Re: [Qemu-devel] [PATCH 16/18] hw: add QEMU model for Faraday RTCtimer

[蘇國榮]

Sorry for the inconveniences, our e-mail server has a limitation on the rate of 
sending mails. 
It prevents the git-send-email to do that for me. I'll try to use gmail later 
when I get the QOM issues fixed.

Best Regards
Dante Su

-----Original Message-----
From: address@hidden [mailto:address@hidden On Behalf Of Andreas Farber
Sent: Friday, January 18, 2013 6:43 PM
To: Dante Kuo-Jung Su(蘇國榮)
Cc: address@hidden; address@hidden; Paolo Bonzini; address@hidden; KONRAD 
Frédéric
Subject: Re: [Qemu-devel] [PATCH 16/18] hw: add QEMU model for Faraday RTCtimer

Kuo-Jung, please thread your messages together (e.g., using
git-send-email) and prepend a cover letter, right now this is a badly 
reviewable mess of individual patches on the list.

Am 18.01.2013 09:44, schrieb KONRAD Frédéric:
> On 18/01/2013 07:32, Dante wrote:
>> Signed-off-by: Kuo-Jung Su <address@hidden>
>> ---
>>   hw/ftrtc011.c |  308
>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 308 insertions(+)
>>   create mode 100644 hw/ftrtc011.c
>>
>> diff --git a/hw/ftrtc011.c b/hw/ftrtc011.c new file mode 100644 index 
>> 0000000..466cbb6
>> --- /dev/null
>> +++ b/hw/ftrtc011.c
>> @@ -0,0 +1,308 @@
>> +/*
>> + * QEMU model of the FTRTC011 RTC Timer
>> + *
>> + * Copyright (C) 2012 Faraday Technology
>> + * Copyright (C) 2012 Dante Su <address@hidden>
>> + *
>> + * 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 "sysemu/sysemu.h"
>> +#include "qemu/timer.h"
>> +
>> +/* Hardware registers */
>> +#define REG_SEC            0x00
>> +#define REG_MIN            0x04
>> +#define REG_HOUR        0x08
>> +#define REG_DAY            0x0C
>> +
>> +#define REG_ALARM_SEC    0x10
>> +#define REG_ALARM_MIN    0x14
>> +#define REG_ALARM_HOUR    0x18
>> +
>> +#define REG_CR            0x20
>> +#define REG_WSEC        0x24
>> +#define REG_WMIN        0x28
>> +#define REG_WHOUR        0x2C
>> +#define REG_WDAY        0x30
>> +#define REG_ISR            0x34
>> +
>> +#define REG_REV            0x3C
>> +#define REG_CURRENT        0x44

You would be well advised to put these constants into their own header file so 
that they can be reused for qtest test cases. Please take a look at the 
existing rtc code and test cases.

>> +
>> +enum ftrtc011_irqpin {
>> +    IRQ_ALARM_LEVEL = 0,
>> +    IRQ_ALARM_EDGE,
>> +    IRQ_SEC,
>> +    IRQ_MIN,
>> +    IRQ_HOUR,
>> +    IRQ_DAY,
>> +};
>> +
>> +typedef struct {

Please name the struct, usually like the typedef.

>> +    SysBusDevice busdev;

parent_obj please and please separate from the remaining fields.

>> +    MemoryRegion mmio;
>> +
>> +    qemu_irq irq[6];
>> +
>> +    QEMUTimer *qtimer;
>> +
>> +    uint8_t sec;
>> +    uint8_t min;
>> +    uint8_t hr;
>> +    uint32_t day;
>> +
>> +    uint8_t alarm_sec;
>> +    uint8_t alarm_min;
>> +    uint8_t alarm_hr;
>> +
>> +    uint32_t cr;
>> +    uint32_t isr;
>> +
>> +} ftrtc011_state;

CamelCase please.

These comments may apply to other patches in the series as well, please check 
on your own.

>> +
>> +/* Update interrupts.  */
>> +static inline void ftrtc011_update_irq(ftrtc011_state *s) {
>> +    uint32_t mask = ((s->cr >> 1) & 0x1f) & s->isr;
>> +
>> +    qemu_set_irq(s->irq[IRQ_ALARM_LEVEL], (mask & 0x10) ? 1 : 0);
>> +
>> +    if (mask) {
>> +        if (mask & 0x01)

Please use scripts/checkpatch.pl, it will complain about missing braces for if 
statements. You can automate this as a commit hook:
http://blog.vmsplice.net/2011/03/how-to-automatically-run-checkpatchpl.html

>> +            qemu_irq_pulse(s->irq[IRQ_SEC]);
>> +        if (mask & 0x02)
>> +            qemu_irq_pulse(s->irq[IRQ_MIN]);
>> +        if (mask & 0x04)
>> +            qemu_irq_pulse(s->irq[IRQ_HOUR]);
>> +        if (mask & 0x08)
>> +            qemu_irq_pulse(s->irq[IRQ_DAY]);
>> +        if (mask & 0x10)
>> +            qemu_irq_pulse(s->irq[IRQ_ALARM_EDGE]);
>> +    }
>> +}
>> +
>> +static uint64_t ftrtc011_mem_read(void *opaque, hwaddr addr, 
>> +unsigned
>> int size)
>> +{
>> +    ftrtc011_state *s = opaque;
>> +    uint32_t rc = 0;
>> +
>> +    switch (addr) {
>> +    case REG_SEC:
>> +        return s->sec;
>> +    case REG_MIN:
>> +        return s->min;
>> +    case REG_HOUR:
>> +        return s->hr;
>> +    case REG_DAY:
>> +        return s->day;
>> +    case REG_ALARM_SEC:
>> +        return s->alarm_sec;
>> +    case REG_ALARM_MIN:
>> +        return s->alarm_min;
>> +    case REG_ALARM_HOUR:
>> +        return s->alarm_hr;
>> +    case REG_CR:
>> +        return s->cr;
>> +    case REG_ISR:
>> +        return s->isr;
>> +    case REG_REV:
>> +        return 0x00010000;
>> +    case REG_CURRENT:
>> +        return (s->day << 17) | (s->hr << 12) | (s->min << 6) |
>> (s->sec);
>> +    default:
>> +        break;
>> +    }
>> +
>> +    return rc;
>> +}
>> +
>> +static void ftrtc011_mem_write(void *opaque, hwaddr addr, uint64_t
>> val, unsigned int size)
>> +{
>> +    ftrtc011_state *s = opaque;
>> +
>> +    switch (addr) {
>> +    case REG_ALARM_SEC:
>> +        s->alarm_sec = (uint32_t)val;
>> +        break;
>> +    case REG_ALARM_MIN:
>> +        s->alarm_min = (uint32_t)val;
>> +        break;
>> +    case REG_ALARM_HOUR:
>> +        s->alarm_hr = (uint32_t)val;
>> +        break;
>> +    case REG_WSEC:
>> +        s->sec = (uint32_t)val;
>> +        break;
>> +    case REG_WMIN:
>> +        s->min = (uint32_t)val;
>> +        break;
>> +    case REG_WHOUR:
>> +        s->hr = (uint32_t)val;
>> +        break;
>> +    case REG_WDAY:
>> +        s->day = (uint32_t)val;
>> +        break;
>> +    case REG_CR:
>> +        s->cr = (uint32_t)val;
>> +        if (s->cr & 0x01) {
>> +            qemu_mod_timer(s->qtimer, qemu_get_clock_ns(vm_clock) +
>> get_ticks_per_sec());
>> +        } else {
>> +            qemu_del_timer(s->qtimer);
>> +        }
>> +        break;
>> +    case REG_ISR:
>> +        s->isr &= ~((uint32_t)val);
>> +        ftrtc011_update_irq(s);
>> +        break;
>> +    default:
>> +        break;
>> +    }
>> +}
>> +
>> +static const MemoryRegionOps bus_ops = {
>> +    .read  = ftrtc011_mem_read,
>> +    .write = ftrtc011_mem_write,
>> +    .endianness = DEVICE_LITTLE_ENDIAN,
>> +    .valid = {
>> +        .min_access_size = 4,
>> +        .max_access_size = 4
>> +    }
>> +};
>> +
>> +static void ftrtc011_timer_tick(void *opaque) {
>> +    ftrtc011_state *s = (ftrtc011_state *)opaque;
> Didn't that need OBJECT_CHECK for QOM? (See below).

C-Casting an opaque value to the type passed in is perfectly valid, especially 
in the context of timers and interrupts.

My personal preference is to explicitly cast to parent types via QOM macros 
though (in this case SYS_BUS_DEVICE() or DEVICE()).

>> +
>> +    s->sec += 1;
>> +    if (s->cr & (1 << 1))
>> +        s->isr |= (1 << 0);
>> +
>> +    if (s->sec >= 60) {
>> +        s->sec = 0;
>> +        s->min += 1;
>> +        if (s->cr & (1 << 2))
>> +            s->isr |= (1 << 1);
>> +
>> +        if (s->min >= 60) {
>> +            s->min = 0;
>> +            s->hr += 1;
>> +            if (s->cr & (1 << 3))
>> +                s->isr |= (1 << 2);
>> +
>> +            if (s->hr >= 24) {
>> +                s->hr = 0;
>> +                s->day += 1;
>> +                if (s->cr & (1 << 4))
>> +                    s->isr |= (1 << 3);
>> +            }
>> +        }
>> +    }
>> +
>> +    /* alarm interrupt */
>> +    if (s->cr & (1 << 5)) {
>> +        if ((s->sec | (s->min << 8) | (s->hr << 16)) ==
>> +            (s->alarm_sec | (s->alarm_min << 8) | (s->alarm_hr <<
>> 16))) {
>> +            s->isr |= (1 << 4);
>> +        }
>> +    }
>> +
>> +    /* send interrupt signal */
>> +    ftrtc011_update_irq(s);
>> +
>> +    qemu_mod_timer(s->qtimer, qemu_get_clock_ns(vm_clock) +
>> get_ticks_per_sec());
>> +}
>> +
>> +static void ftrtc011_reset(DeviceState *d) {
>> +    ftrtc011_state *s = DO_UPCAST(ftrtc011_state, busdev.qdev, d);
> As far as I understand this is not QOM (someone correct me if I'm wrong):
> 
> You must introduce QOM macros in header:
> 
> #define TYPE_FTRTC011 "ftrtc011"
> #define TRTC011_GET_CLASS(obj) \
>         OBJECT_GET_CLASS(ftrtc011_state, obj, TYPE_FTRTC011) #define 
> FTRTC011_CLASS(klass) \
>         OBJECT_CLASS_CHECK(ftrtc011_state, klass, TYPE_FTRTC011) 
> #define FTRTC011(obj) \
>         OBJECT_CHECK(ftrtc011_state, (obj), TYPE_FTRTC011)
> 
> Then use FTRTC011(d) instead of DO_UPCAST(...)

Correct. The TYPE_FTRTC011 should then also be used for the TypeInfo .name 
below. If it is used from some machine or other device, it should go into a 
header so that it can be reused there.

The general rationale is to not depend on implementation details related to our 
class/object inheritence (parent_obj vs. qdev vs. busdev vs. dev
etc.) and to remain compatible with switching to an object-oriented language in 
the future.

> 
>> +
>> +    s->sec = 0;
>> +    s->min = 0;
>> +    s->hr  = 0;
>> +    s->day = 0;
>> +    s->alarm_sec = 0;
>> +    s->alarm_min = 0;
>> +    s->alarm_hr  = 0;
>> +
>> +    ftrtc011_update_irq(s);
>> +}
>> +
>> +static int ftrtc011_init(SysBusDevice *dev)

In reference to my patchset
https://lists.gnu.org/archive/html/qemu-devel/2013-01/msg02752.html
could you please make this
static void ftrtc011_realizefn(DeviceState *dev, Error **errp) from the start?

>> +{
>> +    ftrtc011_state *s = FROM_SYSBUS(typeof(*s), dev);
> The same.
>> +
>> +    s->qtimer = qemu_new_timer_ns(vm_clock, ftrtc011_timer_tick, s);
>> +

>> +    memory_region_init_io(&s->mmio, &bus_ops, s, "ftrtc011", 0x1000);
>> +    sysbus_init_mmio(dev, &s->mmio);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_ALARM_LEVEL]);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_ALARM_EDGE]);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_SEC]);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_MIN]);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_HOUR]);
>> +    sysbus_init_irq(dev, &s->irq[IRQ_DAY]);

These do not depend on any properties, so please move them to a .instance_init 
function instead:
static void ftrtc011_initfn(Object *obj)

>> +
>> +    return 0;
>> +}
>> +
>> +static const VMStateDescription vmstate_ftrtc011 = {
>> +    .name = "ftrtc011",
>> +    .version_id = 1,
>> +    .minimum_version_id = 1,
>> +    .minimum_version_id_old = 1,
>> +    .fields = (VMStateField[]) {
>> +        VMSTATE_UINT8(sec, ftrtc011_state),
>> +        VMSTATE_UINT8(min, ftrtc011_state),
>> +        VMSTATE_UINT8(hr, ftrtc011_state),
>> +        VMSTATE_UINT32(day, ftrtc011_state),
>> +        VMSTATE_UINT8(alarm_sec, ftrtc011_state),
>> +        VMSTATE_UINT8(alarm_min, ftrtc011_state),
>> +        VMSTATE_UINT8(alarm_hr, ftrtc011_state),
>> +        VMSTATE_UINT32(cr, ftrtc011_state),
>> +        VMSTATE_UINT32(isr, ftrtc011_state),
>> +        VMSTATE_END_OF_LIST()
>> +    }
>> +};
>> +
>> +static void ftrtc011_class_init(ObjectClass *klass, void *data) {
>> +    SysBusDeviceClass *sdc = SYS_BUS_DEVICE_CLASS(klass);
>> +    DeviceClass *k = DEVICE_CLASS(klass);
>> +
>> +    sdc->init  = ftrtc011_init;
>> +    k->vmsd    = &vmstate_ftrtc011;
>> +    k->reset   = ftrtc011_reset;
>> +    k->no_user = 1;
>> +}
>> +
>> +static TypeInfo ftrtc011_info = {
> static *const* TypeInfo ftrtc011_info

...and optionally ftrtc011_type_info for clarity.

>> +    .name           = "ftrtc011",
>> +    .parent         = TYPE_SYS_BUS_DEVICE,
>> +    .instance_size  = sizeof(ftrtc011_state),
>> +    .class_init     = ftrtc011_class_init,
>> +};
>> +
>> +static void ftrtc011_register_types(void) {
>> +    type_register_static(&ftrtc011_info);
>> +}
>> +
>> +type_init(ftrtc011_register_types)

Regards,
Andreas

--
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg


********************* Confidentiality Notice ************************
This electronic message and any attachments may contain
confidential and legally privileged information or
information which is otherwise protected from disclosure.
If you are not the intended recipient,please do not disclose
the contents, either in whole or in part, to anyone,and
immediately delete the message and any attachments from
your computer system and destroy all hard copies.
Thank you for your cooperation.
***********************************************************************