Re: [PATCH 7/9] irqchip: add a RISC-V PLIC driver

From: Atish Patra
Date: Mon Jul 30 2018 - 23:21:47 EST


On 7/27/18 5:04 PM, Atish Patra wrote:
On 7/26/18 7:38 AM, Christoph Hellwig wrote:
This patch adds a driver for the Platform Level Interrupt Controller (PLIC)
specified as part of the RISC-V supervisor level ISA manual, in the memory
layout implemented by SiFive and qemu.

The PLIC connects global interrupt sources to the local interrupt controller
on each hart.

This driver is based on the driver in the RISC-V tree from Palmer Dabbelt,
but has been almost entirely rewritten since.

Signed-off-by: Christoph Hellwig <hch@xxxxxx>

I tried to boot HighFive Unleashed with the patch series after applying
all the patches from riscv-all branch except timer & irq patches. It
gets stuck pretty early.

Here is my github repo with all the changes:
https://github.com/atishp04/riscv-linux/commits/master_chris_cleanup_hifive

I am still looking into it.


I found the issue. As per PLIC documentation, a hart context is a given privilege mode on a given hart. Thus, cpu context ID & cpu numbers are not same. Here is the PLIC register Maps in U54 core:

Ref: https://static.dev.sifive.com/U54-MC-RVCoreIP.pdf

Memory address for Interrupt enable
Address
0x0C00-2080 Hart 1 M-mode enables
0x0C00 2094 End of Hart 1 M-mode enables

0x0C00-2100 Hart 1 S-mode enables
0x0C00-2114 End of Hart 1 S-mode enables

Memory map Claim/Threshold
Address
0x0C20-1000 4B M-mode priority threshold
0x0C20-1004 4B M-mode claim/complete
0x0C20-2000 4B S-mode priority threshold
0x0C20-2004 4B S-mode claim/complete

The original PLIC patch was calculating based on handle->contextid which will assume numbers on a HighFive Unleashed board as 2 4 6 8.

In this patch, context id is assigned as cpu numbers which will be 1 2 3 4. Thus it will lead to incorrect plic address access as shown below.

CPU1 enable register:
plic: plic_toggle: In for hwirq = [1] ctxid [1] reg = [0x2080]
plic: plic_toggle: In for hwirq = [32] ctxid [1] reg = [0x2084]


Palmer: Did I miss something?

FWIW, here is the boot log.
--------- Boot log -------------------------------------------
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=5
[ 0.000000] NR_IRQS: 0, nr_irqs: 0, preallocated irqs: 0
[ 0.000000] plic: mapped 53 interrupts to 4 (out of 9) handlers.
[ 0.000000] clocksource: riscv_clocksource: mask: 0xffffffffffffffff
max_cycles: 0x1d854df40, max_idle_ns: 3526361616960 ns
[ 0.000000] Calibrating delay loop (skipped), value calculated using
timer frequency.. 2.00 BogoMIPS (lpj=10000)
[ 0.010000] pid_max: default: 32768 minimum: 301
[ 0.010000] Mount-cache hash table entries: 16384 (order: 5, 131072
bytes)
[ 0.020000] Mountpoint-cache hash table entries: 16384 (order: 5,
131072 bytes)
[ 0.020000] Hierarchical SRCU implementation.
[ 0.030000] smp: Bringing up secondary CPUs ...
---
drivers/irqchip/Kconfig | 13 ++
drivers/irqchip/Makefile | 1 +
drivers/irqchip/irq-riscv-plic.c | 219 +++++++++++++++++++++++++++++++
3 files changed, 233 insertions(+)
create mode 100644 drivers/irqchip/irq-riscv-plic.c

diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
index e9233db16e03..5ac6dd922a0d 100644
--- a/drivers/irqchip/Kconfig
+++ b/drivers/irqchip/Kconfig
@@ -372,3 +372,16 @@ config QCOM_PDC
IRQs for Qualcomm Technologies Inc (QTI) mobile chips.
endmenu
+
+config RISCV_PLIC
+ bool "Platform-Level Interrupt Controller"
+ depends on RISCV
+ default y
+ help
+ This enables support for the PLIC chip found in standard RISC-V
+ systems. The PLIC controls devices interrupts and connects them to
+ each core's local interrupt controller. Aside from timer and
+ software interrupts, all other interrupt sources (MSI, GPIO, etc)
+ are subordinate to the PLIC.
+
+ If you don't know what to do here, say Y.
diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index 15f268f646bf..bf9238da8a31 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -87,3 +87,4 @@ obj-$(CONFIG_MESON_IRQ_GPIO) += irq-meson-gpio.o
obj-$(CONFIG_GOLDFISH_PIC) += irq-goldfish-pic.o
obj-$(CONFIG_NDS32) += irq-ativic32.o
obj-$(CONFIG_QCOM_PDC) += qcom-pdc.o
+obj-$(CONFIG_RISCV_PLIC) += irq-riscv-plic.o
diff --git a/drivers/irqchip/irq-riscv-plic.c b/drivers/irqchip/irq-riscv-plic.c
new file mode 100644
index 000000000000..274fe2cba544
--- /dev/null
+++ b/drivers/irqchip/irq-riscv-plic.c
@@ -0,0 +1,219 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2017 SiFive
+ * Copyright (C) 2018 Christoph Hellwig
+ */
+#define pr_fmt(fmt) "plic: " fmt
+#include <linux/cpumask.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/irqchip.h>
+#include <linux/irqdomain.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/platform_device.h>
+#include <linux/spinlock.h>
+
+/*
+ * From the RISC-V Priviledged Spec v1.10:
+ *
+ * Global interrupt sources are assigned small unsigned integer identifiers,
+ * beginning at the value 1. An interrupt ID of 0 is reserved to mean "no
+ * interrupt". Interrupt identifiers are also used to break ties when two or
+ * more interrupt sources have the same assigned priority. Smaller values of
+ * interrupt ID take precedence over larger values of interrupt ID.
+ *
+ * While the RISC-V supervisor spec doesn't define the maximum number of
+ * devices supported by the PLIC, the largest number supported by devices
+ * marked as 'riscv,plic0' (which is the only device type this driver supports,
+ * and is the only extant PLIC as of now) is 1024. As mentioned above, device
+ * 0 is defined to be non-existent so this device really only supports 1023
+ * devices.
+ */
+#define MAX_DEVICES 1024
+#define MAX_CONTEXTS 15872
+

Is there any way we can preserve some of the comments in the original
patch about memory-mapped control registers or at least a reference
where to find the register offset calculations?

IMHO, it is helpful for anybody who is not familiar with the details.
+/*
+ * Each interrupt source has a priority register associated with it.
+ * We always hardwire it to one in Linux.
+ */
+#define PRIORITY_BASE 0
+#define PRIORITY_PER_ID 4
+
+/*
+ * Each hart context has a vector of interupt enable bits associated with it.
+ * There's one bit for each interrupt source.
+ */
+#define ENABLE_BASE 0x2000
+#define ENABLE_PER_HART 0x80
+
+/*
+ * Each hart context has a set of control registers associated with it. Right
+ * now there's only two: a source priority threshold over which the hart will
+ * take an interrupt, and a register to claim interrupts.
+ */
+#define CONTEXT_BASE 0x200000
+#define CONTEXT_PER_HART 0x1000
+#define CONTEXT_THRESHOLD 0x00
+#define CONTEXT_CLAIM 0x04
+
+static void __iomem *plic_regs;
+
+static inline void __iomem *plic_hart_offset(int ctxid)
+{
+ return plic_regs + CONTEXT_BASE + ctxid * CONTEXT_PER_HART;
+}
+
+/*
+ * Protect mask operations on the registers given that we can't assume that
+ * atomic memory operations work on them.
+ */
+static DEFINE_SPINLOCK(plic_toggle_lock);
+
+static inline void plic_toggle(int ctxid, int hwirq, int enable)
+{
+ u32 __iomem *reg = plic_regs + ENABLE_BASE + ctxid * ENABLE_PER_HART;

shouldn't it be
u32 __iomem *reg = plic_regs + ENABLE_BASE + ctxid * ENABLE_PER_HART +
(hwirq / 32) * 4;

Without that change, plic_handle_irq() gets called before irq mapping as
plic was not disabled for that irq.

As per comment in the original patch,
--------------------------------------------------------------
* base + 0x002000: Enable bits for sources 0-31 on context 0
* base + 0x002004: Enable bits for sources 32-63 on context 0
--------------------------------------------------------------

+ u32 hwirq_mask = 1 << (hwirq % 32);
+
+ spin_lock(&plic_toggle_lock);
+ if (enable)
+ writel(readl(reg) | hwirq_mask, reg);
+ else
+ writel(readl(reg) & ~hwirq_mask, reg);
+ spin_unlock(&plic_toggle_lock);
+}
+
+static inline void plic_irq_toggle(struct irq_data *d, int enable)
+{
+ int cpu;
+
+ writel(enable, plic_regs + PRIORITY_BASE + d->hwirq * PRIORITY_PER_ID);
+ for_each_present_cpu(cpu)
+ plic_toggle(cpu, d->hwirq, enable);
+}
+
+static void plic_irq_enable(struct irq_data *d)
+{
+ plic_irq_toggle(d, 1);
+}
+
+static void plic_irq_disable(struct irq_data *d)
+{
+ plic_irq_toggle(d, 0);
+}
+
+static struct irq_chip plic_chip = {
+ .name = "riscv,plic0",
+ /*
+ * There is no need to mask/unmask PLIC interrupts. They are "masked"
+ * by reading claim and "unmasked" when writing it back.
+ */
+ .irq_enable = plic_irq_enable,
+ .irq_disable = plic_irq_disable,
+};
+
+static int plic_irqdomain_map(struct irq_domain *d, unsigned int irq,
+ irq_hw_number_t hwirq)
+{
+ irq_set_chip_and_handler(irq, &plic_chip, handle_simple_irq);
+ irq_set_chip_data(irq, NULL);
+ irq_set_noprobe(irq);
+ return 0;
+}
+
+static const struct irq_domain_ops plic_irqdomain_ops = {
+ .map = plic_irqdomain_map,
+ .xlate = irq_domain_xlate_onecell,
+};
+
+static struct irq_domain *plic_irqdomain;
+
+/*
+ * Handling an interrupt is a two-step process: first you claim the interrupt
+ * by reading the claim register, then you complete the interrupt by writing
+ * that source ID back to the same claim register. This automatically enables
+ * and disables the interrupt, so there's nothing else to do.
+ */
+static void plic_handle_irq(struct pt_regs *regs)
+{
+ void __iomem *claim =
+ plic_hart_offset(smp_processor_id()) + CONTEXT_CLAIM;
+ irq_hw_number_t hwirq;
+
+ csr_clear(sie, SIE_STIE);
+ while ((hwirq = readl(claim))) {
+ int irq = irq_find_mapping(plic_irqdomain, hwirq);
+
+ if (unlikely(irq <= 0)) {
+ pr_warn_ratelimited("can't find mapping for hwirq %lu\n",
+ hwirq);

Ratlimiting the warning message here didn't help as ack_bad_irq() still
print message still flooded the console without any useful info.

Regards,
Atish
+ ack_bad_irq(irq);
+ } else {
+ generic_handle_irq(irq);
+ }
+ writel(hwirq, claim);
+ }
+ csr_set(sie, SIE_STIE);
+}
+
+static int __init plic_init(struct device_node *node,
+ struct device_node *parent)
+{
+ int error = 0, nr_mapped = 0, nr_handlers, cpu;
+ u32 nr_irqs;
+
+ if (plic_regs) {
+ pr_warning("PLIC already present.\n");
+ return -ENXIO;
+ }
+
+ plic_regs = of_iomap(node, 0);
+ if (WARN_ON(!plic_regs))
+ return -EIO;
+
+ error = -EINVAL;
+ of_property_read_u32(node, "riscv,ndev", &nr_irqs);
+ if (WARN_ON(!nr_irqs))
+ goto out_iounmap;
+
+ nr_handlers = of_irq_count(node);
+ if (WARN_ON(!nr_handlers))
+ goto out_iounmap;
+ if (WARN_ON(nr_handlers < num_possible_cpus()))
+ goto out_iounmap;
+
+ error = -ENOMEM;
+ plic_irqdomain = irq_domain_add_linear(node, nr_irqs + 1,
+ &plic_irqdomain_ops, NULL);
+ if (WARN_ON(!plic_irqdomain))
+ goto out_iounmap;
+
+ /*
+ * We assume that each present hart is wire up to the PLIC.
+ * If that isn't the case in the future this code will need to be
+ * modified.
+ */
+ for_each_present_cpu(cpu) {
+ irq_hw_number_t hwirq;
+
+ /* priority must be > threshold to trigger an interrupt */
+ writel(0, plic_hart_offset(cpu) + CONTEXT_THRESHOLD);
In correct context id as explained above.

+ for (hwirq = 1; hwirq <= nr_irqs; ++hwirq)
+ plic_toggle(cpu, hwirq, 0);

In correct context id as explained above.


Regards,
Atish
+ nr_mapped++;
+ }
+
+ pr_info("mapped %d interrupts to %d (out of %d) handlers.\n",
+ nr_irqs, nr_mapped, nr_handlers);
+ set_handle_irq(plic_handle_irq);
+ return 0;
+
+out_iounmap:
+ iounmap(plic_regs);
+ return error;
+}
+
+IRQCHIP_DECLARE(plic0, "riscv,plic0", plic_init);