I had some problems debugging multicore, then I verified cache consistency between two clusters under uboot, and I found that every time the two clusters communicated with each other, 0x140 bytes did not align, and other data was normal.
I try to test with the core0 of cluster0 and core0 of cluster1. The test steps are as follows:
1 core0 of cluster0 writes 0x66666666 to 0x80000000 to 0x80000400
The core0 of 2 cluster1 reads 0x80000000 to 0x80000400
3 core0 of cluster1 writes 0x99999999 to 0x80000200 to 0x80000400
4 core0 of cluster0 writes 0x99999999 to 0x80000000 to 0x80000400
In theory, the core0 of cluster1 should be able to read 0x666666666 in the address range of 0x80000000~0x80000400, but the world did not read it, I guess it may be because the address state machine is M, ncore did not move. At the same time, the core0 of cluster0 does not read 0x99999999, for a similar reason.
Question:
How do I verify that ncore takes effect in uboot
