On 04.03.2020 11:33, Russell King - ARM Linux admin wrote:
On Wed, Mar 04, 2020 at 10:56:06AM +0200, Laurentiu Tudor wrote:
On 04.03.2020 00:17, Russell King - ARM Linux admin wrote:
On Tue, Mar 03, 2020 at 05:55:05PM +0200, Laurentiu Tudor wrote:
From c98dc05cdd45ae923654f2427985bd28bcde4bb2 Mon Sep 17 00:00:00 2001
From: Laurentiu Tudor <[email protected]>
Date: Thu, 13 Feb 2020 11:59:12 +0200
Subject: [PATCH 1/4] bus: fsl-mc: add custom .dma_configure implementation
Content-Type: text/plain; charset="us-ascii"
The devices on this bus are not discovered by way of device tree
but by queries to the firmware. It makes little sense to trick the
generic of layer into thinking that these devices are of related so
that we can get our dma configuration. Instead of doing that, add
our custom dma configuration implementation.
Firstly, applying this to v5.5 results in a build failure, due to a
missing linux/iommu.h include.
Secondly, this on its own appears to make the DPAA2 network interfaces
completely disappear. Looking in /sys/bus/fsl-mc/drivers/*, none of
the DPAA2 drivers are bound to anything, and looking in
/sys/bus/fsl-mc/devices/, there is:
lrwxrwxrwx 1 root root 0 Mar 3 22:06 dprc.1 ->
../../../devices/platform/soc/80c000000.fsl-mc/dprc.1
This is booting with u-boot, so using DT rather than ACPI.
Signed-off-by: Laurentiu Tudor <[email protected]>
---
drivers/bus/fsl-mc/fsl-mc-bus.c | 42 ++++++++++++++++++++++++++++++++-
1 file changed, 41 insertions(+), 1 deletion(-)
diff --git a/drivers/bus/fsl-mc/fsl-mc-bus.c b/drivers/bus/fsl-mc/fsl-mc-bus.c
index 36eb25f82c8e..3df015eedae4 100644
--- a/drivers/bus/fsl-mc/fsl-mc-bus.c
+++ b/drivers/bus/fsl-mc/fsl-mc-bus.c
@@ -132,11 +132,51 @@ static int fsl_mc_bus_uevent(struct device *dev, struct
kobj_uevent_env *env)
static int fsl_mc_dma_configure(struct device *dev)
{
struct device *dma_dev = dev;
+ struct iommu_fwspec *fwspec;
+ const struct iommu_ops *iommu_ops;
+ struct fsl_mc_device *mc_dev = to_fsl_mc_device(dev);
+ int ret;
+ u32 icid;
+
+ /* Skip DMA setup for devices that are not DMA masters */
+ if (dev->type == &fsl_mc_bus_dpmcp_type ||
+ dev->type == &fsl_mc_bus_dpbp_type ||
+ dev->type == &fsl_mc_bus_dpcon_type ||
+ dev->type == &fsl_mc_bus_dpio_type)
+ return 0;
while (dev_is_fsl_mc(dma_dev))
dma_dev = dma_dev->parent;
- return of_dma_configure(dev, dma_dev->of_node, 0);
+ fwspec = dev_iommu_fwspec_get(dma_dev);
+ if (!fwspec)
+ return -ENODEV;
The problem appears to be here - fwspec is NULL for dprc.1.
Ok, that's because the iommu config is missing from the DT node that's
exposing the MC firmware. I've attached a fresh set of patches that include
on top the missing config and a workaround that makes MC work over SMMU.
Also added the missing #include, thanks for pointing out.
Let me know how it goes.
Shouldn't patch 6 should be first in the series, so that patch 1 does
not cause a regression and bisectability damage?
Correct, width one comment: both 5 and 6 should go first. Once iommu-map
is added in the device tree the workaround for MC with the
arm-smmu.disable_bypass boot arg will no longer work.
---
Best Regards, Laurentiu
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
