habanalabs/gaudi: use 8KB aligned address for TPC kernels

I$ prefetch is enabled when sending a TPC kernel to initialize the TPC
memory, and it has a restriction that the base address will be aligned
to 8KB.
Currently the base address is 128 bytes from the start address of the
device SRAM, so prefetching will start 128 bytes before the actual
kernel memory.
Modify the kernel address to be 8KB aligned.
Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

drivers/misc/habanalabs/gaudi/gaudi.c

@@ -984,9 +984,10 @@ static int _gaudi_init_tpc_mem(struct hl_device *hdev,
 	init_tpc_mem_pkt->ctl = cpu_to_le32(ctl);
 
 	init_tpc_mem_pkt->src_addr = cpu_to_le64(tpc_kernel_src_addr);
-	dst_addr = (prop->sram_user_base_address &
-			GAUDI_PKT_LIN_DMA_DST_ADDR_MASK) >>
-			GAUDI_PKT_LIN_DMA_DST_ADDR_SHIFT;
+
+	/* TPC_CMD is configured with I$ prefetch enabled, so address should be aligned to 8KB */
+	dst_addr = FIELD_PREP(GAUDI_PKT_LIN_DMA_DST_ADDR_MASK,
+				round_up(prop->sram_user_base_address, SZ_8K));
 	init_tpc_mem_pkt->dst_addr |= cpu_to_le64(dst_addr);
 
 	job = hl_cs_allocate_job(hdev, QUEUE_TYPE_EXT, true);