I don't think it is possible using OpenCL. LDS is used to store data allocated in local address space and only can be used to share data among WIs within a WG.
As per OpenCL spec:
" Variables allocated in the __local address space inside a kernel function are allocated for each work-group executing the kernel and exist only for the lifetime of the work-group executing the kernel."
Similarly, as per AMD Optimization guide:
"Local memory is software-controlled “scratchpad” memory. ... The scratchpad allows the kernel to explicitly load items into the memory; they exist in local memory until the kernel replaces them, or until the work-group ends. To declare a block of local memory, use the __local keyword; for example: __local float localBuffer"
Normally, h/w resource manager allocates the LDS space to a new WG as soon as one WG finishes. So, it can not be used to communicate between two kernel calls.
I suspected as much. Ty.