From: will.deacon@arm.com (Will Deacon)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v8 3/3] iommu/arm-smmu-v3: Add workaround for Cavium ThunderX2 erratum #126
Date: Tue, 20 Jun 2017 19:00:39 +0100 [thread overview]
Message-ID: <20170620180038.GC28035@arm.com> (raw)
In-Reply-To: <1497968259-16390-4-git-send-email-gakula@caviumnetworks.com>
On Tue, Jun 20, 2017 at 07:47:39PM +0530, Geetha sowjanya wrote:
> From: Geetha Sowjanya <geethasowjanya.akula@cavium.com>
>
> Cavium ThunderX2 SMMU doesn't support MSI and also doesn't have unique irq
> lines for gerror, eventq and cmdq-sync.
>
> SHARED_IRQ option is set as a errata workaround, which allows to share the irq
> line by register single irq handler for all the interrupts.
>
> Signed-off-by: Geetha sowjanya <gakula@caviumnetworks.com>
> ---
> .../devicetree/bindings/iommu/arm,smmu-v3.txt | 5 ++
> drivers/iommu/arm-smmu-v3.c | 73 ++++++++++++++++----
> 2 files changed, 64 insertions(+), 14 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/iommu/arm,smmu-v3.txt b/Documentation/devicetree/bindings/iommu/arm,smmu-v3.txt
> index 6ecc48c..44b40e0 100644
> --- a/Documentation/devicetree/bindings/iommu/arm,smmu-v3.txt
> +++ b/Documentation/devicetree/bindings/iommu/arm,smmu-v3.txt
> @@ -55,6 +55,11 @@ the PCIe specification.
> Set for Caviun ThunderX2 silicon that doesn't support
> SMMU page1 register space.
>
> +- cavium,cn9900-broken-unique-irqline
> + : Use single irq line for all the SMMUv3 interrupts.
> + Set for Caviun ThunderX2 silicon that doesn't support
> + MSI and also doesn't have unique irq lines for gerror,
> + eventq and cmdq-sync.
I think we're better off just supporting a new (optional) named interrupt
as "combined", and then allowing that to be used instead of the others.
> ** Example
>
> smmu at 2b400000 {
> diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c
> index 2dea4a9..6c0c632 100644
> --- a/drivers/iommu/arm-smmu-v3.c
> +++ b/drivers/iommu/arm-smmu-v3.c
> @@ -598,6 +598,7 @@ struct arm_smmu_device {
>
> #define ARM_SMMU_OPT_SKIP_PREFETCH (1 << 0)
> #define ARM_SMMU_OPT_PAGE0_REGS_ONLY (1 << 1)
> +#define ARM_SMMU_OPT_SHARED_IRQ (1 << 2)
Please call this COMBINED instead of SHARED (similarly elsewhere). That
said, not sure we need this.
> u32 options;
>
> struct arm_smmu_cmdq cmdq;
> @@ -665,6 +666,7 @@ struct arm_smmu_option_prop {
> static struct arm_smmu_option_prop arm_smmu_options[] = {
> { ARM_SMMU_OPT_SKIP_PREFETCH, "hisilicon,broken-prefetch-cmd" },
> { ARM_SMMU_OPT_PAGE0_REGS_ONLY, "cavium,cn9900-broken-page1-regspace"},
> + { ARM_SMMU_OPT_SHARED_IRQ, "cavium,cn9900-broken-unique-irqline"},
> { 0, NULL},
> };
>
> @@ -1313,6 +1315,21 @@ static irqreturn_t arm_smmu_gerror_handler(int irq, void *dev)
> writel(gerror, smmu->base + ARM_SMMU_GERRORN);
> return IRQ_HANDLED;
> }
> +/* Shared irq handler*/
> +static irqreturn_t arm_smmu_shared_irq_thread(int irq, void *dev)
> +{
> + struct arm_smmu_device *smmu = dev;
> + irqreturn_t ret;
> +
> + ret = arm_smmu_gerror_handler(irq, dev);
> + if (ret == IRQ_NONE) {
> + arm_smmu_evtq_thread(irq, dev);
> + arm_smmu_cmdq_sync_handler(irq, dev);
> + if (smmu->features & ARM_SMMU_FEAT_PRI)
> + arm_smmu_priq_thread(irq, dev);
> + }
> + return IRQ_HANDLED;
> +}
This isn't quite right, because you're now running low-level handlers (like
the gerror handler) in threaded context. You're better off registering a
low-level handler too (see below) which can kick gerror and cmdq_sync
before unconditionally returning IRQ_WAKE_THREAD.
>
> /* IO_PGTABLE API */
> static void __arm_smmu_tlb_sync(struct arm_smmu_device *smmu)
> @@ -2230,18 +2247,9 @@ static void arm_smmu_setup_msis(struct arm_smmu_device *smmu)
> devm_add_action(dev, arm_smmu_free_msis, dev);
> }
>
> -static int arm_smmu_setup_irqs(struct arm_smmu_device *smmu)
> +static void arm_smmu_setup_unique_irqs(struct arm_smmu_device *smmu)
> {
> - int ret, irq;
> - u32 irqen_flags = IRQ_CTRL_EVTQ_IRQEN | IRQ_CTRL_GERROR_IRQEN;
> -
> - /* Disable IRQs first */
> - ret = arm_smmu_write_reg_sync(smmu, 0, ARM_SMMU_IRQ_CTRL,
> - ARM_SMMU_IRQ_CTRLACK);
> - if (ret) {
> - dev_err(smmu->dev, "failed to disable irqs\n");
> - return ret;
> - }
> + int irq, ret;
>
> arm_smmu_setup_msis(smmu);
>
> @@ -2284,10 +2292,46 @@ static int arm_smmu_setup_irqs(struct arm_smmu_device *smmu)
> if (ret < 0)
> dev_warn(smmu->dev,
> "failed to enable priq irq\n");
> - else
> - irqen_flags |= IRQ_CTRL_PRIQ_IRQEN;
> }
> }
> +}
> +
> +static void arm_smmu_setup_shared_irqs(struct arm_smmu_device *smmu)
> +{
> + int ret, irq;
> +
> + /* Single irq is used for all queues, request single interrupt lines */
> + irq = smmu->evtq.q.irq;
> + if (irq) {
> + ret = devm_request_threaded_irq(smmu->dev, irq, NULL,
As above, stick your low-level handler in instead of NULL here.
> + arm_smmu_shared_irq_thread,
> + IRQF_ONESHOT | IRQF_SHARED,
Why do you need IRQF_SHARED here?
> + "arm-smmu-v3-shared_irq", smmu);
Call this "combined" instead of shared, to avoid confusion with the IRQ
flags.
> + if (ret < 0)
> + dev_warn(smmu->dev, "failed to enable shared irq\n");
> + }
> +}
> +
> +static int arm_smmu_setup_irqs(struct arm_smmu_device *smmu)
> +{
> + int ret;
> + u32 irqen_flags = IRQ_CTRL_EVTQ_IRQEN | IRQ_CTRL_GERROR_IRQEN;
> +
> + /* Disable IRQs first */
> + ret = arm_smmu_write_reg_sync(smmu, 0, ARM_SMMU_IRQ_CTRL,
> + ARM_SMMU_IRQ_CTRLACK);
> + if (ret) {
> + dev_err(smmu->dev, "failed to disable irqs\n");
> + return ret;
> + }
> +
> + if (smmu->options & ARM_SMMU_OPT_SHARED_IRQ)
> + arm_smmu_setup_shared_irqs(smmu);
> + else
> + arm_smmu_setup_unique_irqs(smmu);
I'd rather just have something like:
irq = platform_get_irq_byname(pdev, "combined");
in the arm_smmu_device_probe function. If we find it's there, we use that
in preference to the other interrupts.
Will
next prev parent reply other threads:[~2017-06-20 18:00 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-20 14:17 [PATCH v8 0/3] Cavium ThunderX2 SMMUv3 errata workarounds Geetha sowjanya
2017-06-20 14:17 ` [PATCH v8 1/3] ACPI/IORT: Fixup SMMUv3 resource size for Cavium ThunderX2 SMMUv3 model Geetha sowjanya
2017-06-20 18:01 ` Will Deacon
2017-06-20 19:27 ` Lorenzo Pieralisi
2017-06-20 14:17 ` [PATCH v8 2/3] iommu/arm-smmu-v3: Add workaround for Cavium ThunderX2 erratum #74 Geetha sowjanya
2017-06-20 18:06 ` Will Deacon
2017-06-20 14:17 ` [PATCH v8 3/3] iommu/arm-smmu-v3: Add workaround for Cavium ThunderX2 erratum #126 Geetha sowjanya
2017-06-20 18:00 ` Will Deacon [this message]
2017-06-21 6:39 ` Geetha Akula
2017-06-21 9:08 ` Will Deacon
2017-06-21 9:30 ` Marc Zyngier
2017-06-21 18:19 ` Robert Richter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170620180038.GC28035@arm.com \
--to=will.deacon@arm.com \
--cc=linux-arm-kernel@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).