Doing that and read it back with the value, but smt else is overwriting it to zero.. idk what..that should be done in rk_iommu_enable() (write dt page p_addr to iommu reg). also a failure should be logged
panfrost0: <Mali Midgard/Bifrost GPU> mem 0xfdea0000-0xfdea3fff irq 19,20,21 on ofwbus0
panfrost0: Mali GPU clock is unknown
device_attach: panfrost0 attach returned 6
panfrost0: Mali GPU clock is unknown
I repost the DTE_ADDR to offset 0 from the vop, and renables paging.. that seems to work..Doing that and read it back with the value, but smt else is overwriting it to zero.. idk what..
vop1 seems to have some iommu control registers in its own register space at offset 0x300I repost the DTE_ADDR to offset 0 from the vop, and renables paging.. that seems to work..
Yeah also in the base-dtsi. the regs are shown.vop1 seems to have some iommu control registers in its own register space at offset 0x300
vop2 has no such thing. i don't know how they are suppose to work as they are not used in the linux driver but
https://rockchip.fr/RK3288 TRM/rk3288-chapter-27-visual-output-processor-(vop).pdf (search for dte_addr)
and here https://rockchip.fr/Rockchip RK3399 TRM V1.4 Part1.pdf page 555
yeah, but I have no board to o test v2.differences from iommu v1 to v2 are what, 20 lines ?
you probably run a quartz64 cloned dtb i posted here, not the official radxa one
source is this imageI've been tempted so hardly and I've bought the "Radxa Cubie A7Z". Do you have or do you know where I can grab the DTB file adapted for FreeBSD ?