From: Fabien Chouteau <chouteau@adacore.com>
To: "Edgar E. Iglesias" <edgar.iglesias@gmail.com>
Cc: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP as defined in GRLIB IP Core User's Manual.
Date: Thu, 09 Dec 2010 12:32:03 +0100 [thread overview]
Message-ID: <4D00BE33.4030707@adacore.com> (raw)
In-Reply-To: <20101209110631.GD20208@edde.se.axis.com>
On 12/09/2010 12:06 PM, Edgar E. Iglesias wrote:
> 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<chouteau@adacore.com>
>>>> ---
>>>> 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?
No, just one time.
> 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
Let's go with the pulse ;)
--
Fabien Chouteau
next prev parent reply other threads:[~2010-12-09 11:32 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-12-06 9:26 [Qemu-devel] [PATCH 0/6] [RFC] New SPARC machine: Leon3 Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer as defined in GRLIB IP Core User's Manual Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP " Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 3/6] [RFC] Emulation of GRLIB APB UART " Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 4/6] [RFC] Header file for the GRLIB components Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 5/6] [RFC] Emulation of Leon3 Fabien Chouteau
2010-12-06 9:26 ` [Qemu-devel] [PATCH 6/6] [RFC] SPARCV8 asr17 register support Fabien Chouteau
2010-12-06 18:01 ` Blue Swirl
2010-12-07 11:51 ` Fabien Chouteau
2010-12-11 9:59 ` Blue Swirl
2010-12-13 17:01 ` Fabien Chouteau
2010-12-06 17:53 ` [Qemu-devel] [PATCH 5/6] [RFC] Emulation of Leon3 Blue Swirl
2010-12-07 11:40 ` Fabien Chouteau
2010-12-11 9:56 ` Blue Swirl
2010-12-13 15:51 ` Fabien Chouteau
2010-12-13 18:18 ` Blue Swirl
2010-12-15 17:47 ` Fabien Chouteau
2010-12-17 19:14 ` Blue Swirl
2010-12-20 6:46 ` Edgar E. Iglesias
2010-12-20 9:40 ` Fabien Chouteau
2010-12-20 20:09 ` Blue Swirl
2010-12-20 9:25 ` Fabien Chouteau
2010-12-20 19:27 ` Blue Swirl
2010-12-12 14:41 ` Andreas Färber
2010-12-13 17:00 ` Fabien Chouteau
2010-12-06 17:31 ` [Qemu-devel] [PATCH 4/6] [RFC] Header file for the GRLIB components Blue Swirl
2010-12-07 11:04 ` Fabien Chouteau
2010-12-06 17:29 ` [Qemu-devel] [PATCH 3/6] [RFC] Emulation of GRLIB APB UART as defined in GRLIB IP Core User's Manual Blue Swirl
2010-12-07 10:55 ` Fabien Chouteau
2010-12-06 17:25 ` [Qemu-devel] [PATCH 2/6] [RFC] Emulation of GRLIB IRQMP " Blue Swirl
2010-12-07 10:43 ` Fabien Chouteau
2010-12-11 10:31 ` Blue Swirl
2010-12-13 16:23 ` Fabien Chouteau
2010-12-13 18:13 ` Blue Swirl
2010-12-09 10:32 ` Edgar E. Iglesias
2010-12-09 11:03 ` Fabien Chouteau
2010-12-09 11:06 ` Edgar E. Iglesias
2010-12-09 11:32 ` Fabien Chouteau [this message]
2010-12-06 17:12 ` [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer " Blue Swirl
2010-12-07 9:55 ` Fabien Chouteau
2010-12-08 8:30 ` Edgar E. Iglesias
2010-12-08 9:39 ` Fabien Chouteau
2010-12-08 21:02 ` Edgar E. Iglesias
2010-12-08 22:51 ` Edgar E. Iglesias
2010-12-09 10:04 ` Fabien Chouteau
2010-12-09 10:22 ` Edgar E. Iglesias
2010-12-06 10:44 ` [Qemu-devel] [PATCH 0/6] [RFC] New SPARC machine: Leon3 Artyom Tarasenko
2010-12-06 15:07 ` Fabien Chouteau
2010-12-06 18:12 ` Blue Swirl
2010-12-07 17:43 ` Fabien Chouteau
2010-12-06 18:05 ` Blue Swirl
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=4D00BE33.4030707@adacore.com \
--to=chouteau@adacore.com \
--cc=edgar.iglesias@gmail.com \
--cc=qemu-devel@nongnu.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).