From: Fabien Chouteau <chouteau@adacore.com>
To: Blue Swirl <blauwirbel@gmail.com>
Cc: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer as defined in GRLIB IP Core User's Manual.
Date: Tue, 07 Dec 2010 10:55:33 +0100 [thread overview]
Message-ID: <4CFE0495.6060906@adacore.com> (raw)
In-Reply-To: <AANLkTimBn3PAfkOY8aEH9WzrwdFk9juXwHO5CDpfaF-n@mail.gmail.com>
On 12/06/2010 06:12 PM, Blue Swirl wrote:
> On Mon, Dec 6, 2010 at 9:26 AM, Fabien Chouteau<chouteau@adacore.com> wrote:
>>
>> Signed-off-by: Fabien Chouteau<chouteau@adacore.com>
>> ---
>> hw/grlib_gptimer.c | 448 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>> 1 files changed, 448 insertions(+), 0 deletions(-)
>>
>> diff --git a/hw/grlib_gptimer.c b/hw/grlib_gptimer.c
>> new file mode 100644
>> index 0000000..41edbe4
>> --- /dev/null
>> +++ b/hw/grlib_gptimer.c
>> @@ -0,0 +1,448 @@
>> +/*
>> + * QEMU GRLIB GPTimer Emulator
>> + *
>> + * 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 "qemu-timer.h"
>> +
>> +#include "grlib.h"
>> +
>> +/* #define DEBUG_TIMER */
>
> The usual convention is
> //#define DEBUG_TIMER
> for easy editing.
>
Actually, it's easier for me with the /* */, but OK.
> However, very often the much more powerful tracepoints can replace
> debug statements.
>
>> +
>> +#ifdef DEBUG_TIMER
>> +#define DPRINTF(fmt, ...) \
>> + do { printf("GPTIMER: " fmt , ## __VA_ARGS__); } while (0)
>> +#else
>> +#define DPRINTF(fmt, ...)
>> +#endif
>> +
>> +#define UNIT_REG_SIZE 16 /* Size of memory mapped regs for the unit */
>> +#define GPTIMER_REG_SIZE 16 /* Size of memory mapped regs for a GPTimer */
>> +
>> +#define GPTIMER_MAX_TIMERS 8
>> +
>> +/* GPTimer Config register fields */
>> +#define GPTIMER_ENABLE (1<< 0)
>> +#define GPTIMER_RESTART (1<< 1)
>> +#define GPTIMER_LOAD (1<< 2)
>> +#define GPTIMER_INT_ENABLE (1<< 3)
>> +#define GPTIMER_INT_PENDING (1<< 4)
>> +#define GPTIMER_CHAIN (1<< 5) /* Not supported */
>> +#define GPTIMER_DEBUG_HALT (1<< 6) /* Not supported */
>> +
>> +/* Memory mapped register offsets */
>> +#define SCALER_OFFSET 0x00
>> +#define SCALER_RELOAD_OFFSET 0x04
>> +#define CONFIG_OFFSET 0x08
>> +#define COUNTER_OFFSET 0x00
>> +#define COUNTER_RELOAD_OFFSET 0x04
>> +#define TIMER_BASE 0x10
>> +
>> +typedef struct GPTimer GPTimer;
>> +typedef struct GPTimerUnit GPTimerUnit;
>> +
>> +struct GPTimer
>> +{
>> + QEMUBH *bh;
>> + struct ptimer_state *ptimer;
>> +
>> + qemu_irq irq;
>> + int id;
>> + GPTimerUnit *unit;
>> +
>> + /* registers */
>> + uint32_t counter;
>> + uint32_t reload;
>> + uint32_t config;
>> +};
>> +
>> +struct GPTimerUnit
>> +{
>> + SysBusDevice busdev;
>> +
>> + uint32_t nr_timers; /* Number of timers available */
>> + uint32_t freq_hz; /* System frequency */
>> + uint32_t irq_line; /* Base irq line */
>> +
>> + GPTimer *timers;
>> +
>> + /* registers */
>> + uint32_t scaler;
>> + uint32_t reload;
>> + uint32_t config;
>> +};
>> +
>> +DeviceState *grlib_gptimer_create(target_phys_addr_t base,
>> + uint32_t nr_timers,
>> + uint32_t freq,
>> + qemu_irq *cpu_irqs,
>> + int base_irq)
>
> This function belongs to leon3.c.
I don't see why. GPTimer is a peripheral and you may want to use it in
an other system.
>> +{
>> + DeviceState *dev;
>> + int i;
>> +_ir
>> + dev = qdev_create(NULL, "grlib,gptimer");
>> + qdev_prop_set_uint32(dev, "nr-timers", nr_timers);
>> + qdev_prop_set_uint32(dev, "frequency", freq);
>> + qdev_prop_set_uint32(dev, "irq-line", base_irq);
>
> Base irq is not device property, but part of board configuration. Thus
> leon3.c should just pass&cpu_irqs[base_irq] to this function.
>
I need this property to put the IRQ line in the configuration register.
Is there a way to get this number from a qemu_irq structure?
>> +
>> + if (qdev_init(dev)) {
>> + return NULL;
>> + }
>> +
>> + sysbus_mmio_map(sysbus_from_qdev(dev), 0, base);
>> +
>> + for (i = 0; i< nr_timers; i++)
>> + sysbus_connect_irq(sysbus_from_qdev(dev), i, cpu_irqs[base_irq + i]);
>> +
>> + return dev;
>> +}
>> +
>> +static void grlib_gptimer_enable(GPTimer *timer)
>> +{
>> + assert(timer != NULL);
>> +
>> + DPRINTF("%s id:%d\n", __func__, timer->id);
>> +
>> + ptimer_stop(timer->ptimer);
>> +
>> + if (!(timer->config& GPTIMER_ENABLE)) {
>> + /* Timer disabled */
>> + DPRINTF("%s id:%d Timer disabled (config 0x%x)\n", __func__,
>> + timer->id, timer->config);
>> + return;
>> + }
>> +
>> + /* ptimer is triggered when the counter reach 0 but GPTimer is triggered at
>> + underflow. Set count + 1 to simulate the GPTimer behavior. */
>> +
>> + DPRINTF("%s id:%d set count 0x%x and run\n",
>> + __func__,
>> + timer->id,
>> + timer->counter + 1);
>> +
>> + ptimer_set_count(timer->ptimer, timer->counter + 1);
>> + ptimer_run(timer->ptimer, 1);
>> +}
>> +
>> +static void grlib_gptimer_restart(GPTimer *timer)
>> +{
>> + assert(timer != NULL);
>> +
>> + DPRINTF("%s id:%d reload val: 0x%x\n", __func__, timer->id, timer->reload);
>> +
>> + timer->counter = timer->reload;
>> + grlib_gptimer_enable(timer);
>> +}
>> +
>> +static void grlib_gptimer_set_scaler(GPTimerUnit *unit, uint32_t scaler)
>> +{
>> + int i = 0;
>> + uint32_t value = 0;
>> +
>> + assert(unit != NULL);
>> +
>> +
>> + if (scaler> 0) {
>> + value = unit->freq_hz / (scaler + 1);
>> + } else {
>> + value = unit->freq_hz;
>> + }
>> +
>> + DPRINTF("%s scaler:%d freq:0x%x\n", __func__, scaler, value);
>> +
>> + for (i = 0; i< unit->nr_timers; i++) {
>> + ptimer_set_freq(unit->timers[i].ptimer, value);
>> + }
>> +}
>> +
>> +static void grlib_gptimer_hit(void *opaque)
>> +{
>> + GPTimer *timer = opaque;
>> + assert(timer != NULL);
>> +
>> + DPRINTF("%s id:%d\n", __func__, timer->id);
>> +
>> + /* Timer expired */
>> +
>> + if (timer->config& GPTIMER_INT_ENABLE) {
>> + /* Set the pending bit (only unset by write in the config register) */
>> + timer->config&= GPTIMER_INT_PENDING;
>> + qemu_set_irq(timer->irq, 1);
>> + }
>> +
>> + if (timer->config& GPTIMER_RESTART) {
>> + grlib_gptimer_restart(timer);
>> + }
>> +}
>> +
>> +static uint32_t grlib_gptimer_readl (void *opaque, target_phys_addr_t addr)
>
> Extra space between function name and its arguments.
Fixed.
>
>> +{
>> + GPTimerUnit *unit = opaque;
>> + uint32_t value = 0;
>> +
>> + addr&= 0xff;
>> +
>> + assert(unit != NULL);
>> +
>> + /* Unit registers */
>> + switch (addr)
>> + {
>> + case SCALER_OFFSET:
>> + DPRINTF("%s scaler: 0x%x\n", __func__, unit->scaler);
>> + return unit->scaler;
>> +
>> + case SCALER_RELOAD_OFFSET:
>> + DPRINTF("%s reload: 0x%x\n", __func__, unit->reload);
>> + return unit->reload;
>> +
>> + case CONFIG_OFFSET:
>> + DPRINTF("%s unit config: 0x%x\n", __func__, unit->config);
>> + return unit->config;
>> +
>> + default:
>> + break;
>> + }
>> +
>> + target_phys_addr_t timer_addr = (addr % TIMER_BASE);
>> + int id = (addr - TIMER_BASE) / TIMER_BASE;
>
> Variables should be declared at the beginning of the function. But I'd
> actually make two functions and use the first one for the general
> registers above and second one for the timer registers below. Then the
> opaque passed to the latter could be GPTimer instead of GPTimerUnit.
>
If you want, I can put the variables at the beginning of the function,
but I think that split the function will just obfuscate the code.
>> +
>> + if (id>= 0&& id< unit->nr_timers) {
>> +
>> + /* GPTimer registers */
>> + switch (timer_addr)
>> + {
>> + case COUNTER_OFFSET:
>> + value = ptimer_get_count (unit->timers[id].ptimer);
>> + DPRINTF("%s counter value for timer %d: 0x%x\n",
>> + __func__, id, value);
>> + return value;
>> +
>> + case COUNTER_RELOAD_OFFSET:
>> + value = unit->timers[id].reload;
>> + DPRINTF("%s reload value for timer %d: 0x%x\n",
>> + __func__, id, value);
>> + return value;
>> +
>> + case CONFIG_OFFSET:
>> + DPRINTF("%s config for timer %d: 0x%x\n",
>> + __func__, id, unit->timers[id].config);
>> + return unit->timers[id].config;
>> +
>> + default:
>> + break;
>> + }
>> +
>> + }
>> +
>> + DPRINTF("read unknown register 0x%04x\n", (int)addr);
>> + return 0;
>> +}
>> +
>> +static void
>> +grlib_gptimer_writel (void *opaque, target_phys_addr_t addr, uint32_t value)
>> +{
>> + GPTimerUnit *unit = opaque;
>> +
>> + addr&= 0xff;
>> +
>> + assert(unit != NULL);
>> +
>> + /* Unit registers */
>> + switch (addr)
>> + {
>> + case SCALER_OFFSET:
>> + value&= 0xFFFF; /* clean up the value */
>> + unit->scaler = value;
>> + return;
>> +
>> + case SCALER_RELOAD_OFFSET:
>> + value&= 0xFFFF; /* clean up the value */
>> + unit->reload = value;
>> + grlib_gptimer_set_scaler(unit, value);
>> + return;
>> +
>> + case CONFIG_OFFSET:
>> + /* Read Only (disable timer freeze not supported) */
>> + return;
>> +
>> + default:
>> + break;
>> + }
>> +
>> + target_phys_addr_t timer_addr = (addr % TIMER_BASE);
>> + int id = (addr - TIMER_BASE) / TIMER_BASE;
>> +
>> + if (id>= 0&& id< unit->nr_timers) {
>> +
>> + /* GPTimer registers */
>> + switch (timer_addr)
>> + {
>> + case COUNTER_OFFSET:
>> + DPRINTF("%s counter value for timer %d: 0x%x\n",
>> + __func__, id, value);
>> + unit->timers[id].counter = value;
>> + grlib_gptimer_enable(&unit->timers[id]);
>> + return;
>> +
>> + case COUNTER_RELOAD_OFFSET:
>> + DPRINTF("%s reload value for timer %d: 0x%x\n",
>> + __func__, id, value);
>> + unit->timers[id].reload = value;
>> + return;
>> +
>> + case CONFIG_OFFSET:
>> + DPRINTF("%s config for timer %d: 0x%x\n", __func__, id, value);
>> +
>> + unit->timers[id].config = value;
>> +
>> + /* gptimer_restart calls gptimer_enable, so if "enable" and
>> + "load" bits are present, we just have to call restart. */
>> +
>> + if (value& GPTIMER_LOAD) {
>> + grlib_gptimer_restart(&unit->timers[id]);
>> + } else if (value& GPTIMER_ENABLE) {
>> + grlib_gptimer_enable(&unit->timers[id]);
>> + }
>> +
>> + /* This fields must always be read as 0 */
>> + value&= ~(GPTIMER_LOAD& GPTIMER_DEBUG_HALT);
>> +
>> + unit->timers[id].config = value;
>> + return;
>> +
>> + default:
>> + break;
>> + }
>> +
>> + }
>> +
>> + DPRINTF("write unknown register 0x%04x\n", (int)addr);
>> +}
>> +
>> +static CPUReadMemoryFunc *grlib_gptimer_read[] = {
>
> 'const'
>
Fixed.
>> + NULL, NULL, grlib_gptimer_readl,
>> +};
>> +
>> +static CPUWriteMemoryFunc *grlib_gptimer_write[] = {
>> + NULL, NULL, grlib_gptimer_writel,
>> +};
>> +
>> +static void grlib_gptimer_reset(void *opaque)
>> +{
>> + /* int i = 0; */
>> + /* GPTimerUnit *unit = (GPTimerUnit *)opaque; */
>> + /* assert(unit != NULL); */
>> +
>> + /* unit->scaler = 0; */
>> + /* unit->reload = 0; */
>> + /* unit->config = 0; */
>> +
>> + /* unit->config = unit->nr_timers; */
>> + /* unit->config |= unit->irq_line<< 3; */
>> + /* unit->config |= 1<< 8; /\* separate interrupt *\/ */
>> + /* unit->config |= 1<< 9; /\* Disable timer freeze *\/ */
>> +
>> +
>> + /* for (i = 0; i< unit->nr_timers; i++) { */
>> + /* GPTimer *timer =&unit->timers[i]; */
>> +
>> + /* timer->counter = 0; */
>> + /* timer->reload = 0; */
>> + /* timer->config = 0; */
>> + /* ptimer_stop(timer->ptimer); */
>> + /* ptimer_set_count(timer->ptimer, 0); */
>> + /* ptimer_set_freq(timer->ptimer, unit->freq_hz); */
>> + /* } */
>
> Why dead code?
No reason, this is just a mistake, the code should be un-commented.
>
>> +
>> +}
>> +
>> +static int grlib_gptimer_init(SysBusDevice *dev)
>> +{
>> + GPTimerUnit *unit = FROM_SYSBUS(typeof (*unit), dev);
>> + unsigned int i;
>> + int timer_regs;
>> +
>> + assert(unit->nr_timers> 0);
>> + assert(unit->nr_timers<= GPTIMER_MAX_TIMERS);
>> + unit->timers = qemu_mallocz(sizeof unit->timers[0] * unit->nr_timers);
>> +
>> + for (i = 0; i< unit->nr_timers; i++) {
>> + GPTimer *timer =&unit->timers[i];
>> +
>> + timer->unit = unit;
>> + timer->bh = qemu_bh_new(grlib_gptimer_hit, timer);
>> + timer->ptimer = ptimer_init(timer->bh);
>> + timer->id = i;
>> +
>> + /* One IRQ line for each timer */
>> + sysbus_init_irq(dev,&timer->irq);
>> +
>> + ptimer_set_freq(timer->ptimer, unit->freq_hz);
>> + }
>> +
>> + qemu_register_reset(grlib_gptimer_reset, unit);
>
> This can be replaced by qdev.reset field.
All right, it's clearer indeed.
--
Fabien Chouteau
next prev parent reply other threads:[~2010-12-07 9:56 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
2010-12-06 17:12 ` [Qemu-devel] [PATCH 1/6] [RFC] Emulation of GRLIB GPTimer " Blue Swirl
2010-12-07 9:55 ` Fabien Chouteau [this message]
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=4CFE0495.6060906@adacore.com \
--to=chouteau@adacore.com \
--cc=blauwirbel@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).