qemu/hw/misc/macio/cuda.c
Mark Cave-Ayland d6c666ad81 cuda: decrease time delay before raising VIA SR interrupt and remove fast path
In order to handle a race condition in the MacOS 9 CUDA driver, a
delay was introduced when raising the VIA SR interrupt inspired by
similar code in MacOnLinux.

During original testing of the MacOS 9 patches it was found that the
30us delay used in MacOnLinux did not work reliably within QEMU, and a
value of 300us was required to function correctly.

Recent experiments have shown two things: firstly when booting Linux,
MacOS 9 and MacOS X the fast path which bypasses the delay is never
triggered once the OS kernel is loaded making it effectively
useless. Rather than leave this code in place where a guest could
potentially enable it by accident and break itself, we might as well
just remove it.

Secondly the previous reliability issues are no longer present, and
this value can be reduced down to 20us with no apparent ill
effects. This has the benefit of considerably improving the
responsiveness of the ADB keyboard and mouse within the guest.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2019-02-17 21:54:02 +11:00

628 lines
19 KiB
C

/*
* QEMU PowerMac CUDA device support
*
* Copyright (c) 2004-2007 Fabrice Bellard
* Copyright (c) 2007 Jocelyn Mayer
*
* 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 "qemu/osdep.h"
#include "hw/hw.h"
#include "hw/ppc/mac.h"
#include "hw/input/adb.h"
#include "hw/misc/mos6522.h"
#include "hw/misc/macio/cuda.h"
#include "qemu/timer.h"
#include "sysemu/sysemu.h"
#include "qemu/cutils.h"
#include "qemu/log.h"
#include "trace.h"
/* Bits in B data register: all active low */
#define TREQ 0x08 /* Transfer request (input) */
#define TACK 0x10 /* Transfer acknowledge (output) */
#define TIP 0x20 /* Transfer in progress (output) */
/* commands (1st byte) */
#define ADB_PACKET 0
#define CUDA_PACKET 1
#define ERROR_PACKET 2
#define TIMER_PACKET 3
#define POWER_PACKET 4
#define MACIIC_PACKET 5
#define PMU_PACKET 6
#define CUDA_TIMER_FREQ (4700000 / 6)
/* CUDA returns time_t's offset from Jan 1, 1904, not 1970 */
#define RTC_OFFSET 2082844800
static void cuda_receive_packet_from_host(CUDAState *s,
const uint8_t *data, int len);
/* MacOS uses timer 1 for calibration on startup, so we use
* the timebase frequency and cuda_get_counter_value() with
* cuda_get_load_time() to steer MacOS to calculate calibrate its timers
* correctly for both TCG and KVM (see commit b981289c49 "PPC: Cuda: Use cuda
* timer to expose tbfreq to guest" for more information) */
static uint64_t cuda_get_counter_value(MOS6522State *s, MOS6522Timer *ti)
{
MOS6522CUDAState *mcs = container_of(s, MOS6522CUDAState, parent_obj);
CUDAState *cs = container_of(mcs, CUDAState, mos6522_cuda);
/* Reverse of the tb calculation algorithm that Mac OS X uses on bootup */
uint64_t tb_diff = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL),
cs->tb_frequency, NANOSECONDS_PER_SECOND) -
ti->load_time;
return (tb_diff * 0xBF401675E5DULL) / (cs->tb_frequency << 24);
}
static uint64_t cuda_get_load_time(MOS6522State *s, MOS6522Timer *ti)
{
MOS6522CUDAState *mcs = container_of(s, MOS6522CUDAState, parent_obj);
CUDAState *cs = container_of(mcs, CUDAState, mos6522_cuda);
uint64_t load_time = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL),
cs->tb_frequency, NANOSECONDS_PER_SECOND);
return load_time;
}
static void cuda_set_sr_int(void *opaque)
{
CUDAState *s = opaque;
MOS6522CUDAState *mcs = &s->mos6522_cuda;
MOS6522State *ms = MOS6522(mcs);
MOS6522DeviceClass *mdc = MOS6522_DEVICE_GET_CLASS(ms);
mdc->set_sr_int(ms);
}
static void cuda_delay_set_sr_int(CUDAState *s)
{
int64_t expire;
trace_cuda_delay_set_sr_int();
expire = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + s->sr_delay_ns;
timer_mod(s->sr_delay_timer, expire);
}
/* NOTE: TIP and TREQ are negated */
static void cuda_update(CUDAState *s)
{
MOS6522CUDAState *mcs = &s->mos6522_cuda;
MOS6522State *ms = MOS6522(mcs);
int packet_received, len;
packet_received = 0;
if (!(ms->b & TIP)) {
/* transfer requested from host */
if (ms->acr & SR_OUT) {
/* data output */
if ((ms->b & (TACK | TIP)) != (s->last_b & (TACK | TIP))) {
if (s->data_out_index < sizeof(s->data_out)) {
trace_cuda_data_send(ms->sr);
s->data_out[s->data_out_index++] = ms->sr;
cuda_delay_set_sr_int(s);
}
}
} else {
if (s->data_in_index < s->data_in_size) {
/* data input */
if ((ms->b & (TACK | TIP)) != (s->last_b & (TACK | TIP))) {
ms->sr = s->data_in[s->data_in_index++];
trace_cuda_data_recv(ms->sr);
/* indicate end of transfer */
if (s->data_in_index >= s->data_in_size) {
ms->b = (ms->b | TREQ);
}
cuda_delay_set_sr_int(s);
}
}
}
} else {
/* no transfer requested: handle sync case */
if ((s->last_b & TIP) && (ms->b & TACK) != (s->last_b & TACK)) {
/* update TREQ state each time TACK change state */
if (ms->b & TACK) {
ms->b = (ms->b | TREQ);
} else {
ms->b = (ms->b & ~TREQ);
}
cuda_delay_set_sr_int(s);
} else {
if (!(s->last_b & TIP)) {
/* handle end of host to cuda transfer */
packet_received = (s->data_out_index > 0);
/* always an IRQ at the end of transfer */
cuda_delay_set_sr_int(s);
}
/* signal if there is data to read */
if (s->data_in_index < s->data_in_size) {
ms->b = (ms->b & ~TREQ);
}
}
}
s->last_acr = ms->acr;
s->last_b = ms->b;
/* NOTE: cuda_receive_packet_from_host() can call cuda_update()
recursively */
if (packet_received) {
len = s->data_out_index;
s->data_out_index = 0;
cuda_receive_packet_from_host(s, s->data_out, len);
}
}
static void cuda_send_packet_to_host(CUDAState *s,
const uint8_t *data, int len)
{
int i;
trace_cuda_packet_send(len);
for (i = 0; i < len; i++) {
trace_cuda_packet_send_data(i, data[i]);
}
memcpy(s->data_in, data, len);
s->data_in_size = len;
s->data_in_index = 0;
cuda_update(s);
cuda_delay_set_sr_int(s);
}
static void cuda_adb_poll(void *opaque)
{
CUDAState *s = opaque;
uint8_t obuf[ADB_MAX_OUT_LEN + 2];
int olen;
olen = adb_poll(&s->adb_bus, obuf + 2, s->adb_poll_mask);
if (olen > 0) {
obuf[0] = ADB_PACKET;
obuf[1] = 0x40; /* polled data */
cuda_send_packet_to_host(s, obuf, olen + 2);
}
timer_mod(s->adb_poll_timer, qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) +
(NANOSECONDS_PER_SECOND / (1000 / s->autopoll_rate_ms)));
}
/* description of commands */
typedef struct CudaCommand {
uint8_t command;
const char *name;
bool (*handler)(CUDAState *s,
const uint8_t *in_args, int in_len,
uint8_t *out_args, int *out_len);
} CudaCommand;
static bool cuda_cmd_autopoll(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
int autopoll;
if (in_len != 1) {
return false;
}
autopoll = (in_data[0] != 0);
if (autopoll != s->autopoll) {
s->autopoll = autopoll;
if (autopoll) {
timer_mod(s->adb_poll_timer,
qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) +
(NANOSECONDS_PER_SECOND / (1000 / s->autopoll_rate_ms)));
} else {
timer_del(s->adb_poll_timer);
}
}
return true;
}
static bool cuda_cmd_set_autorate(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 1) {
return false;
}
/* we don't want a period of 0 ms */
/* FIXME: check what real hardware does */
if (in_data[0] == 0) {
return false;
}
s->autopoll_rate_ms = in_data[0];
if (s->autopoll) {
timer_mod(s->adb_poll_timer,
qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) +
(NANOSECONDS_PER_SECOND / (1000 / s->autopoll_rate_ms)));
}
return true;
}
static bool cuda_cmd_set_device_list(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 2) {
return false;
}
s->adb_poll_mask = (((uint16_t)in_data[0]) << 8) | in_data[1];
return true;
}
static bool cuda_cmd_powerdown(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 0) {
return false;
}
qemu_system_shutdown_request(SHUTDOWN_CAUSE_GUEST_SHUTDOWN);
return true;
}
static bool cuda_cmd_reset_system(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 0) {
return false;
}
qemu_system_reset_request(SHUTDOWN_CAUSE_GUEST_RESET);
return true;
}
static bool cuda_cmd_set_file_server_flag(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 1) {
return false;
}
qemu_log_mask(LOG_UNIMP,
"CUDA: unimplemented command FILE_SERVER_FLAG %d\n",
in_data[0]);
return true;
}
static bool cuda_cmd_set_power_message(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
if (in_len != 1) {
return false;
}
qemu_log_mask(LOG_UNIMP,
"CUDA: unimplemented command SET_POWER_MESSAGE %d\n",
in_data[0]);
return true;
}
static bool cuda_cmd_get_time(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
uint32_t ti;
if (in_len != 0) {
return false;
}
ti = s->tick_offset + (qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL)
/ NANOSECONDS_PER_SECOND);
out_data[0] = ti >> 24;
out_data[1] = ti >> 16;
out_data[2] = ti >> 8;
out_data[3] = ti;
*out_len = 4;
return true;
}
static bool cuda_cmd_set_time(CUDAState *s,
const uint8_t *in_data, int in_len,
uint8_t *out_data, int *out_len)
{
uint32_t ti;
if (in_len != 4) {
return false;
}
ti = (((uint32_t)in_data[0]) << 24) + (((uint32_t)in_data[1]) << 16)
+ (((uint32_t)in_data[2]) << 8) + in_data[3];
s->tick_offset = ti - (qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL)
/ NANOSECONDS_PER_SECOND);
return true;
}
static const CudaCommand handlers[] = {
{ CUDA_AUTOPOLL, "AUTOPOLL", cuda_cmd_autopoll },
{ CUDA_SET_AUTO_RATE, "SET_AUTO_RATE", cuda_cmd_set_autorate },
{ CUDA_SET_DEVICE_LIST, "SET_DEVICE_LIST", cuda_cmd_set_device_list },
{ CUDA_POWERDOWN, "POWERDOWN", cuda_cmd_powerdown },
{ CUDA_RESET_SYSTEM, "RESET_SYSTEM", cuda_cmd_reset_system },
{ CUDA_FILE_SERVER_FLAG, "FILE_SERVER_FLAG",
cuda_cmd_set_file_server_flag },
{ CUDA_SET_POWER_MESSAGES, "SET_POWER_MESSAGES",
cuda_cmd_set_power_message },
{ CUDA_GET_TIME, "GET_TIME", cuda_cmd_get_time },
{ CUDA_SET_TIME, "SET_TIME", cuda_cmd_set_time },
};
static void cuda_receive_packet(CUDAState *s,
const uint8_t *data, int len)
{
uint8_t obuf[16] = { CUDA_PACKET, 0, data[0] };
int i, out_len = 0;
for (i = 0; i < ARRAY_SIZE(handlers); i++) {
const CudaCommand *desc = &handlers[i];
if (desc->command == data[0]) {
trace_cuda_receive_packet_cmd(desc->name);
out_len = 0;
if (desc->handler(s, data + 1, len - 1, obuf + 3, &out_len)) {
cuda_send_packet_to_host(s, obuf, 3 + out_len);
} else {
qemu_log_mask(LOG_GUEST_ERROR,
"CUDA: %s: wrong parameters %d\n",
desc->name, len);
obuf[0] = ERROR_PACKET;
obuf[1] = 0x5; /* bad parameters */
obuf[2] = CUDA_PACKET;
obuf[3] = data[0];
cuda_send_packet_to_host(s, obuf, 4);
}
return;
}
}
qemu_log_mask(LOG_GUEST_ERROR, "CUDA: unknown command 0x%02x\n", data[0]);
obuf[0] = ERROR_PACKET;
obuf[1] = 0x2; /* unknown command */
obuf[2] = CUDA_PACKET;
obuf[3] = data[0];
cuda_send_packet_to_host(s, obuf, 4);
}
static void cuda_receive_packet_from_host(CUDAState *s,
const uint8_t *data, int len)
{
int i;
trace_cuda_packet_receive(len);
for (i = 0; i < len; i++) {
trace_cuda_packet_receive_data(i, data[i]);
}
switch(data[0]) {
case ADB_PACKET:
{
uint8_t obuf[ADB_MAX_OUT_LEN + 3];
int olen;
olen = adb_request(&s->adb_bus, obuf + 2, data + 1, len - 1);
if (olen > 0) {
obuf[0] = ADB_PACKET;
obuf[1] = 0x00;
cuda_send_packet_to_host(s, obuf, olen + 2);
} else {
/* error */
obuf[0] = ADB_PACKET;
obuf[1] = -olen;
obuf[2] = data[1];
olen = 0;
cuda_send_packet_to_host(s, obuf, olen + 3);
}
}
break;
case CUDA_PACKET:
cuda_receive_packet(s, data + 1, len - 1);
break;
}
}
static uint64_t mos6522_cuda_read(void *opaque, hwaddr addr, unsigned size)
{
CUDAState *s = opaque;
MOS6522CUDAState *mcs = &s->mos6522_cuda;
MOS6522State *ms = MOS6522(mcs);
addr = (addr >> 9) & 0xf;
return mos6522_read(ms, addr, size);
}
static void mos6522_cuda_write(void *opaque, hwaddr addr, uint64_t val,
unsigned size)
{
CUDAState *s = opaque;
MOS6522CUDAState *mcs = &s->mos6522_cuda;
MOS6522State *ms = MOS6522(mcs);
addr = (addr >> 9) & 0xf;
mos6522_write(ms, addr, val, size);
}
static const MemoryRegionOps mos6522_cuda_ops = {
.read = mos6522_cuda_read,
.write = mos6522_cuda_write,
.endianness = DEVICE_BIG_ENDIAN,
.valid = {
.min_access_size = 1,
.max_access_size = 1,
},
};
static const VMStateDescription vmstate_cuda = {
.name = "cuda",
.version_id = 5,
.minimum_version_id = 5,
.fields = (VMStateField[]) {
VMSTATE_STRUCT(mos6522_cuda.parent_obj, CUDAState, 0, vmstate_mos6522,
MOS6522State),
VMSTATE_UINT8(last_b, CUDAState),
VMSTATE_UINT8(last_acr, CUDAState),
VMSTATE_INT32(data_in_size, CUDAState),
VMSTATE_INT32(data_in_index, CUDAState),
VMSTATE_INT32(data_out_index, CUDAState),
VMSTATE_UINT8(autopoll, CUDAState),
VMSTATE_UINT8(autopoll_rate_ms, CUDAState),
VMSTATE_UINT16(adb_poll_mask, CUDAState),
VMSTATE_BUFFER(data_in, CUDAState),
VMSTATE_BUFFER(data_out, CUDAState),
VMSTATE_UINT32(tick_offset, CUDAState),
VMSTATE_TIMER_PTR(adb_poll_timer, CUDAState),
VMSTATE_TIMER_PTR(sr_delay_timer, CUDAState),
VMSTATE_END_OF_LIST()
}
};
static void cuda_reset(DeviceState *dev)
{
CUDAState *s = CUDA(dev);
s->data_in_size = 0;
s->data_in_index = 0;
s->data_out_index = 0;
s->autopoll = 0;
}
static void cuda_realize(DeviceState *dev, Error **errp)
{
CUDAState *s = CUDA(dev);
SysBusDevice *sbd;
MOS6522State *ms;
DeviceState *d;
struct tm tm;
/* Pass IRQ from 6522 */
d = DEVICE(&s->mos6522_cuda);
ms = MOS6522(d);
sbd = SYS_BUS_DEVICE(s);
sysbus_pass_irq(sbd, SYS_BUS_DEVICE(ms));
qemu_get_timedate(&tm, 0);
s->tick_offset = (uint32_t)mktimegm(&tm) + RTC_OFFSET;
s->sr_delay_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, cuda_set_sr_int, s);
s->sr_delay_ns = 20 * SCALE_US;
s->adb_poll_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, cuda_adb_poll, s);
s->adb_poll_mask = 0xffff;
s->autopoll_rate_ms = 20;
}
static void cuda_init(Object *obj)
{
CUDAState *s = CUDA(obj);
SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
sysbus_init_child_obj(obj, "mos6522-cuda", &s->mos6522_cuda,
sizeof(s->mos6522_cuda), TYPE_MOS6522_CUDA);
memory_region_init_io(&s->mem, obj, &mos6522_cuda_ops, s, "cuda", 0x2000);
sysbus_init_mmio(sbd, &s->mem);
qbus_create_inplace(&s->adb_bus, sizeof(s->adb_bus), TYPE_ADB_BUS,
DEVICE(obj), "adb.0");
}
static Property cuda_properties[] = {
DEFINE_PROP_UINT64("timebase-frequency", CUDAState, tb_frequency, 0),
DEFINE_PROP_END_OF_LIST()
};
static void cuda_class_init(ObjectClass *oc, void *data)
{
DeviceClass *dc = DEVICE_CLASS(oc);
dc->realize = cuda_realize;
dc->reset = cuda_reset;
dc->vmsd = &vmstate_cuda;
dc->props = cuda_properties;
set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
}
static const TypeInfo cuda_type_info = {
.name = TYPE_CUDA,
.parent = TYPE_SYS_BUS_DEVICE,
.instance_size = sizeof(CUDAState),
.instance_init = cuda_init,
.class_init = cuda_class_init,
};
static void mos6522_cuda_portB_write(MOS6522State *s)
{
MOS6522CUDAState *mcs = container_of(s, MOS6522CUDAState, parent_obj);
CUDAState *cs = container_of(mcs, CUDAState, mos6522_cuda);
cuda_update(cs);
}
static void mos6522_cuda_reset(DeviceState *dev)
{
MOS6522State *ms = MOS6522(dev);
MOS6522DeviceClass *mdc = MOS6522_DEVICE_GET_CLASS(ms);
mdc->parent_reset(dev);
ms->timers[0].frequency = CUDA_TIMER_FREQ;
ms->timers[1].frequency = (SCALE_US * 6000) / 4700;
}
static void mos6522_cuda_class_init(ObjectClass *oc, void *data)
{
DeviceClass *dc = DEVICE_CLASS(oc);
MOS6522DeviceClass *mdc = MOS6522_DEVICE_CLASS(oc);
dc->reset = mos6522_cuda_reset;
mdc->portB_write = mos6522_cuda_portB_write;
mdc->get_timer1_counter_value = cuda_get_counter_value;
mdc->get_timer2_counter_value = cuda_get_counter_value;
mdc->get_timer1_load_time = cuda_get_load_time;
mdc->get_timer2_load_time = cuda_get_load_time;
}
static const TypeInfo mos6522_cuda_type_info = {
.name = TYPE_MOS6522_CUDA,
.parent = TYPE_MOS6522,
.instance_size = sizeof(MOS6522CUDAState),
.class_init = mos6522_cuda_class_init,
};
static void cuda_register_types(void)
{
type_register_static(&mos6522_cuda_type_info);
type_register_static(&cuda_type_info);
}
type_init(cuda_register_types)