0xac/0xf0
[   81.818155]  invoke_syscall+0x48/0x114
[   81.821917]  el0_svc_common.constprop.0+0x44/0xfc
[   81.826635]  do_el0_svc+0x2c/0x94
[   81.829961]  el0_svc+0x28/0x80
[   81.833027]  el0t_64_sync_handler+0xa8/0x130
[   81.837311]  el0t_64_sync+0x1a0/0x1a4
[   81.840992] Code: aa0003f4 97ff02f9 d37f7e61 8b334033 (f9400000)
[   81.847099] ---[ end trace 46cf421377a59de1 ]---

quoted

+}
+