Thread (17 messages) flat view 17 messages, 5 authors, 2012-10-24

[PATCH 2/2] arm: mvebu: Add hardware I/O Coherency support

From: Gregory CLEMENT <hidden>
Date: 2012-10-24 11:53:48
Also in: lkml

On 10/24/2012 01:48 PM, Gregory CLEMENT wrote:
On 10/24/2012 01:36 PM, Arnd Bergmann wrote:
quoted
On Wednesday 24 October 2012, Gregory CLEMENT wrote:
quoted
+void __init armada_370_xp_coherency_iocache_init(void)
+{
+       /* When the coherency fabric is available, the Armada XP and
+        * Aramada 370 are close to a coherent architecture, so we based
+        * our dma ops on the coherent one, and just changes the
+        * operations which need a arch io sync */
+       if (of_find_compatible_node(NULL, NULL, "marvell,coherency-fabric")) {
+               struct dma_map_ops *dma_ops = &armada_xp_dma_ops;
+               memcpy(dma_ops, &arm_coherent_dma_ops, sizeof(*dma_ops));
+               dma_ops->map_page = armada_xp_dma_map_page;
+               dma_ops->unmap_page = armada_xp_dma_unmap_page;
+               dma_ops->unmap_sg = arm_dma_ops.unmap_sg;
+               dma_ops->sync_single_for_cpu = armada_xp_dma_sync;
+               dma_ops->sync_single_for_device = armada_xp_dma_sync;
+               dma_ops->sync_sg_for_cpu = arm_dma_ops.sync_sg_for_cpu;
+               dma_ops->sync_sg_for_device = arm_dma_ops.sync_sg_for_device;
+       }
+       bus_register_notifier(&platform_bus_type, &armada_xp_platform_nb);
I think it would be cleaner to statically define the operations in a constant
structure and point directly to the functions you need. If necessary, use
multiple structures.
My problem was that these functions are not exposed, only arm_dma_op and
arm_coherent_dma_ops are exported.
Or do you think about something like this:

struct dma_map_ops *dma_ops = {
	.alloc			= arm_dma_ops.arm_coherent_dma_alloc,
	.free			= arm_dma_ops.arm_coherent_dma_free,
	.mmap			= arm_dma_ops.arm_dma_mmap,
	.get_sgtable		= arm_dma_ops.arm_dma_get_sgtable,
	.map_sg			= arm_dma_ops.arm_dma_map_sg,
	.set_dma_mask		= arm_dma_ops.arm_dma_set_mask,
	.map_page 		= armada_xp_dma_map_page,
	.unmap_page 		= armada_xp_dma_unmap_page,
	.unmap_sg 		= arm_dma_ops.unmap_sg,
	.sync_single_for_cpu 	= armada_xp_dma_sync,
	.sync_single_for_device = armada_xp_dma_sync,
	.sync_sg_for_cpu 	= arm_dma_ops.sync_sg_for_cpu,
	.sync_sg_for_device	= arm_dma_ops.sync_sg_for_device,
};
I was too fast to reply this struct is wrong, it should be as this one:

struct dma_map_ops *dma_ops = {
	.alloc			= arm_coherent_dma_ops.arm_coherent_dma_alloc,
	.free			= arm_coherent_dma_ops.arm_coherent_dma_free,
	.mmap			= arm_coherent_dma_ops.arm_dma_mmap,
	.get_sgtable		= arm_coherent_dma_ops.arm_dma_get_sgtable,
	.map_sg			= arm_coherent_dma_ops.arm_dma_map_sg,
	.set_dma_mask		= arm_coherent_dma_ops.arm_dma_set_mask,
	.map_page 		= armada_xp_dma_map_page,
	.unmap_page 		= armada_xp_dma_unmap_page,
	.unmap_sg 		= arm_dma_ops.unmap_sg,
	.sync_single_for_cpu 	= armada_xp_dma_sync,
	.sync_single_for_device = armada_xp_dma_sync,
	.sync_sg_for_cpu 	= arm_dma_ops.sync_sg_for_cpu,
	.sync_sg_for_device 	= arm_dma_ops.sync_sg_for_device,
};

Thanks,
Gregory
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help