Tool/software:
Hi,
After booting the system, opening multiple SSH terminals and executing multiple pipelines may cause the system to hang intermittently. Both the number of pipelines launched and the time intervals between their launches can affect the occurrence of the freeze.
The shorter the time interval between pipeline executions, the higher the probability of a hang.
cms
is launched first, followed almost immediately by ir
and rgb
, the system will freeze consistently.The greater the number of pipelines launched, the higher the probability of a hang.
cms
immediately after booting has an extremely low probability of freezing.ir
after cms
has successfully started has a low probability of freezing.rgb
after both cms
and ir
are running results in a significantly higher probability of freezing.Test Result: Scenario where the initially launched pipeline freezes.
Pipelines started later may freeze at the GST log message related to clock synchronization, with no subsequent ISP-related logs printed.
The dmesg
output is as follows:
By using the devmem2
command to inspect the relevant registers, it was observed that the values of registers 0x30102040
, 0x30102060
, 0x30102080
, 0x30101060
, and 0x31011028
remained identical during both GST freezes and normal GST operation. Specifically, the values were:0x8010006C
, 0x801000AB
, 0x801000EB
, 0x20000000
, and 0x000E0100
.
Regarding this issue, I have the following questions:
0x30102020 + 20*j
, are there other registers that can be used to check the DMA’s operational status?
TIPS:
The log messages printed after data reception are a debugging tool to monitor whether camera image data frames are successfully received.
Thanks,
Yuwei
Hello Yuwei,
I suspect that this is a Gstreamer issue. Can you try running yavta or v4l2 capture in a similar way and see if you can catch raw images without being frozen? This can help us rule out any CSI Rx related issue.
Thank you.
Jianzhong