[PATCH] Disable BROKEN MSM_IOMMU

[Nicolas Chauvet]

MSM_IOMMU doesn't play nice in multi-platform kernel
This is causing probe issue as it tries to register unconditionally

Specially experienced on tegra_drm where IOMMU support was enabled
for tegra30 and later since 3.19. It has prevented the fallback
to non-iommu in the tegra20 case.

This need additionals "workaround" patches for EXYNOS, OMAP and Rockchip.
v2 for thoses patches are reported at:
http://lists.linuxfoundation.org/pipermail/iommu/2015-February/012226.html

This config change and thoses additionals patches would need
to be backported down to 3.19 fedora kernels
---
 config-armv7 |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/config-armv7 b/config-armv7
index 7f76e38..4e62e3a 100644
--- a/config-armv7
+++ b/config-armv7
@@ -356,7 +356,7 @@ CONFIG_PHY_QCOM_APQ8064_SATA=m
 CONFIG_PHY_QCOM_IPQ806X_SATA=m
 CONFIG_USB_DWC3_QCOM=m
 CONFIG_CRYPTO_DEV_QCE=m
-CONFIG_MSM_IOMMU=y
+# CONFIG_MSM_IOMMU is not set
 CONFIG_DRM_MSM=m
 CONFIG_DRM_MSM_FBDEV=y
 CONFIG_USB_EHCI_MSM=m
-- 
1.7.2.1

[Rob Clark]

On Feb 24 2015 or thereabouts, Nicolas Chauvet wrote:
>  MSM_IOMMU doesn't play nice in multi-platform kernel
>  This is causing probe issue as it tries to register unconditionally
>  
>  Specially experienced on tegra_drm where IOMMU support was enabled
>  for tegra30 and later since 3.19. It has prevented the fallback
>  to non-iommu in the tegra20 case.
>  
>  This need additionals "workaround" patches for EXYNOS, OMAP and Rockchip.
>  v2 for thoses patches are reported at:
>  http://lists.linuxfoundation.org/pipermail/iommu/2015-February/012226.html
>  
>  This config change and thoses additionals patches would need
>  to be backported down to 3.19 fedora kernels 

fwiw, there was a patch upstream to mark MSM_IOMMU as depends on
BROKEN.. not sure if that would be sufficient

(either way, I don't really mind, the current upstream msm-iommu is not
actually used by any driver, at least not until I get time to rename
qcom-iommu and rework it to fit the one-true-iommu-bindings..)

BR,
-R

>  ---
>   config-armv7 |    2 +-
>   1 files changed, 1 insertions(+), 1 deletions(-)
>  
>  diff --git a/config-armv7 b/config-armv7
>  index 7f76e38..4e62e3a 100644
>  --- a/config-armv7
>  +++ b/config-armv7
>  @@ -356,7 +356,7 @@ CONFIG_PHY_QCOM_APQ8064_SATA=m
>   CONFIG_PHY_QCOM_IPQ806X_SATA=m
>   CONFIG_USB_DWC3_QCOM=m
>   CONFIG_CRYPTO_DEV_QCE=m
>  -CONFIG_MSM_IOMMU=y
>  +# CONFIG_MSM_IOMMU is not set
>   CONFIG_DRM_MSM=m
>   CONFIG_DRM_MSM_FBDEV=y
>   CONFIG_USB_EHCI_MSM=m
>  -- 
>  1.7.2.1
>

[Peter Robinson]

On Tue, Feb 24, 2015 at 11:27 AM, Nicolas Chauvet <kwizart(a)gmail.com&gt; wrote:
>  MSM_IOMMU doesn't play nice in multi-platform kernel
>  This is causing probe issue as it tries to register unconditionally
>
>  Specially experienced on tegra_drm where IOMMU support was enabled
>  for tegra30 and later since 3.19. It has prevented the fallback
>  to non-iommu in the tegra20 case.
>
>  This need additionals "workaround" patches for EXYNOS, OMAP and Rockchip.
>  v2 for thoses patches are reported at:
>  http://lists.linuxfoundation.org/pipermail/iommu/2015-February/012226.html
>
>  This config change and thoses additionals patches would need
>  to be backported down to 3.19 fedora kernels 
I've disabled that driver and pushed it, I'll review the rest of the
patches shortly.

Peter