On Fri, Aug 14, 2015 at 08:45:22AM +0700, Suravee Suthikulpanit wrote:

On 8/13/15 04:51, Jeremy Linton wrote:

quoted

ACPI configurations can now mark devices as noncoherent,
support that choice.

Signed-off-by: Jeremy Linton <redacted>
---
 include/acpi/acpi_bus.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/include/acpi/acpi_bus.h b/include/acpi/acpi_bus.h
index 83061ca..7ecb8e4 100644
--- a/include/acpi/acpi_bus.h
+++ b/include/acpi/acpi_bus.h

@@ -399,7 +399,7 @@ static inline bool acpi_check_dma(struct acpi_device *adev, bool *coherent)
 	 * case 1. Do not support and disable DMA.
 	 * case 2. Support but rely on arch-specific cache maintenance for
 	 *         non-coherence DMA operations.
-	 * Currently, we implement case 1 above.
+	 * Currently, we implement case 2 above.
 	 *
 	 * For the case when _CCA is missing (i.e. cca_seen=0) and
 	 * platform specifies ACPI_CCA_REQUIRED, we do not support DMA,

@@ -407,7 +407,8 @@ static inline bool acpi_check_dma(struct acpi_device *adev, bool *coherent)
 	 *
 	 * See acpi_init_coherency() for more info.
 	 */
-	if (adev->flags.coherent_dma) {
+	if (adev->flags.coherent_dma ||
+	    (adev->flags.cca_seen && IS_ENABLED(CONFIG_ARM64))) {
 		ret = true;
 		if (coherent)
 			*coherent = adev->flags.coherent_dma;

This change was in my earlier revisions for the original patch series to add
ACPI CCA support. At the time, this was pushed back since we were not sure
whether this would be a useful case, and whether such hardware exists.

Would it be useful to document somewhere (may be in the GIT commit message)
about which hardware might need this?

So far, it's the ARM Juno development board (the emphasis here is on
being able to use it for development, not a production system).

I think the commit log should also give you credit for the original
implementation.

Arnd/Catalin, any feedback on this?

That's where it was left in the previous thread:

https://lkml.org/lkml/2015/5/21/376

(and I'll refrain from further comments ;))

-- 
Catalin