[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20170427154310.GA21049@red-moon>
Date: Thu, 27 Apr 2017 16:43:10 +0100
From: Lorenzo Pieralisi <lorenzo.pieralisi@....com>
To: Arnd Bergmann <arnd@...db.de>
Cc: Thomas Gleixner <tglx@...utronix.de>,
Jason Cooper <jason@...edaemon.net>,
Marc Zyngier <marc.zyngier@....com>,
Hanjun Guo <hanjun.guo@...aro.org>,
Tomasz Nowicki <tn@...ihalf.com>,
Will Deacon <will.deacon@....com>,
"Rafael J. Wysocki" <rafael@...nel.org>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] irqchip/gicv3-its: use IORT only if IOMMU support is
enabled
On Thu, Apr 27, 2017 at 04:26:12PM +0200, Arnd Bergmann wrote:
> When the IOMMU subsystem is disabled, we cannot build ARM64 kernels with ACPI:
>
> drivers/acpi/arm64/iort.c: In function 'iort_iommu_xlate':
> drivers/acpi/arm64/iort.c:647:22: error: 'struct iommu_fwspec' has no member named 'ops'
Damn it. This is triggered by 5a1bb638d567 ("drivers: acpi: Handle IOMMU
lookup failure with deferred probing or error") in -next.
There are a couple of things to fix here but not with this patch
(it is not correct to add a dependency on IOMMU on GIC V3 ITS).
I am not even sure that the series containing the above commit will
get merged so it is not easy to decide how to fix it up.
The ACPI_IORT Kconfig selection _must_ be reworked but not now, it
is too late in the cycle IMO.
I suspect the only thing I can do is to send Joerg a fix-up in case
the series containing the above commit is actually merged (he will
send it upstream through the IOMMU tree).
At -rc1 I will definitely have a patch ready to rework the ACPI_IORT
Kconfig selection because it is just wrong at the moment.
Thanks for the heads-up,
Lorenzo
> This adds a Kconfig dependency to only enable ACPI_IORT when it acutally
> builds, along with addititonal dummy helper functions to make its
> users build without IOMMU/IORT.
>
> Fixes: 3f010cf19732 ("irqchip/gicv3-its: Probe ITS in the ACPI way")
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
> drivers/irqchip/Kconfig | 2 +-
> include/linux/acpi_iort.h | 5 ++++-
> 2 files changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> index 478f8ace2664..f0516740c03f 100644
> --- a/drivers/irqchip/Kconfig
> +++ b/drivers/irqchip/Kconfig
> @@ -39,7 +39,7 @@ config ARM_GIC_V3_ITS
> bool
> depends on PCI
> depends on PCI_MSI
> - select ACPI_IORT if ACPI
> + select ACPI_IORT if ACPI && IOMMU
>
> config ARM_NVIC
> bool
> diff --git a/include/linux/acpi_iort.h b/include/linux/acpi_iort.h
> index 3ff9acea8616..918906d189d3 100644
> --- a/include/linux/acpi_iort.h
> +++ b/include/linux/acpi_iort.h
> @@ -26,10 +26,10 @@
> #define IORT_IRQ_MASK(irq) (irq & 0xffffffffULL)
> #define IORT_IRQ_TRIGGER_MASK(irq) ((irq >> 32) & 0xffffffffULL)
>
> +#ifdef CONFIG_ACPI_IORT
> int iort_register_domain_token(int trans_id, struct fwnode_handle *fw_node);
> void iort_deregister_domain_token(int trans_id);
> struct fwnode_handle *iort_find_domain_token(int trans_id);
> -#ifdef CONFIG_ACPI_IORT
> void acpi_iort_init(void);
> bool iort_node_match(u8 type);
> u32 iort_msi_map_rid(struct device *dev, u32 req_id);
> @@ -40,6 +40,9 @@ int iort_pmsi_get_dev_id(struct device *dev, u32 *dev_id);
> void iort_set_dma_mask(struct device *dev);
> const struct iommu_ops *iort_iommu_configure(struct device *dev);
> #else
> +static inline int iort_register_domain_token(int trans_id, struct fwnode_handle *fw_node) { return -ENOSYS; }
> +static inline void iort_deregister_domain_token(int trans_id) {}
> +static inline struct fwnode_handle *iort_find_domain_token(int trans_id) { return NULL;}
> static inline void acpi_iort_init(void) { }
> static inline bool iort_node_match(u8 type) { return false; }
> static inline u32 iort_msi_map_rid(struct device *dev, u32 req_id)
> --
> 2.9.0
>
Powered by blists - more mailing lists