[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined i
From: |
Edgar E. Iglesias |
Subject: |
Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual. |
Date: |
Thu, 9 Dec 2010 12:06:31 +0100 |
User-agent: |
Mutt/1.5.20 (2009-06-14) |
On Thu, Dec 09, 2010 at 12:03:35PM +0100, Fabien Chouteau wrote:
> On 12/09/2010 11:32 AM, Edgar E. Iglesias wrote:
> > On Mon, Dec 06, 2010 at 10:26:03AM +0100, Fabien Chouteau wrote:
> >>
> >> Signed-off-by: Fabien Chouteau<address@hidden>
> >> ---
> >> hw/grlib_irqmp.c | 416
> >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >> 1 files changed, 416 insertions(+), 0 deletions(-)
> >>
> >> diff --git a/hw/grlib_irqmp.c b/hw/grlib_irqmp.c
> >> new file mode 100644
> >> index 0000000..69e1553
> >> --- /dev/null
> >> +++ b/hw/grlib_irqmp.c
> >> @@ -0,0 +1,416 @@
> >> +/*
> >> + * QEMU GRLIB IRQMP Emulator
> >> + *
> >> + * (Multiprocessor and extended interrupt not supported)
> >> + *
> >> + * Copyright (c) 2010 AdaCore
> >> + *
> >> + * Permission is hereby granted, free of charge, to any person obtaining
> >> a copy
> >> + * of this software and associated documentation files (the "Software"),
> >> to deal
> >> + * in the Software without restriction, including without limitation the
> >> rights
> >> + * to use, copy, modify, merge, publish, distribute, sublicense, and/or
> >> sell
> >> + * copies of the Software, and to permit persons to whom the Software is
> >> + * furnished to do so, subject to the following conditions:
> >> + *
> >> + * The above copyright notice and this permission notice shall be
> >> included in
> >> + * all copies or substantial portions of the Software.
> >> + *
> >> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
> >> EXPRESS OR
> >> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
> >> MERCHANTABILITY,
> >> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> >> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
> >> OTHER
> >> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
> >> ARISING FROM,
> >> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
> >> IN
> >> + * THE SOFTWARE.
> >> + */
> >> +
> >> +#include "sysbus.h"
> >> +#include "cpu.h"
> >> +
> >> +#include "grlib.h"
> >> +
> >> +/* #define DEBUG_IRQ */
> >> +
> >> +#ifdef DEBUG_IRQ
> >> +#define DPRINTF(fmt, ...) \
> >> + do { printf("IRQMP: " fmt , ## __VA_ARGS__); } while (0)
> >> +#else
> >> +#define DPRINTF(fmt, ...)
> >> +#endif
> >> +
> >> +#define IRQMP_MAX_CPU 16
> >> +#define IRQMP_REG_SIZE 256 /* Size of memory mapped registers */
> >> +
> >> +/* Memory mapped register offsets */
> >> +#define LEVEL_OFFSET 0x00
> >> +#define PENDING_OFFSET 0x04
> >> +#define FORCE0_OFFSET 0x08
> >> +#define CLEAR_OFFSET 0x0C
> >> +#define MP_STATUS_OFFSET 0x10
> >> +#define BROADCAST_OFFSET 0x14
> >> +#define MASK_OFFSET 0x40
> >> +#define FORCE_OFFSET 0x80
> >> +#define EXTENDED_OFFSET 0xC0
> >> +
> >> +typedef struct IRQMP
> >> +{
> >> + SysBusDevice busdev;
> >> +
> >> + CPUSPARCState *env;
> >> +} IRQMP;
> >> +
> >> +typedef struct IRQMPState
> >> +{
> >> + uint32_t level;
> >> + uint32_t pending;
> >> + uint32_t clear;
> >> + uint32_t broadcast;
> >> +
> >> + uint32_t mask[IRQMP_MAX_CPU];
> >> + uint32_t force[IRQMP_MAX_CPU];
> >> + uint32_t extended[IRQMP_MAX_CPU];
> >> +
> >> + IRQMP *parent;
> >> +} IRQMPState;
> >> +
> >> +IRQMPState grlib_irqmp_state;
> >> +
> >> +void grlib_irqmp_set_irq(void *opaque, int irq, int level);
> >> +
> >> +DeviceState *grlib_irqmp_create(target_phys_addr_t base,
> >> + CPUState *env,
> >> + qemu_irq **cpu_irqs,
> >> + uint32_t nr_irqs)
> >> +{
> >> + DeviceState *dev;
> >> +
> >> + assert(cpu_irqs != NULL);
> >> +
> >> + dev = qdev_create(NULL, "grlib,irqmp");
> >> + qdev_prop_set_ptr(dev, "cpustate", env);
> >> +
> >> + if (qdev_init(dev)) {
> >> + return NULL;
> >> + }
> >> +
> >> + sysbus_mmio_map(sysbus_from_qdev(dev), 0, base);
> >> +
> >> + *cpu_irqs = qemu_allocate_irqs(grlib_irqmp_set_irq,
> >> +&grlib_irqmp_state,
> >> + nr_irqs);
> >> +
> >> + return dev;
> >> +}
> >> +
> >> +static void grlib_irqmp_check_irqs(CPUState *env)
> >> +{
> >> + uint32_t pend = 0;
> >> + uint32_t level0 = 0;
> >> + uint32_t level1 = 0;
> >> +
> >> + assert(env != NULL);
> >> +
> >> + /* IRQ for CPU 0 (no SMP support) */
> >> + pend = (grlib_irqmp_state.pending | grlib_irqmp_state.force[0])
> >> +& grlib_irqmp_state.mask[0];
> >> +
> >> +
> >> + level0 = pend& ~grlib_irqmp_state.level;
> >> + level1 = pend& grlib_irqmp_state.level;
> >> +
> >> + DPRINTF("pend:0x%04x force:0x%04x mask:0x%04x lvl1:0x%04x
> >> lvl0:0x%04x\n",
> >> + grlib_irqmp_state.pending, grlib_irqmp_state.force[0],
> >> + grlib_irqmp_state.mask[0], level1, level0);
> >> +
> >> + /* Trigger level1 interrupt first and level0 if there is no level1 */
> >> + if (level1 != 0) {
> >> + env->pil_in = level1;
> >> + } else {
> >> + env->pil_in = level0;
> >> + }
> >> +
> >> + if (env->pil_in&& (env->interrupt_index == 0 ||
> >> + (env->interrupt_index& ~15) == TT_EXTINT)) {
> >> + unsigned int i;
> >> +
> >> + for (i = 15; i> 0; i--) {
> >> + if (env->pil_in& (1<< i)) {
> >> + int old_interrupt = env->interrupt_index;
> >> +
> >> + env->interrupt_index = TT_EXTINT | i;
> >> + if (old_interrupt != env->interrupt_index) {
> >> + DPRINTF("Set CPU IRQ %d\n", i);
> >> + cpu_interrupt(env, CPU_INTERRUPT_HARD);
> >> + }
> >> + break;
> >> + }
> >> + }
> >> + } else if (!env->pil_in&& (env->interrupt_index& ~15) == TT_EXTINT)
> >> {
> >> + DPRINTF("Reset CPU IRQ %d\n", env->interrupt_index& 15);
> >> + env->interrupt_index = 0;
> >> + cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
> >> + }
> >> +}
> >> +
> >> +void grlib_irqmp_ack(CPUSPARCState *env, int intno)
> >> +{
> >> + assert(env != NULL);
> >> +
> >> + uint32_t mask;
> >> +
> >> + intno&= 15;
> >> + mask = 1<< intno;
> >> +
> >> + DPRINTF("grlib_irqmp_ack %d\n", intno);
> >> +
> >> + /* Clear registers */
> >> + grlib_irqmp_state.pending&= ~mask;
> >> + grlib_irqmp_state.force[0]&= ~mask; /* Only CPU 0 (No SMP support) */
> >> +
> >> + grlib_irqmp_check_irqs(env);
> >> +}
> >> +
> >> +void grlib_irqmp_set_irq(void *opaque, int irq, int level)
> >> +{
> >> + IRQMPState *s = opaque;
> >> + int i = 0;
> >> +
> >> + assert(opaque != NULL);
> >> + assert(s->parent != NULL);
> >> +
> >> + if (level) {
> >> + DPRINTF("Raise CPU IRQ %d\n", irq);
> >> +
> >> + if (s->broadcast& 1<< irq) {
> >> + /* Broadcasted IRQ */
> >> + for (i = 0; i< IRQMP_MAX_CPU; i++) {
> >> + s->force[i] |= 1<< irq;
> >> + }
> >> + } else {
> >> + s->pending |= 1<< irq;
> >> + }
> >> + grlib_irqmp_check_irqs(s->parent->env);
> >> +
> >> + } else {
> >> +
> >> + DPRINTF("Lower CPU IRQ %d\n", irq);
> >> + if (s->broadcast& 1<< irq) {
> >> + /* Broadcasted IRQ */
> >> + for (i = 0; i< IRQMP_MAX_CPU; i++) {
> >> + s->force[i]&= ~(1<< irq);
> >> + }
> >> + } else {
> >> + s->pending&= ~(1<< irq);
> >> + }
> >
> > If you use the edge triggered interrupt model in the devices, then you
> > shouldn't clear the pending bit here. A pulse from the device should
> > set it and it should only get cleared when the CPU acks it.
> >
> > The model you've coded here indicates that the devices use a level
> > triggered approach. And the clearing of the pending bit in
> > grlib_irqmp_ack becomes meaningless...
>
> OK, so I use qemu_irq_pulse and do nothing in grlib_irqmp_set_irq when
> level == 0...
Right. But I'm still not convinced about the edge triggered nature of the
device irq signals.
One question to ask is:
What happens if you setup a timer without auto-reload and let an
interrupt hit. In the interrupt handler you dont restart the timer
and you don't clear the IP bit. You just return.
Will you get re-hit by the timer interrupt over and over?
If yes, the device interrupt line is connected to the IP bit and
needs to be acked by writing a 1 to the config reg. You shouldn't
use qemu_irq_pulse.
If no, the device interrupt is signaled as a pulse.
Cheers
- Re: [Qemu-devel] [PATCH 4/6] [RFC] Header file for the GRLIB components., (continued)
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Blue Swirl, 2010/12/06
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Fabien Chouteau, 2010/12/07
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Blue Swirl, 2010/12/11
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Fabien Chouteau, 2010/12/13
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Blue Swirl, 2010/12/13
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Edgar E. Iglesias, 2010/12/09
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Fabien Chouteau, 2010/12/09
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual.,
Edgar E. Iglesias <=
- Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual., Fabien Chouteau, 2010/12/09
Re: [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer as defined in GRLIB IP Core User's Manual., Blue Swirl, 2010/12/06
Re: [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer as defined in GRLIB IP Core User's Manual., Edgar E. Iglesias, 2010/12/08
Re: [Qemu-devel] [PATCH 0/6] [RFC] New SPARC machine: Leon3, Artyom Tarasenko, 2010/12/06