I am using imx6 custom board, i am facing CMA latency and allocation failure issue.
Is there any way to reduce latency of CMA allocation ?
Any help would be appreciated.
There could be several root causes. It could be that you are requiring large memory allocations and this process can increase latency depending on the memory configuration. If this is the case you could try setting a memory region on the device tree for the nodes requiring large amounts of memory, but there is no much information on how to configure this on NXP documentation.
I don't want to go with static allocation, CMA has also another issue sometime it will failed to allocate memory, so CMA is not reliable, let me know is there any other options ?
GCMA and CONFIG_CMA zone are the options available but they have also their own drawback, so thats why they are not in mainline kernel.
Let me know if you know any other way for dynamic allocation ?
Retrieving data ...