My experiance with Ubuntu 15.04 release 15.07


#1

Hi,

I just successfully installed Ubuntu on the Dragon Board.
Th system boots fast. I played around with the Epiphany browser. It doesn’t work well.
It crashes often and sometimes the whole system freezes. I can’t tell if there was a kernel crash as I don’t have a serial port. I installed Firefox browser and it fails to startup. I have some general comments:

  1. In spite of all the hype, I don’t such much value in running 64bit on a platform with 1GB memory. This would be true of Android Lollipop release as well.
  2. If Cortex-A53 brings some new instructions, it may be worthwhile to recompile all the packages with the new compiler options while still running a 32bit build. This would be similar to what Raspberry Pi did with Raspbian.
  3. It would be nice if there was enough documentation and access to any proprietary libraries and modules so that we can make our own flavor of Linux.

Regards
Vasant


#2

Vasant,

Both browsers are having issues with multimedia files. The results are not predictable. I have been using two popular videos on Youtube to test (trying to avoid cloud issues by using something popular). Sometimes they play, other times they do not.

I think Firefox is starting with a multimedia page so it fails to start. From a command line, try:

firefox ‘http://www.google.com’ &

and Firefox should start for you. Pages like google or wikipedia will work well. Youtube not so well.

Sean


#3

Please could you provide your dmesg output?

This is what I am observing when playing a youtube video:

[ 2834.437962] =============================================
[ 2834.438509] [ INFO: possible recursive locking detected ]
[ 2834.443809] 4.0.0-linaro-lt-qcom #1 Not tainted
[ 2834.449186] ---------------------------------------------
[ 2834.453537] multiqueue0:src/1239 is trying to acquire lock:
[ 2834.459085] (&mlist->lock){+.+…}, at: [] get_registered_mmap_buf+0x68/0xf0
[ 2834.468455]
but task is already holding lock:
[ 2834.473148] (&mlist->lock){+.+…}, at: [] msm_vdec_start_streaming+0x420/0x51c
[ 2834.482951]
other info that might help us debug this:
[ 2834.487994] Possible unsafe locking scenario:

[ 2834.494500] CPU0
[ 2834.500298] ----
[ 2834.502728] lock(&mlist->lock);
[ 2834.508099] lock(&mlist->lock);
[ 2834.511398]
*** DEADLOCK ***

[ 2834.511776] May be due to missing lock nesting notation

[ 2834.517685] 3 locks held by multiqueue0:src/1239:
[ 2834.524618] #0: (&v4l2_dev->ioctl_lock){+.+.+.}, at: [] v4l2_ioctl+0xdc/0x174
[ 2834.534333] #1: (&inst->bufq[CAPTURE_PORT].lock){+.+.+.}, at: [] msm_vdec_streamon+0x50/0x108
[ 2834.543971] #2: (&mlist->lock){+.+…}, at: [] msm_vdec_start_streaming+0x420/0x51c
[ 2834.552658]
stack backtrace:
[ 2834.557697] CPU: 1 PID: 1239 Comm: multiqueue0:src Not tainted 4.0.0-linaro-lt-qcom #1
[ 2834.557699] Hardware name: Qualcomm Technologies, Inc. APQ 8016 SBC (DT)
[ 2834.557701] Call trace:
[ 2834.557707] [] dump_backtrace+0x0/0x124
[ 2834.557719] [] show_stack+0x10/0x1c
[ 2834.557724] [] dump_stack+0x80/0xc4
[ 2834.557729] [] __lock_acquire+0x158c/0x16b8
[ 2834.557733] [] lock_acquire+0x50/0x78
[ 2834.557738] [] mutex_lock_nested+0x54/0x3b8
[ 2834.557742] [] get_registered_mmap_buf+0x64/0xf0
[ 2834.557746] [] msm_comm_qbuf+0x3d4/0x8f4
[ 2834.557750] [] msm_vdec_start_streaming+0x464/0x51c
[ 2834.557756] [] vb2_start_streaming+0x64/0x178
[ 2834.557761] [] vb2_internal_streamon+0x134/0x16c
[ 2834.557765] [] vb2_streamon+0x38/0x60
[ 2834.557769] [] msm_vdec_streamon+0x58/0x108
[ 2834.557774] [] msm_vidc_streamon+0x28/0x38
[ 2834.557777] [] msm_v4l2_streamon+0x18/0x24
[ 2834.557782] [] v4l_streamon+0x1c/0x28
[ 2834.557787] [] __video_do_ioctl+0x23c/0x2b0
[ 2834.557792] [] video_usercopy+0x328/0x46c
[ 2834.557796] [] video_ioctl2+0x10/0x1c
[ 2834.557800] [] v4l2_ioctl+0xfc/0x174
[ 2834.557806] [] do_vfs_ioctl+0x37c/0x60c
[ 2834.557810] [] SyS_ioctl+0x80/0x98


#4

Very similar results:

[ 442.905900] msm_vidc: info: Opening video instance: ffffffc02a0ae000, 1

[ 443.144795] =============================================
[ 443.145343] [ INFO: possible recursive locking detected ]
[ 443.150649] 4.0.0-linaro-lt-qcom #1 Not tainted
[ 443.156015] ---------------------------------------------
[ 443.160362] multiqueue0:src/1616 is trying to acquire lock:
[ 443.165918] (&mlist->lock){+.+…}, at: [<ffffffc000566cb4>] get_registered_
mmap_buf+0x68/0xf0
[ 443.175287]
but task is already holding lock:
[ 443.179979] (&mlist->lock){+.+…}, at: [<ffffffc000569554>] msm_vdec_start_
streaming+0x420/0x51c
[ 443.189786]
other info that might help us debug this:
[ 443.194829] Possible unsafe locking scenario:

[ 443.201334] CPU0
[ 443.207129] ----
[ 443.209559] lock(&mlist->lock);
[ 443.214931] lock(&mlist->lock);
[ 443.218229]
*** DEADLOCK ***

[ 443.218608] May be due to missing lock nesting notation

[ 443.224520] 3 locks held by multiqueue0:src/1616:
[ 443.231450] #0: (&v4l2_dev->ioctl_lock){+.+.+.}, at: [<ffffffc000540950>] v
4l2_ioctl+0xdc/0x174
[ 443.241166] #1: (&inst->bufq[CAPTURE_PORT].lock){+.+.+.}, at: [<ffffffc0005
6b8dc>] msm_vdec_streamon+0x50/0x108
[ 443.250806] #2: (&mlist->lock){+.+…}, at: [<ffffffc000569554>] msm_vdec_s
tart_streaming+0x420/0x51c
[ 443.259490]
stack backtrace:
[ 443.264525] CPU: 0 PID: 1616 Comm: multiqueue0:src Not tainted 4.0.0-linaro-l
t-qcom #1
[ 443.268968] Hardware name: Qualcomm Technologies, Inc. APQ 8016 SBC (DT)
[ 443.276776] Call trace:
[ 443.283612] [<ffffffc000089b00>] dump_backtrace+0x0/0x124
[ 443.285797] [<ffffffc000089c34>] show_stack+0x10/0x1c
[ 443.291359] [<ffffffc0007f04ec>] dump_stack+0x80/0xc4
[ 443.296387] [<ffffffc0000f0508>] __lock_acquire+0x158c/0x16b8
[ 443.301423] [<ffffffc0000f0dfc>] lock_acquire+0x50/0x78
[ 443.307151] [<ffffffc0007f5284>] mutex_lock_nested+0x54/0x3b8
[ 443.312189] [<ffffffc000566cb0>] get_registered_mmap_buf+0x64/0xf0
[ 443.318093] [<ffffffc00055f7d4>] msm_comm_qbuf+0x3d4/0x8f4
[ 443.324164] [<ffffffc000569598>] msm_vdec_start_streaming+0x464/0x51c
[ 443.329639] [<ffffffc000555218>] vb2_start_streaming+0x64/0x178
[ 443.336147] [<ffffffc000557778>] vb2_internal_streamon+0x134/0x16c
[ 443.341876] [<ffffffc0005577e8>] vb2_streamon+0x38/0x60
[ 443.348122] [<ffffffc00056b8e4>] msm_vdec_streamon+0x58/0x108
[ 443.353246] [<ffffffc000565fcc>] msm_vidc_streamon+0x28/0x38
[ 443.359148] [<ffffffc00055af60>] msm_v4l2_streamon+0x18/0x24
[ 443.364878] [<ffffffc000541c14>] v4l_streamon+0x1c/0x28
[ 443.370518] [<ffffffc000544570>] __video_do_ioctl+0x23c/0x2b0
[ 443.375467] [<ffffffc0005441d4>] video_usercopy+0x328/0x46c
[ 443.381368] [<ffffffc000544328>] video_ioctl2+0x10/0x1c
[ 443.386748] [<ffffffc000540970>] v4l2_ioctl+0xfc/0x174
[ 443.391960] [<ffffffc0001bcbcc>] do_vfs_ioctl+0x37c/0x60c
[ 443.397169] [<ffffffc0001bcedc>] SyS_ioctl+0x80/0x98
[ 451.233077] ------------[ cut here ]------------
[ 451.233139] WARNING: CPU: 2 PID: 1570 at …/drivers/media/v4l2-core/videobuf2
-core.c:2139 __vb2_queue_cancel+0xfc/0x158()


#5

the crash log indicates a problem with the hwcodecs (msm_v4l2/msm_vidc driver) which has been included in 15.07 release. Epiphany is using Gstreamer by default for Multimedia, so it will end up using the hwcodecs through the Gst v4l2dec plugin in the release… in the mean time, can you please try to remove the hw plugin:

apt-get remove gstreamer1.0-plugins-v4l2

that doesn’t really fix the problem… just a workaround for now, until we fix the real problem.


#6

I removed gstreamer1.0-plugins-v4l2. The browser launch, but cannot play video. Processor utilization for the webkit soars to 100%.

Any ideas when the problem will be fixed?

Thanks, Sean