hw/sd/aspeed_sdhci: New device
RFC v2 1/1 hw/sd/aspeed_sdhci: New device
The Aspeed SOCs have two SD/MMC controllers. Add a device that encapsulates both of these controllers and models the Aspeed-specific registers and behavior.
Tested by reading from mmcblk0 in Linux:
qemu-system-arm -machine romulus-bmc -nographic -serial mon:stdio \
-drive file=_tmp/flash-romulus,format=raw,if=mtd \
-device sd-card,drive=sd0 -drive file=_tmp/kernel,format=raw,if=sd
---
This patch applies on top of Cedric's set of recent Aspeed changes. Therefore, I'm sending as an RFC rather than a patch.
Changes since v1:
- Move slot realize code into the Aspeed SDHCI realize function
- Fix interrupt handling by creating input irqs and connecting them to the slot irqs.
- Removed card device creation code
hw/arm/aspeed.c | 1 - hw/arm/aspeed_soc.c | 24 ++++++ hw/sd/Makefile.objs | 1 + hw/sd/aspeed_sdhci.c | 190 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/aspeed_soc.h | 3 + include/hw/sd/aspeed_sdhci.h | 35 ++++++++ 6 files changed, 253 insertions(+), 1 deletion(-) create mode 100644 hw/sd/aspeed_sdhci.c create mode 100644 include/hw/sd/aspeed_sdhci.h
hw/arm/aspeed.c +0 -1
@@ -480,7 +480,6 @@ static void aspeed_machine_class_init(ObjectClass *oc, void *data) mc->desc = board->desc; mc->init = aspeed_machine_init; mc->max_cpus =ASPEED_CPUS_NUM ; - mc->no_sdcard =1 ; mc->no_floppy =1 ; mc->no_cdrom =1 ; mc->no_parallel =1 ;
hw/arm/aspeed_soc.c +24 -0
@@ -22,6 +22,7 @@#include "qemu/error-report.h" #include "hw/i2c/aspeed_i2c.h" #include "net/net.h" +#include "sysemu/blockdev.h" #define ASPEED_SOC_IOMEM_SIZE 0x00200000
@@ -62,6 +63,7 @@ static const hwaddr aspeed_soc_ast2500_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_SRAM] = 0x1E720000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,
@@ -100,6 +102,7 @@ static const hwaddr aspeed_soc_ast2600_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_VIDEO] = 0x1E700000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,
@@ -146,6 +149,7 @@ static const int aspeed_soc_ast2400_irqmap[] = { [ASPEED_ETH1] = 2, [ASPEED_ETH2] = 3, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 26, };#define aspeed_soc_ast2500_irqmap aspeed_soc_ast2400_irqmap
@@ -163,6 +167,7 @@ static const int aspeed_soc_ast2600_irqmap[] = { [ASPEED_SDMC] = 0, [ASPEED_SCU] = 12, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 43, [ASPEED_ADC] = 46, [ASPEED_GPIO] = 40, [ASPEED_RTC] = 13,
@@ -350,6 +355,15 @@ static void aspeed_soc_init(Object *obj)sysbus_init_child_obj (obj,"fsi[*]" ,OBJECT (&s->fsi[0 ]),sizeof (s->fsi[0 ]),TYPE_ASPEED_FSI ); } + +sysbus_init_child_obj (obj,"sdhci" ,OBJECT (&s->sdhci),sizeof (s->sdhci), +TYPE_ASPEED_SDHCI ); + +/* Init sd card slot class here so that they're under the correct parent */ +for (i =0 ; i <ASPEED_SDHCI_NUM_SLOTS ; ++i) { +sysbus_init_child_obj (obj,"sdhci_slot[*]" ,OBJECT (&s->sdhci.slots[i]), +sizeof (s->sdhci.slots[i]),TYPE_SYSBUS_SDHCI ); + } }/*
@@ -680,6 +694,16 @@ static void aspeed_soc_realize(DeviceState *dev, Error **errp)sysbus_connect_irq (SYS_BUS_DEVICE (&s->fsi[0 ]),0 ,aspeed_soc_get_irq (s,ASPEED_FSI1 )); } + +/* SD/SDIO - set the reg address so slot memory mapping can be set up */ + s->sdhci.ioaddr = sc->info->memmap[ASPEED_SDHCI ]; +object_property_set_bool (OBJECT (&s->sdhci), true,"realized" , &err); +if (err) { +error_propagate (errp, err); +return ; + } +sysbus_connect_irq (SYS_BUS_DEVICE (&s->sdhci),0 , +aspeed_soc_get_irq (s,ASPEED_SDHCI )); }static Property aspeed_soc_properties[] = {DEFINE_PROP_UINT32 ("num-cpus" , AspeedSoCState, num_cpus,0 ),
hw/sd/Makefile.objs +1 -0
@@ -8,3 +8,4 @@ obj-$(CONFIG_MILKYMIST) += milkymist-memcard.o obj-$(CONFIG_OMAP) += omap_mmc.o obj-$(CONFIG_PXA2XX) += pxa2xx_mmci.o obj-$(CONFIG_RASPI) += bcm2835_sdhost.o +obj-$(CONFIG_ASPEED_SOC) += aspeed_sdhci.o
hw/sd/aspeed_sdhci.c new file +190 -0
190 lines not on this page · raw message
include/hw/arm/aspeed_soc.h +3 -0
@@ -29,6 +29,7 @@#include "hw/misc/aspeed_pwm.h" #include "hw/misc/aspeed_lpc.h" #include "hw/misc/aspeed_fsi.h" +#include "hw/sd/aspeed_sdhci.h" #define ASPEED_SPIS_NUM 2#define ASPEED_WDTS_NUM 4
@@ -62,6 +63,7 @@ typedef struct AspeedSoCState {AspeedPWMState pwm;AspeedLPCState lpc;AspeedFsiState fsi[2 ]; +AspeedSDHCIState sdhci; } AspeedSoCState;#define TYPE_ASPEED_SOC "aspeed-soc"
@@ -107,6 +109,7 @@ enum {ASPEED_ADC ,ASPEED_VIDEO ,ASPEED_SRAM , +ASPEED_SDHCI ,ASPEED_GPIO ,ASPEED_RTC , ASPEED_TIMER1,
include/hw/sd/aspeed_sdhci.h new file +35 -0
@@ -0,0 +1,35 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#ifndef ASPEED_SDHCI_H +#define ASPEED_SDHCI_H + +#include "hw/sd/sdhci.h" + +#define TYPE_ASPEED_SDHCI "aspeed.sdhci" +#define ASPEED_SDHCI (obj) OBJECT_CHECK(AspeedSDHCIState, (obj), \ + TYPE_ASPEED_SDHCI) + +#define ASPEED_SDHCI_CAPABILITIES 0x01E80080 +#define ASPEED_SDHCI_NUM_SLOTS 2 +#define ASPEED_SDHCI_NUM_REGS (ASPEED_SDHCI_REG_SIZE / sizeof(uint32_t)) +#define ASPEED_SDHCI_REG_SIZE 0x100 + +typedef struct AspeedSDHCIState { +SysBusDevice parent; + +SDHCIState slots[ASPEED_SDHCI_NUM_SLOTS ]; + +hwaddr ioaddr; +MemoryRegion iomem; +qemu_irq irq; + +uint32_t regs[ASPEED_SDHCI_NUM_REGS ]; +}AspeedSDHCIState ; + +#endif /* ASPEED_SDHCI_H */
Signature · 1 line
1.8.3.1
Hello Eddie,
Quoted 9 lines · Eddie James
The Aspeed SOCs have two SD/MMC controllers. Add a device that encapsulates both of these controllers and models the Aspeed-specific registers and behavior.
Tested by reading from mmcblk0 in Linux:
qemu-system-arm -machine romulus-bmc -nographic -serial mon:stdio \
-drive file=_tmp/flash-romulus,format=raw,if=mtd \
-device sd-card,drive=sd0 -drive file=_tmp/kernel,format=raw,if=sd
Quoted 4 lines
Signed-off-by Eddie James <[email protected]>---
This patch applies on top of Cedric's set of recent Aspeed changes. Therefore, I'm sending as an RFC rather than a patch.
yes. we can worked that out when the patch is reviewed. You can base on mainline when ready. My tree serves as an overall test base.
Quoted 5 lines
Changes since v1:
- Move slot realize code into the Aspeed SDHCI realize function
- Fix interrupt handling by creating input irqs and connecting them to the slot irqs.
- Removed card device creation code
I think all the code is here but it needs some more reshuffling :)
The raspi machine is a good source for modelling pratices.
Quoted 27 lines
hw/arm/aspeed.c | 1 - hw/arm/aspeed_soc.c | 24 ++++++ hw/sd/Makefile.objs | 1 + hw/sd/aspeed_sdhci.c | 190 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/aspeed_soc.h | 3 + include/hw/sd/aspeed_sdhci.h | 35 ++++++++ 6 files changed, 253 insertions(+), 1 deletion(-) create mode 100644 hw/sd/aspeed_sdhci.c create mode 100644 include/hw/sd/aspeed_sdhci.hhw/arm/aspeed.c +0
-1@@ -480,7 +480,6 @@ static void aspeed_machine_class_init(ObjectClass *oc, void *data) mc->desc = board->desc; mc->init = aspeed_machine_init; mc->max_cpus =ASPEED_CPUS_NUM ; - mc->no_sdcard =1 ; mc->no_floppy =1 ; mc->no_cdrom =1 ; mc->no_parallel =1 ;hw/arm/aspeed_soc.c +0
-0
hw/arm/aspeed_soc.c line 25
Quoted 4 lines
hw/arm/aspeed_soc.c +1
-0@@ -22,6 +22,7 @@#include "qemu/error-report.h" #include "hw/i2c/aspeed_i2c.h" #include "net/net.h" +#include "sysemu/blockdev.h"
I would expect block devices to be handled at the machine level in aspeed.c, like the flash devices are. Something like :
/* Create and plug in the SD cards */
for (i = 0; i < ASPEED_SDHCI_NUM_SLOTS; i++) {
BusState *bus;
DriveInfo *di = drive_get_next(IF_SD);
BlockBackend *blk = di ? blk_by_legacy_dinfo(di) : NULL;
DeviceState *carddev;
bus = qdev_get_child_bus(DEVICE(&s->soc), "sd-bus");
if (!bus) {
error_report("No SD bus found for SD card %d", i);
exit(1);
}
carddev = qdev_create(bus, TYPE_SD_CARD);
qdev_prop_set_drive(carddev, "drive", blk, &error_fatal);
object_property_set_bool(OBJECT(carddev), true, "realized",
&error_fatal);
}Quoted 36 lines
hw/arm/aspeed_soc.c +4
-0#define ASPEED_SOC_IOMEM_SIZE 0x00200000@@ -62,6 +63,7 @@ static const hwaddr aspeed_soc_ast2500_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_SRAM] = 0x1E720000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -100,6 +102,7 @@ static const hwaddr aspeed_soc_ast2600_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_VIDEO] = 0x1E700000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -146,6 +149,7 @@ static const int aspeed_soc_ast2400_irqmap[] = { [ASPEED_ETH1] = 2, [ASPEED_ETH2] = 3, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 26, };#define aspeed_soc_ast2500_irqmap aspeed_soc_ast2400_irqmap@@ -163,6 +167,7 @@ static const int aspeed_soc_ast2600_irqmap[] = { [ASPEED_SDMC] = 0, [ASPEED_SCU] = 12, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 43, [ASPEED_ADC] = 46, [ASPEED_GPIO] = 40, [ASPEED_RTC] = 13,
hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 360
Quoted 6 lines
hw/arm/aspeed_soc.c +3
-0@@ -350,6 +355,15 @@ static void aspeed_soc_init(Object *obj)sysbus_init_child_obj (obj,"fsi[*]" ,OBJECT (&s->fsi[0 ]),sizeof (s->fsi[0 ]),TYPE_ASPEED_FSI ); } + +sysbus_init_child_obj (obj,"sdhci" ,OBJECT (&s->sdhci),sizeof (s->sdhci), +TYPE_ASPEED_SDHCI );
This is the Aspeed SD host interface. May be we should call it sdhost ?
I suppose this is our "sd-bus" device ?
hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 363
Quoted 2 lines
hw/arm/aspeed_soc.c +2
-0+/* Init sd card slot class here so that they're under the correct parent */ +for (i =0 ; i <ASPEED_SDHCI_NUM_SLOTS ; ++i) {
and these are the slots, I would put them at the SoC level.
hw/arm/aspeed_soc.c static void aspeed_soc_realize(DeviceState *dev, Error **errp) · line 699
Quoted 13 lines
hw/arm/aspeed_soc.c +6
-0+sysbus_init_child_obj (obj,"sdhci_slot[*]" ,OBJECT (&s->sdhci.slots[i]), +sizeof (s->sdhci.slots[i]),TYPE_SYSBUS_SDHCI ); + } }/* @@ -680,6 +694,16 @@ static void aspeed_soc_realize(DeviceState *dev, Error **errp)sysbus_connect_irq (SYS_BUS_DEVICE (&s->fsi[0 ]),0 ,aspeed_soc_get_irq (s,ASPEED_FSI1 )); } + +/* SD/SDIO - set the reg address so slot memory mapping can be set up */ + s->sdhci.ioaddr = sc->info->memmap[ASPEED_SDHCI ];
That's weird. We do all mappings in the SoC.
I think you should be realizing the slots here also. See the other SoCs, XlnxZynqMPState for instance.
Quoted 63 lines
hw/arm/aspeed_soc.c +7
-0+object_property_set_bool (OBJECT (&s->sdhci), true,"realized" , &err); +if (err) { +error_propagate (errp, err); +return ; + } +sysbus_connect_irq (SYS_BUS_DEVICE (&s->sdhci),0 , +aspeed_soc_get_irq (s,ASPEED_SDHCI )); }static Property aspeed_soc_properties[] = {
DEFINE_PROP_UINT32("num-cpus", AspeedSoCState, num_cpus, 0),hw/sd/Makefile.objs +1
-0@@ -8,3 +8,4 @@ obj-$(CONFIG_MILKYMIST) += milkymist-memcard.o obj-$(CONFIG_OMAP) += omap_mmc.o obj-$(CONFIG_PXA2XX) += pxa2xx_mmci.o obj-$(CONFIG_RASPI) += bcm2835_sdhost.o +obj-$(CONFIG_ASPEED_SOC) += aspeed_sdhci.ohw/sd/aspeed_sdhci.c new file +37
-0@@ -0,0 +1,190 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qemu/error-report.h" +#include "hw/sd/aspeed_sdhci.h" +#include "qapi/error.h" + +#define ASPEED_SDHCI_INFO 0x00 +#define ASPEED_SDHCI_INFO_RESET 0x00030000 +#define ASPEED_SDHCI_DEBOUNCE 0x04 +#define ASPEED_SDHCI_DEBOUNCE_RESET 0x00000005 +#define ASPEED_SDHCI_BUS 0x08 +#define ASPEED_SDHCI_SDIO_140 0x10 +#define ASPEED_SDHCI_SDIO_148 0x18 +#define ASPEED_SDHCI_SDIO_240 0x20 +#define ASPEED_SDHCI_SDIO_248 0x28 +#define ASPEED_SDHCI_WP_POL 0xec +#define ASPEED_SDHCI_CARD_DET 0xf0 +#define ASPEED_SDHCI_IRQ_STAT 0xfc + +#define TO_REG (addr) ((addr) / sizeof(uint32_t)) + +static uint64_t aspeed_sdhci_read (void *opaque,hwaddr addr,unsigned int size) +{ +uint32_t val =0 ; +AspeedSDHCIState *sdhci = opaque; + +switch (addr) { +case ASPEED_SDHCI_SDIO_140 : + val = (uint32_t )sdhci->slots[0 ].capareg;
hw/sd/aspeed_sdhci.c line 47
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+break ; +case ASPEED_SDHCI_SDIO_148 : + val = (uint32_t )sdhci->slots[0 ].maxcurr; +break ; +case ASPEED_SDHCI_SDIO_240 : + val = (uint32_t )sdhci->slots[1 ].capareg; +break ; +case ASPEED_SDHCI_SDIO_248 : + val = (uint32_t )sdhci->slots[1 ].maxcurr; +break ;
We could mirror the 16bytes segment for [ SDHC_CAPAB ... SDHC_MAXCURR + 4 ] of each slot under the MMIO region of the Aspeed SD controller at offset:
(slot + 1) * 0x10. If that worked, we wouldn't need these redirections.
I think you need alias regions.
hw/sd/aspeed_sdhci.c line 51
Quoted 4 lines
hw/sd/aspeed_sdhci.c new file +4
-0+default : +if (addr <ASPEED_SDHCI_REG_SIZE ) { + val = sdhci->regs[TO_REG (addr)]; + }
we could return some errors for not implemented registers.
Quoted 13 lines
hw/sd/aspeed_sdhci.c new file +13
-0+ } + +return (uint64_t )val; +} + +static void aspeed_sdhci_write (void *opaque,hwaddr addr,uint64_t val, +unsigned int size) +{ +AspeedSDHCIState *sdhci = opaque; + +switch (addr) { +case ASPEED_SDHCI_SDIO_140 : + sdhci->slots[0 ].capareg = (uint64_t )(uint32_t )val;
hw/sd/aspeed_sdhci.c line 74
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+break ; +case ASPEED_SDHCI_SDIO_148 : + sdhci->slots[0 ].maxcurr = (uint64_t )(uint32_t )val; +break ; +case ASPEED_SDHCI_SDIO_240 : + sdhci->slots[1 ].capareg = (uint64_t )(uint32_t )val; +break ; +case ASPEED_SDHCI_SDIO_248 : + sdhci->slots[1 ].maxcurr = (uint64_t )(uint32_t )val; +break ;
I think these regs are readonly.
Quoted 18 lines
hw/sd/aspeed_sdhci.c new file +18
-0+default : +if (addr <ASPEED_SDHCI_REG_SIZE ) { + sdhci->regs[TO_REG (addr)] = (uint32_t )val; + } + } +} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness =DEVICE_NATIVE_ENDIAN , + .valid.min_access_size =4 , + .valid.max_access_size =4 , +}; + +static void aspeed_sdhci_set_irq (void *opaque,int n,int level) +{ +AspeedSDHCIState *sdhci = opaque;
hw/sd/aspeed_sdhci.c line 102
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ +if (level) { + sdhci->regs[TO_REG (ASPEED_SDHCI_IRQ_STAT )] |=BIT (n); + +qemu_irq_raise (sdhci->irq); + }else { + sdhci->regs[TO_REG (ASPEED_SDHCI_IRQ_STAT )] &= ~BIT (n); + +qemu_irq_lower (sdhci->irq); + }
Doesn't that work the other way around ?
The SDHCI objects trigger their IRQ which call the irq_notify() handler in which we need to deduce the slot number to update ASPEED_SDHCI_IRQ_STAT and raise/lower the Aspeed SD host IRQ. I think that's how it should work.
Quoted 27 lines
hw/sd/aspeed_sdhci.c new file +27
-0+} + +static void aspeed_sdhci_realize (DeviceState *dev,Error **errp) +{ +Error *err =NULL ; +SysBusDevice *sbd =SYS_BUS_DEVICE (dev); +AspeedSDHCIState *sdhci =ASPEED_SDHCI (dev); + +/* Create input irqs for the slots */ +qdev_init_gpio_in_named_with_opaque (DEVICE (sbd), aspeed_sdhci_set_irq, + sdhci,NULL ,ASPEED_SDHCI_NUM_SLOTS ); + +for (int i =0 ; i <ASPEED_SDHCI_NUM_SLOTS ; ++i) { +hwaddr slot_addr = sdhci->ioaddr + (0x100 * (i +1 )); +Object *sdhci_slot =OBJECT (&sdhci->slots[i]); +SysBusDevice *sbd_slot =SYS_BUS_DEVICE (&sdhci->slots[i]); + +object_property_set_int (sdhci_slot,2 ,"sd-spec-version" , &err); +if (err) { +error_propagate (errp, err); +return ; + } + +object_property_set_uint (sdhci_slot,ASPEED_SDHCI_CAPABILITIES , +"capareg" , &err); +if (err) { +error_propagate (errp, err);
hw/sd/aspeed_sdhci.c line 139
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+return ; + } + +object_property_set_bool (sdhci_slot, true,"realized" , &err); +if (err) { +error_propagate (errp, err); +return ; + } + +sysbus_mmio_map (sbd_slot,0 , slot_addr);
We should do the mapping at the SoC level and I would also put the slots at the SoC level.
hw/sd/aspeed_sdhci.c line 148
Quoted 8 lines
hw/sd/aspeed_sdhci.c new file +8
-0+sysbus_connect_irq (sbd_slot,0 ,qdev_get_gpio_in (DEVICE (sbd), i)); + } + +sysbus_init_irq (sbd, &sdhci->irq); +memory_region_init_io (&sdhci->iomem,OBJECT (sdhci), &aspeed_sdhci_ops, + sdhci,TYPE_ASPEED_SDHCI ,ASPEED_SDHCI_REG_SIZE ); +sysbus_init_mmio (sbd, &sdhci->iomem); +sysbus_mmio_map (sbd,0 , sdhci->ioaddr);
Here also.
Quoted 89 lines
hw/sd/aspeed_sdhci.c new file +42
-0+} + +static void aspeed_sdhci_reset (DeviceState *dev) +{ +AspeedSDHCIState *sdhci =ASPEED_SDHCI (dev); + +memset (sdhci->regs,0 ,ASPEED_SDHCI_REG_SIZE ); + sdhci->regs[TO_REG (ASPEED_SDHCI_INFO )] =ASPEED_SDHCI_INFO_RESET ; + sdhci->regs[TO_REG (ASPEED_SDHCI_DEBOUNCE )] =ASPEED_SDHCI_DEBOUNCE_RESET ; +} + +static const VMStateDescription vmstate_aspeed_sdhci = { + .name =TYPE_ASPEED_SDHCI , + .version_id =1 , + .fields = (VMStateField []) { +VMSTATE_UINT32_ARRAY (regs, AspeedSDHCIState,ASPEED_SDHCI_NUM_REGS ), +VMSTATE_END_OF_LIST (), + }, +}; + +static void aspeed_sdhci_class_init (ObjectClass *classp,void *data) +{ +DeviceClass *dc =DEVICE_CLASS (classp); + + dc->realize = aspeed_sdhci_realize; + dc->reset = aspeed_sdhci_reset; + dc->vmsd = &vmstate_aspeed_sdhci; +} + +static TypeInfo aspeed_sdhci_info = { + .name =TYPE_ASPEED_SDHCI , + .parent =TYPE_SYS_BUS_DEVICE , + .instance_size =sizeof (AspeedSDHCIState), + .class_init = aspeed_sdhci_class_init, +}; + +static void aspeed_sdhci_register_types (void ) +{ +type_register_static (&aspeed_sdhci_info); +} + +type_init (aspeed_sdhci_register_types)include/hw/arm/aspeed_soc.h +3
-0@@ -29,6 +29,7 @@#include "hw/misc/aspeed_pwm.h" #include "hw/misc/aspeed_lpc.h" #include "hw/misc/aspeed_fsi.h" +#include "hw/sd/aspeed_sdhci.h" #define ASPEED_SPIS_NUM 2#define ASPEED_WDTS_NUM 4@@ -62,6 +63,7 @@ typedef struct AspeedSoCState {AspeedPWMState pwm;AspeedLPCState lpc;AspeedFsiState fsi[2 ]; +AspeedSDHCIState sdhci; } AspeedSoCState;#define TYPE_ASPEED_SOC "aspeed-soc"@@ -107,6 +109,7 @@ enum {ASPEED_ADC ,ASPEED_VIDEO ,ASPEED_SRAM , +ASPEED_SDHCI ,ASPEED_GPIO ,ASPEED_RTC , ASPEED_TIMER1,include/hw/sd/aspeed_sdhci.h new file +13
-0@@ -0,0 +1,35 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#ifndef ASPEED_SDHCI_H +#define ASPEED_SDHCI_H + +#include "hw/sd/sdhci.h" +
include/hw/sd/aspeed_sdhci.h line 23
Quoted 10 lines
include/hw/sd/aspeed_sdhci.h new file +10
-0+#define TYPE_ASPEED_SDHCI "aspeed.sdhci" +#define ASPEED_SDHCI (obj) OBJECT_CHECK(AspeedSDHCIState, (obj), \ + TYPE_ASPEED_SDHCI) + +#define ASPEED_SDHCI_CAPABILITIES 0x01E80080 +#define ASPEED_SDHCI_NUM_SLOTS 2 +#define ASPEED_SDHCI_NUM_REGS (ASPEED_SDHCI_REG_SIZE / sizeof(uint32_t)) +#define ASPEED_SDHCI_REG_SIZE 0x100 + +typedef struct AspeedSDHCIState {
AspeedSDHost may be ? It's the SoC SD controller.
include/hw/sd/aspeed_sdhci.h line 26
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +3
-0+SysBusDevice parent; + +SDHCIState slots[ASPEED_SDHCI_NUM_SLOTS ];
I think the SoC should own the SD slots.
include/hw/sd/aspeed_sdhci.h line 28
Quoted 2 lines
include/hw/sd/aspeed_sdhci.h new file +2
-0+ +hwaddr ioaddr;
not needed.
include/hw/sd/aspeed_sdhci.h line 35
Quoted 8 lines
include/hw/sd/aspeed_sdhci.h new file +7
-0+MemoryRegion iomem; +qemu_irq irq; + + uint32_t regs[ASPEED_SDHCI_NUM_REGS]; +} AspeedSDHCIState; + +#endif /* ASPEED_SDHCI_H */
Thanks,
C.
Quoted 53 lines · Cédric Le Goater
Hello Eddie,
Quoted 9 lines · Eddie James
The Aspeed SOCs have two SD/MMC controllers. Add a device that encapsulates both of these controllers and models the Aspeed-specific registers and behavior.
Tested by reading from mmcblk0 in Linux:
qemu-system-arm -machine romulus-bmc -nographic -serial mon:stdio \
-drive file=_tmp/flash-romulus,format=raw,if=mtd \
-device sd-card,drive=sd0 -drive file=_tmp/kernel,format=raw,if=sdQuoted 4 lines
Signed-off-by Eddie James <[email protected]>---
This patch applies on top of Cedric's set of recent Aspeed changes. Therefore, I'm sending as an RFC rather than a patch.yes. we can worked that out when the patch is reviewed. You can base on mainline when ready. My tree serves as an overall test base.
Quoted 5 lines
Changes since v1:
- Move slot realize code into the Aspeed SDHCI realize function
- Fix interrupt handling by creating input irqs and connecting them to the slot irqs.
- Removed card device creation codeI think all the code is here but it needs some more reshuffling :)
The raspi machine is a good source for modelling pratices.
Quoted 25 lines
hw/arm/aspeed.c | 1 - hw/arm/aspeed_soc.c | 24 ++++++ hw/sd/Makefile.objs | 1 + hw/sd/aspeed_sdhci.c | 190 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/aspeed_soc.h | 3 + include/hw/sd/aspeed_sdhci.h | 35 ++++++++ 6 files changed, 253 insertions(+), 1 deletion(-)create mode 100644
hw/sd/aspeed_sdhci.c
create mode 100644include/hw/sd/aspeed_sdhci.hhw/arm/aspeed.c +0
-1@@ -480,7 +480,6 @@ static void aspeed_machine_class_init(ObjectClass *oc, void *data) mc->desc = board->desc; mc->init = aspeed_machine_init; mc->max_cpus =ASPEED_CPUS_NUM ; - mc->no_sdcard =1 ; mc->no_floppy =1 ; mc->no_cdrom =1 ; mc->no_parallel =1 ;hw/arm/aspeed_soc.c +0
-1--- a/hw/arm/aspeed_soc.c
Quoted 8 lines
hw/arm/aspeed_soc.c line 25
Quoted 6 lines
+++
b/hw/arm/aspeed_soc.chw/arm/aspeed_soc.c +1
-0@@ -22,6 +22,7 @@#include "qemu/error-report.h" #include "hw/i2c/aspeed_i2c.h" #include "net/net.h" +#include "sysemu/blockdev.h" I would expect block devices to be handled at the machine level in
aspeed.c, like the flash devices are. Something like :
OK, I did have that in v1 but Peter mentioned it was typically done at the command line? I guess it can go in aspeed.c too.
Quoted 188 lines
/* Create and plug in the SD cards */ for (i = 0; i < ASPEED_SDHCI_NUM_SLOTS; i++) { BusState *bus; DriveInfo *di = drive_get_next(IF_SD); BlockBackend *blk = di ? blk_by_legacy_dinfo(di) : NULL; DeviceState *carddev; bus = qdev_get_child_bus(DEVICE(&s->soc), "sd-bus"); if (!bus) { error_report("No SD bus found for SD card %d", i); exit(1); } carddev = qdev_create(bus, TYPE_SD_CARD); qdev_prop_set_drive(carddev, "drive", blk, &error_fatal); object_property_set_bool(OBJECT(carddev), true, "realized", &error_fatal); }Quoted 36 lines
hw/arm/aspeed_soc.c +4
-0#define ASPEED_SOC_IOMEM_SIZE 0x00200000@@ -62,6 +63,7 @@ static const hwaddr aspeed_soc_ast2500_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_SRAM] = 0x1E720000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -100,6 +102,7 @@ static const hwaddr aspeed_soc_ast2600_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_VIDEO] = 0x1E700000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -146,6 +149,7 @@ static const int aspeed_soc_ast2400_irqmap[] = { [ASPEED_ETH1] = 2, [ASPEED_ETH2] = 3, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 26, };#define aspeed_soc_ast2500_irqmap aspeed_soc_ast2400_irqmap@@ -163,6 +167,7 @@ static const int aspeed_soc_ast2600_irqmap[] = { [ASPEED_SDMC] = 0, [ASPEED_SCU] = 12, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 43, [ASPEED_ADC] = 46, [ASPEED_GPIO] = 40, [ASPEED_RTC] = 13,hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 360
Quoted 6 lines
hw/arm/aspeed_soc.c +3
-0@@ -350,6 +355,15 @@ static void aspeed_soc_init(Object *obj)sysbus_init_child_obj (obj,"fsi[*]" ,OBJECT (&s->fsi[0 ]),sizeof (s->fsi[0 ]),TYPE_ASPEED_FSI ); } + +sysbus_init_child_obj (obj,"sdhci" ,OBJECT (&s->sdhci),sizeof (s->sdhci), +TYPE_ASPEED_SDHCI );This is the Aspeed SD host interface. May be we should call it sdhost ?
I suppose this is our "sd-bus" device ?
hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 363
Quoted 2 lines
hw/arm/aspeed_soc.c +2
-0+/* Init sd card slot class here so that they're under the correct parent */ +for (i =0 ; i <ASPEED_SDHCI_NUM_SLOTS ; ++i) {and these are the slots, I would put them at the SoC level.
hw/arm/aspeed_soc.c static void aspeed_soc_realize(DeviceState *dev, Error **errp) · line 699
Quoted 13 lines
hw/arm/aspeed_soc.c +6
-0+sysbus_init_child_obj (obj,"sdhci_slot[*]" ,OBJECT (&s->sdhci.slots[i]), +sizeof (s->sdhci.slots[i]),TYPE_SYSBUS_SDHCI ); + } }/* @@ -680,6 +694,16 @@ static void aspeed_soc_realize(DeviceState *dev, Error **errp)sysbus_connect_irq (SYS_BUS_DEVICE (&s->fsi[0 ]),0 ,aspeed_soc_get_irq (s,ASPEED_FSI1 )); } + +/* SD/SDIO - set the reg address so slot memory mapping can be set up */ + s->sdhci.ioaddr = sc->info->memmap[ASPEED_SDHCI ];That's weird. We do all mappings in the SoC.
I think you should be realizing the slots here also. See the other SoCs, XlnxZynqMPState for instance.
Quoted 63 lines
hw/arm/aspeed_soc.c +7
-0+object_property_set_bool (OBJECT (&s->sdhci), true,"realized" , &err); +if (err) { +error_propagate (errp, err); +return ; + } +sysbus_connect_irq (SYS_BUS_DEVICE (&s->sdhci),0 , +aspeed_soc_get_irq (s,ASPEED_SDHCI )); }static Property aspeed_soc_properties[] = {
DEFINE_PROP_UINT32("num-cpus", AspeedSoCState, num_cpus, 0),hw/sd/Makefile.objs +1
-0@@ -8,3 +8,4 @@ obj-$(CONFIG_MILKYMIST) += milkymist-memcard.o obj-$(CONFIG_OMAP) += omap_mmc.o obj-$(CONFIG_PXA2XX) += pxa2xx_mmci.o obj-$(CONFIG_RASPI) += bcm2835_sdhost.o +obj-$(CONFIG_ASPEED_SOC) += aspeed_sdhci.ohw/sd/aspeed_sdhci.c new file +37
-0@@ -0,0 +1,190 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qemu/error-report.h" +#include "hw/sd/aspeed_sdhci.h" +#include "qapi/error.h" + +#define ASPEED_SDHCI_INFO 0x00 +#define ASPEED_SDHCI_INFO_RESET 0x00030000 +#define ASPEED_SDHCI_DEBOUNCE 0x04 +#define ASPEED_SDHCI_DEBOUNCE_RESET 0x00000005 +#define ASPEED_SDHCI_BUS 0x08 +#define ASPEED_SDHCI_SDIO_140 0x10 +#define ASPEED_SDHCI_SDIO_148 0x18 +#define ASPEED_SDHCI_SDIO_240 0x20 +#define ASPEED_SDHCI_SDIO_248 0x28 +#define ASPEED_SDHCI_WP_POL 0xec +#define ASPEED_SDHCI_CARD_DET 0xf0 +#define ASPEED_SDHCI_IRQ_STAT 0xfc + +#define TO_REG (addr) ((addr) / sizeof(uint32_t)) + +static uint64_t aspeed_sdhci_read (void *opaque,hwaddr addr,unsigned int size) +{ +uint32_t val =0 ; +AspeedSDHCIState *sdhci = opaque; + +switch (addr) { +case ASPEED_SDHCI_SDIO_140 : + val = (uint32_t )sdhci->slots[0 ].capareg;hw/sd/aspeed_sdhci.c line 47
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+break ; +case ASPEED_SDHCI_SDIO_148 : + val = (uint32_t )sdhci->slots[0 ].maxcurr; +break ; +case ASPEED_SDHCI_SDIO_240 : + val = (uint32_t )sdhci->slots[1 ].capareg; +break ; +case ASPEED_SDHCI_SDIO_248 : + val = (uint32_t )sdhci->slots[1 ].maxcurr; +break ;We could mirror the 16bytes segment for [
SDHC_CAPAB...SDHC_MAXCURR+ 4 ] of each slot under the MMIO region of the Aspeed SD controller at offset:
(slot + 1) *0x10. If that worked, we wouldn't need these redirections.I think you need alias regions.
hw/sd/aspeed_sdhci.c line 51
Quoted 4 lines
hw/sd/aspeed_sdhci.c new file +4
-0+default : +if (addr <ASPEED_SDHCI_REG_SIZE ) { + val = sdhci->regs[TO_REG (addr)]; + }we could return some errors for not implemented registers.
Quoted 16 lines
hw/sd/aspeed_sdhci.c new file +16
-0+ } + +return (uint64_t )val; +} + +static void aspeed_sdhci_write (void *opaque,hwaddr addr,uint64_t val, +unsigned int size) +{ +AspeedSDHCIState *sdhci = opaque; + +switch (addr) { +case ASPEED_SDHCI_SDIO_140 : + sdhci->slots[0 ].capareg = (uint64_t )(uint32_t )val; +break ; +case ASPEED_SDHCI_SDIO_148 : + sdhci->slots[0 ].maxcurr = (uint64_t )(uint32_t )val;
Quoted 8 lines
hw/sd/aspeed_sdhci.c line 74
Quoted 7 lines
hw/sd/aspeed_sdhci.c new file +7
-0+break ; +case ASPEED_SDHCI_SDIO_240 : + sdhci->slots[1 ].capareg = (uint64_t )(uint32_t )val; +break ; +case ASPEED_SDHCI_SDIO_248 : + sdhci->slots[1 ].maxcurr = (uint64_t )(uint32_t )val; +break ;I think these regs are readonly.
Well the actual regs at slot + 0x40/0x48 are indeed, but not the Aspeed-specific ones that mirror there. I think the idea is that Aspeed-specific code can set it's capabilities differently if desired. This may prevent the use of alias regions here.
Quoted 31 lines
Quoted 19 lines
hw/sd/aspeed_sdhci.c new file +18
-0+default : +if (addr <ASPEED_SDHCI_REG_SIZE ) { + sdhci->regs[TO_REG (addr)] = (uint32_t )val; + } + } +} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness =DEVICE_NATIVE_ENDIAN , + .valid.min_access_size =4 , + .valid.max_access_size =4 , +}; + +static void aspeed_sdhci_set_irq (void *opaque,int n,int level) +{ +AspeedSDHCIState *sdhci = opaque;hw/sd/aspeed_sdhci.c line 102
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ +if (level) { + sdhci->regs[TO_REG (ASPEED_SDHCI_IRQ_STAT )] |=BIT (n); + +qemu_irq_raise (sdhci->irq); + }else { + sdhci->regs[TO_REG (ASPEED_SDHCI_IRQ_STAT )] &= ~BIT (n); + +qemu_irq_lower (sdhci->irq); + }Doesn't that work the other way around ?
Quoted 3 lines
The SDHCI objects trigger their IRQ which call the
irq_notify()handler in which we need to deduce the slot number to updateASPEED_SDHCI_IRQ_STATand raise/lower the Aspeed SD host IRQ. I think that's how it should work.
That's exactly what's happening here. Maybe my variable/function naming is confusing?
Quoted 33 lines
Quoted 33 lines
hw/sd/aspeed_sdhci.c new file +31
-0+} + +static void aspeed_sdhci_realize (DeviceState *dev,Error **errp) +{ +Error *err =NULL ; +SysBusDevice *sbd =SYS_BUS_DEVICE (dev); +AspeedSDHCIState *sdhci =ASPEED_SDHCI (dev); + +/* Create input irqs for the slots */ +qdev_init_gpio_in_named_with_opaque (DEVICE (sbd), aspeed_sdhci_set_irq, + sdhci,NULL ,ASPEED_SDHCI_NUM_SLOTS ); + +for (int i =0 ; i <ASPEED_SDHCI_NUM_SLOTS ; ++i) { +hwaddr slot_addr = sdhci->ioaddr + (0x100 * (i +1 )); +Object *sdhci_slot =OBJECT (&sdhci->slots[i]); +SysBusDevice *sbd_slot =SYS_BUS_DEVICE (&sdhci->slots[i]); + +object_property_set_int (sdhci_slot,2 ,"sd-spec-version" , &err); +if (err) { +error_propagate (errp, err); +return ; + } + +object_property_set_uint (sdhci_slot,ASPEED_SDHCI_CAPABILITIES , +"capareg" , &err); +if (err) { +error_propagate (errp, err); +return ; + } + +object_property_set_bool (sdhci_slot, true,"realized" , &err);
Quoted 8 lines
hw/sd/aspeed_sdhci.c line 139
Quoted 6 lines
hw/sd/aspeed_sdhci.c new file +6
-0+if (err) { +error_propagate (errp, err); +return ; + } + +sysbus_mmio_map (sbd_slot,0 , slot_addr);We should do the mapping at the SoC level and I would also put the slots at the SoC level.
OK. I did that in v1 but Peter made some comments about initializing things in the Aspeed SD code so I moved it all down here...
Quoted 112 lines
hw/sd/aspeed_sdhci.c line 148
Quoted 9 lines
hw/sd/aspeed_sdhci.c new file +8
-0+sysbus_connect_irq (sbd_slot,0 ,qdev_get_gpio_in (DEVICE (sbd), i)); + } + +sysbus_init_irq (sbd, &sdhci->irq); +memory_region_init_io (&sdhci->iomem,OBJECT (sdhci), &aspeed_sdhci_ops, + sdhci,TYPE_ASPEED_SDHCI ,ASPEED_SDHCI_REG_SIZE ); +sysbus_init_mmio (sbd, &sdhci->iomem); +sysbus_mmio_map (sbd,0 , sdhci->ioaddr);Here also.
Quoted 89 lines
hw/sd/aspeed_sdhci.c new file +42
-0+} + +static void aspeed_sdhci_reset (DeviceState *dev) +{ +AspeedSDHCIState *sdhci =ASPEED_SDHCI (dev); + +memset (sdhci->regs,0 ,ASPEED_SDHCI_REG_SIZE ); + sdhci->regs[TO_REG (ASPEED_SDHCI_INFO )] =ASPEED_SDHCI_INFO_RESET ; + sdhci->regs[TO_REG (ASPEED_SDHCI_DEBOUNCE )] =ASPEED_SDHCI_DEBOUNCE_RESET ; +} + +static const VMStateDescription vmstate_aspeed_sdhci = { + .name =TYPE_ASPEED_SDHCI , + .version_id =1 , + .fields = (VMStateField []) { +VMSTATE_UINT32_ARRAY (regs, AspeedSDHCIState,ASPEED_SDHCI_NUM_REGS ), +VMSTATE_END_OF_LIST (), + }, +}; + +static void aspeed_sdhci_class_init (ObjectClass *classp,void *data) +{ +DeviceClass *dc =DEVICE_CLASS (classp); + + dc->realize = aspeed_sdhci_realize; + dc->reset = aspeed_sdhci_reset; + dc->vmsd = &vmstate_aspeed_sdhci; +} + +static TypeInfo aspeed_sdhci_info = { + .name =TYPE_ASPEED_SDHCI , + .parent =TYPE_SYS_BUS_DEVICE , + .instance_size =sizeof (AspeedSDHCIState), + .class_init = aspeed_sdhci_class_init, +}; + +static void aspeed_sdhci_register_types (void ) +{ +type_register_static (&aspeed_sdhci_info); +} + +type_init (aspeed_sdhci_register_types)include/hw/arm/aspeed_soc.h +3
-0@@ -29,6 +29,7 @@#include "hw/misc/aspeed_pwm.h" #include "hw/misc/aspeed_lpc.h" #include "hw/misc/aspeed_fsi.h" +#include "hw/sd/aspeed_sdhci.h" #define ASPEED_SPIS_NUM 2#define ASPEED_WDTS_NUM 4@@ -62,6 +63,7 @@ typedef struct AspeedSoCState {AspeedPWMState pwm;AspeedLPCState lpc;AspeedFsiState fsi[2 ]; +AspeedSDHCIState sdhci; } AspeedSoCState;#define TYPE_ASPEED_SOC "aspeed-soc"@@ -107,6 +109,7 @@ enum {ASPEED_ADC ,ASPEED_VIDEO ,ASPEED_SRAM , +ASPEED_SDHCI ,ASPEED_GPIO ,ASPEED_RTC , ASPEED_TIMER1,include/hw/sd/aspeed_sdhci.h new file +13
-0@@ -0,0 +1,35 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#ifndef ASPEED_SDHCI_H +#define ASPEED_SDHCI_H + +#include "hw/sd/sdhci.h" +include/hw/sd/aspeed_sdhci.h line 23
Quoted 10 lines
include/hw/sd/aspeed_sdhci.h new file +10
-0+#define TYPE_ASPEED_SDHCI "aspeed.sdhci" +#define ASPEED_SDHCI (obj) OBJECT_CHECK(AspeedSDHCIState, (obj), \ + TYPE_ASPEED_SDHCI) + +#define ASPEED_SDHCI_CAPABILITIES 0x01E80080 +#define ASPEED_SDHCI_NUM_SLOTS 2 +#define ASPEED_SDHCI_NUM_REGS (ASPEED_SDHCI_REG_SIZE / sizeof(uint32_t)) +#define ASPEED_SDHCI_REG_SIZE 0x100 + +typedef struct AspeedSDHCIState {AspeedSDHost may be ? It's the SoC SD controller.
Quoted 4 lines
include/hw/sd/aspeed_sdhci.h line 26
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +3
-0+SysBusDevice parent; + +SDHCIState slots[ASPEED_SDHCI_NUM_SLOTS ];I think the SoC should own the SD slots.
Then it would be tricky/impossible to access the slots from the Aspeed SD specific functions? For example to connect the irqs and either mirror the regs or do some alias mapping.
Thanks for the quick review!
Eddie
Quoted 17 lines
include/hw/sd/aspeed_sdhci.h line 28
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +2
-0+ +hwaddr ioaddr;not needed.
include/hw/sd/aspeed_sdhci.h line 35
Quoted 8 lines
include/hw/sd/aspeed_sdhci.h new file +7
-0+MemoryRegion iomem; +qemu_irq irq; + + uint32_t regs[ASPEED_SDHCI_NUM_REGS]; +} AspeedSDHCIState; + +#endif /* ASPEED_SDHCI_H */ Thanks,
C.
Quoted 17 lines · Eddie James
Quoted 15 lines · Cédric Le Goater
Hello Eddie,
Quoted 12 lines · Eddie James
+ sdhci->slots[0].capareg = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_148:
+ sdhci->slots[0].maxcurr = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_240:
+ sdhci->slots[1].capareg = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_248:
+ sdhci->slots[1].maxcurr = (uint64_t)(uint32_t)val; + break;
Quoted 4 lines
Quoted 1 line
I think these regs are readonly.
Well the actual regs at slot + 0x40/0x48 are indeed, but not the Aspeed-specific ones that mirror there. I think the idea is that Aspeed-specific code can set it's capabilities differently if desired. This may prevent the use of alias regions here.
Actually I could be wrong after reading the specs again. It's a little confusing. I'm fine with making it read-only anyway, I doubt there will be any code that needs to write it.
Quoted 15 lines
Quoted 15 lines
Quoted 15 lines
+ default:
+ if (addr <ASPEED_SDHCI_REG_SIZE) {
+ sdhci->regs[TO_REG(addr)] = (uint32_t)val; + }
+ }hw/sd/aspeed_sdhci.c new file +7
-0+} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness = DEVICE_NATIVE_ENDIAN, + .valid.min_access_size = 4,
Quoted 19 lines · Eddie James
Quoted 17 lines · Eddie James
Quoted 15 lines · Cédric Le Goater
Hello Eddie,
Quoted 12 lines · Eddie James
+ sdhci->slots[0].capareg = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_148:
+ sdhci->slots[0].maxcurr = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_240:
+ sdhci->slots[1].capareg = (uint64_t)(uint32_t)val; + break;
+ caseASPEED_SDHCI_SDIO_248:
+ sdhci->slots[1].maxcurr = (uint64_t)(uint32_t)val; + break;
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
I think these regs are readonly.
Well the actual regs at slot + 0x40/0x48 are indeed, but not the Aspeed-specific ones that mirror there. I think the idea is that Aspeed-specific code can set it's capabilities differently if desired. This may prevent the use of alias regions here.
Actually I could be wrong after reading the specs again. It's a little confusing. I'm fine with making it read-only anyway, I doubt there will be any code that needs to write it.
Let's ask Aspeed what was the intent. It is a bit weird indeed.
Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
+ default:
+ if (addr <ASPEED_SDHCI_REG_SIZE) {
+ sdhci->regs[TO_REG(addr)] = (uint32_t)val; + }
+ }hw/sd/aspeed_sdhci.c new file +7
-0+} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness = DEVICE_NATIVE_ENDIAN, + .valid.min_access_size = 4,
Quoted 64 lines · Eddie James
Quoted 52 lines · Cédric Le Goater
Hello Eddie,
Quoted 9 lines · Eddie James
The Aspeed SOCs have two SD/MMC controllers. Add a device that encapsulates both of these controllers and models the Aspeed-specific registers and behavior.
Tested by reading from mmcblk0 in Linux:
qemu-system-arm -machine romulus-bmc -nographic -serial mon:stdio \
-drive file=_tmp/flash-romulus,format=raw,if=mtd \
-device sd-card,drive=sd0 -drive file=_tmp/kernel,format=raw,if=sdQuoted 4 lines
Signed-off-by Eddie James <[email protected]>---
This patch applies on top of Cedric's set of recent Aspeed changes. Therefore, I'm sending as an RFC rather than a patch.yes. we can worked that out when the patch is reviewed. You can base on mainline when ready. My tree serves as an overall test base.
Quoted 5 lines
Changes since v1:
- Move slot realize code into the Aspeed SDHCI realize function
- Fix interrupt handling by creating input irqs and connecting them to the slot irqs.
- Removed card device creation codeI think all the code is here but it needs some more reshuffling :) The raspi machine is a good source for modelling pratices.
Quoted 25 lines
hw/arm/aspeed.c | 1 -
hw/arm/aspeed_soc.c | 24 ++++++
hw/sd/Makefile.objs | 1 +
hw/sd/aspeed_sdhci.c | 190 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/aspeed_soc.h | 3 +
include/hw/sd/aspeed_sdhci.h| 35 ++++++++6 files changed, 253 insertions(+), 1 deletion(-)create mode 100644
hw/sd/aspeed_sdhci.c
create mode 100644include/hw/sd/aspeed_sdhci.hhw/arm/aspeed.c +0
-1@@ -480,7 +480,6 @@ static void aspeed_machine_class_init(ObjectClass *oc, void *data) mc->desc = board->desc; mc->init = aspeed_machine_init; mc->max_cpus = ASPEED_CPUS_NUM; - mc->no_sdcard = 1; mc->no_floppy = 1; mc->no_cdrom = 1; mc->no_parallel = 1;hw/arm/aspeed_soc.c +0
-1--- a/hw/arm/aspeed_soc.cQuoted 8 lines
hw/arm/aspeed_soc.c line 25
Quoted 6 lines
+++
b/hw/arm/aspeed_soc.chw/arm/aspeed_soc.c +1
-0@@ -22,6 +22,7 @@ #include "qemu/error-report.h" #include "hw/i2c/aspeed_i2c.h" #include "net/net.h" +#include "sysemu/blockdev.h" I would expect block devices to be handled at the machine level in
aspeed.c, like the flash devices are. Something like :
Quoted 1 line
OK, I did have that in v1 but Peter mentioned it was typically done at the command line?
yes, indeed. This works also and this is less code which is even better.
Quoted 1 line
I guess it can go in
aspeed.ctoo.
Well, let's do without if we can.
We might be able to get rid of aspeed_board_init_flashes() now.
We should start writing a QEMU cookbook page on the OpenBMC wiki to document the Aspeed machine command line.
Quoted 196 lines
Quoted 186 lines
/* Create and plug in the SD cards */
for (i = 0; i <ASPEED_SDHCI_NUM_SLOTS; i++) { BusState *bus;
DriveInfo *di =drive_get_next(IF_SD); BlockBackend *blk = di ?blk_by_legacy_dinfo(di): NULL; DeviceState *carddev;bus =
qdev_get_child_bus(DEVICE(&s->soc), "sd-bus"); if (!bus) {
error_report("No SD bus found for SD card %d", i); exit(1);
}
carddev =qdev_create(bus, TYPE_SD_CARD);qdev_prop_set_drive(carddev, "drive", blk, &error_fatal);object_property_set_bool(OBJECT(carddev), true, "realized", &error_fatal);
}Quoted 33 lines
hw/arm/aspeed_soc.c +0
-0#define ASPEED_SOC_IOMEM_SIZE 0x00200000 @@ -62,6 +63,7 @@ static const hwaddr aspeed_soc_ast2500_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000,[ASPEED_ADC] =
0x1E6E9000,
[ASPEED_SRAM] =0x1E720000,
+ [ASPEED_SDHCI] =0x1E740000,
[ASPEED_GPIO] =0x1E780000,
[ASPEED_RTC] =0x1E781000,
[ASPEED_TIMER1] =0x1E782000,hw/arm/aspeed_soc.c +3
-0@@ -100,6 +102,7 @@ static const hwaddr aspeed_soc_ast2600_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_VIDEO] = 0x1E700000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -146,6 +149,7 @@ static const int aspeed_soc_ast2400_irqmap[] = { [ASPEED_ETH1] = 2, [ASPEED_ETH2] = 3, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 26, }; #define aspeed_soc_ast2500_irqmap aspeed_soc_ast2400_irqmap@@ -163,6 +167,7 @@ static const int aspeed_soc_ast2600_irqmap[] = { [ASPEED_SDMC] = 0, [ASPEED_SCU] = 12, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 43, [ASPEED_ADC] = 46, [ASPEED_GPIO] = 40, [ASPEED_RTC] = 13,hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 360
Quoted 6 lines
hw/arm/aspeed_soc.c +3
-0@@ -350,6 +355,15 @@ static void aspeed_soc_init(Object *obj) sysbus_init_child_obj(obj, "fsi[*]", OBJECT(&s->fsi[0]), sizeof(s->fsi[0]), TYPE_ASPEED_FSI); } + + sysbus_init_child_obj(obj, "sdhci", OBJECT(&s->sdhci), sizeof(s->sdhci), + TYPE_ASPEED_SDHCI);This is the Aspeed SD host interface. May be we should call it sdhost ?
I suppose this is our "sd-bus" device ?
hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 363
Quoted 2 lines
hw/arm/aspeed_soc.c +2
-0+ /* Init sd card slot class here so that they're under the correct parent */ + for (i = 0; i < ASPEED_SDHCI_NUM_SLOTS; ++i) {and these are the slots, I would put them at the SoC level.
hw/arm/aspeed_soc.c static void aspeed_soc_realize(DeviceState *dev, Error **errp) · line 699
Quoted 12 lines
hw/arm/aspeed_soc.c +6
-0+ sysbus_init_child_obj(obj, "sdhci_slot[*]", OBJECT(&s->sdhci.slots[i]), + sizeof(s->sdhci.slots[i]), TYPE_SYSBUS_SDHCI); + } } /*@@ -680,6 +694,16 @@ static void aspeed_soc_realize(DeviceState *dev, Error **errp) sysbus_connect_irq(SYS_BUS_DEVICE(&s->fsi[0]), 0, aspeed_soc_get_irq(s, ASPEED_FSI1)); } + + /* SD/SDIO - set the reg address so slot memory mapping can be set up */ + s->sdhci.ioaddr = sc->info->memmap[ASPEED_SDHCI];That's weird. We do all mappings in the SoC.
I think you should be realizing the slots here also. See the other SoCs, XlnxZynqMPState for instance.
Quoted 63 lines
hw/arm/aspeed_soc.c +7
-0+ object_property_set_bool(OBJECT(&s->sdhci), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sysbus_connect_irq(SYS_BUS_DEVICE(&s->sdhci), 0, + aspeed_soc_get_irq(s, ASPEED_SDHCI)); } static Property aspeed_soc_properties[] = {
DEFINE_PROP_UINT32("num-cpus", AspeedSoCState, num_cpus, 0),hw/sd/Makefile.objs +0
-0@@ -8,3 +8,4 @@ obj-$(CONFIG_MILKYMIST) += milkymist-memcard.oobj-$(CONFIG_OMAP) += omap_mmc.o
obj-$(CONFIG_PXA2XX) += pxa2xx_mmci.o
obj-$(CONFIG_RASPI) += bcm2835_sdhost.ohw/sd/Makefile.objs +1
-0+obj-$(CONFIG_ASPEED_SOC) += aspeed_sdhci.ohw/sd/aspeed_sdhci.c new file +37
-0@@ -0,0 +1,190 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qemu/error-report.h" +#include "hw/sd/aspeed_sdhci.h" +#include "qapi/error.h" + +#define ASPEED_SDHCI_INFO 0x00 +#define ASPEED_SDHCI_INFO_RESET 0x00030000 +#define ASPEED_SDHCI_DEBOUNCE 0x04 +#define ASPEED_SDHCI_DEBOUNCE_RESET 0x00000005 +#define ASPEED_SDHCI_BUS 0x08 +#define ASPEED_SDHCI_SDIO_140 0x10 +#define ASPEED_SDHCI_SDIO_148 0x18 +#define ASPEED_SDHCI_SDIO_240 0x20 +#define ASPEED_SDHCI_SDIO_248 0x28 +#define ASPEED_SDHCI_WP_POL 0xec +#define ASPEED_SDHCI_CARD_DET 0xf0 +#define ASPEED_SDHCI_IRQ_STAT 0xfc + +#define TO_REG (addr) ((addr) / sizeof(uint32_t)) + +static uint64_t aspeed_sdhci_read (void *opaque,hwaddr addr,unsigned int size) +{ + uint32_t val = 0; + AspeedSDHCIState *sdhci = opaque; + + switch (addr) { + case ASPEED_SDHCI_SDIO_140: + val = (uint32_t)sdhci->slots[0].capareg;hw/sd/aspeed_sdhci.c line 47
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ break; + case ASPEED_SDHCI_SDIO_148: + val = (uint32_t)sdhci->slots[0].maxcurr; + break; + case ASPEED_SDHCI_SDIO_240: + val = (uint32_t)sdhci->slots[1].capareg; + break; + case ASPEED_SDHCI_SDIO_248: + val = (uint32_t)sdhci->slots[1].maxcurr; + break;We could mirror the 16bytes segment for [
SDHC_CAPAB...SDHC_MAXCURR+ 4 ] of each slot under the MMIO region of the Aspeed SD controller at offset:
(slot + 1) *0x10. If that worked, we wouldn't need these redirections.I think you need alias regions.
hw/sd/aspeed_sdhci.c line 51
Quoted 4 lines
hw/sd/aspeed_sdhci.c new file +4
-0+ default: + if (addr < ASPEED_SDHCI_REG_SIZE) { + val = sdhci->regs[TO_REG(addr)]; + }we could return some errors for not implemented registers.
Quoted 16 lines
hw/sd/aspeed_sdhci.c new file +16
-0+ } + + return (uint64_t)val; +} + +static void aspeed_sdhci_write (void *opaque,hwaddr addr,uint64_t val, + unsigned int size) +{ + AspeedSDHCIState *sdhci = opaque; + + switch (addr) { + case ASPEED_SDHCI_SDIO_140: + sdhci->slots[0].capareg = (uint64_t)(uint32_t)val; + break; + case ASPEED_SDHCI_SDIO_148: + sdhci->slots[0].maxcurr = (uint64_t)(uint32_t)val;Quoted 8 lines
hw/sd/aspeed_sdhci.c line 74
Quoted 7 lines
hw/sd/aspeed_sdhci.c new file +7
-0+ break; + case ASPEED_SDHCI_SDIO_240: + sdhci->slots[1].capareg = (uint64_t)(uint32_t)val; + break; + case ASPEED_SDHCI_SDIO_248: + sdhci->slots[1].maxcurr = (uint64_t)(uint32_t)val; + break;I think these regs are readonly.
Quoted 1 line
Well the actual regs at slot + 0x40/0x48 are indeed, but not the Aspeed-specific ones that mirror there. I think the idea is that Aspeed-specific code can set it's capabilities differently if desired. This may prevent the use of alias regions here.
ah yes. The SD controller regs can be used to HW init the slots.
This is unfortunate. So your model is fine. I don't see any other elegant ways to do these settings.
Quoted 31 lines
Quoted 31 lines
Quoted 19 lines
hw/sd/aspeed_sdhci.c new file +18
-0+ default: + if (addr < ASPEED_SDHCI_REG_SIZE) { + sdhci->regs[TO_REG(addr)] = (uint32_t)val; + } + } +} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness = DEVICE_NATIVE_ENDIAN, + .valid.min_access_size = 4, + .valid.max_access_size = 4, +}; + +static void aspeed_sdhci_set_irq (void *opaque,int n,int level) +{ + AspeedSDHCIState *sdhci = opaque;hw/sd/aspeed_sdhci.c line 102
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ + if (level) { + sdhci->regs[TO_REG(ASPEED_SDHCI_IRQ_STAT)] |= BIT(n); + + qemu_irq_raise(sdhci->irq); + } else { + sdhci->regs[TO_REG(ASPEED_SDHCI_IRQ_STAT)] &= ~BIT(n); + + qemu_irq_lower(sdhci->irq); + }Doesn't that work the other way around ?
Quoted 6 lines
Quoted 3 lines
The SDHCI objects trigger their IRQ which call the
irq_notify()handler in which we need to deduce the slot number to updateASPEED_SDHCI_IRQ_STATand raise/lower the Aspeed SD host IRQ. I think that's how it should work.That's exactly what's happening here. Maybe my variable/function naming is confusing?
Sorry. I was looking at the aspeed-4.1 tree which includes your previous patch with the irq_notify() handler and I got confused. This looks good.
Quoted 43 lines
Quoted 33 lines
Quoted 33 lines
hw/sd/aspeed_sdhci.c new file +31
-0+} + +static void aspeed_sdhci_realize (DeviceState *dev,Error **errp) +{ + Error *err = NULL; + SysBusDevice *sbd = SYS_BUS_DEVICE(dev); + AspeedSDHCIState *sdhci = ASPEED_SDHCI(dev); + + /* Create input irqs for the slots */ + qdev_init_gpio_in_named_with_opaque(DEVICE(sbd), aspeed_sdhci_set_irq, + sdhci, NULL, ASPEED_SDHCI_NUM_SLOTS); + + for (int i = 0; i < ASPEED_SDHCI_NUM_SLOTS; ++i) { + hwaddr slot_addr = sdhci->ioaddr + (0x100 * (i + 1)); + Object *sdhci_slot = OBJECT(&sdhci->slots[i]); + SysBusDevice *sbd_slot = SYS_BUS_DEVICE(&sdhci->slots[i]); + + object_property_set_int(sdhci_slot, 2, "sd-spec-version", &err); + if (err) { + error_propagate(errp, err); + return;5f + } + + object_property_set_uint(sdhci_slot, ASPEED_SDHCI_CAPABILITIES, + "capareg", &err); + if (err) { + error_propagate(errp, err); + return; + } + + object_property_set_bool(sdhci_slot, true, "realized", &err);Quoted 8 lines
hw/sd/aspeed_sdhci.c line 139
Quoted 6 lines
hw/sd/aspeed_sdhci.c new file +6
-0+ if (err) { + error_propagate(errp, err); + return; + } + + sysbus_mmio_map(sbd_slot, 0, slot_addr);We should do the mapping at the SoC level and I would also put the slots at the SoC level.
Quoted 2 lines
OK. I did that in v1 but Peter made some comments about initializing things in the Aspeed SD code so I moved it all down here...
ok. we can keep most of the code here but not the mapping.
Would it be possible to add the memory regions of the SDHCIState objects as subregions of the AspeedSDHCIState memory region and do the mapping at the SoC level.
Quoted 110 lines
Quoted 110 lines
hw/sd/aspeed_sdhci.c line 148
Quoted 9 lines
hw/sd/aspeed_sdhci.c new file +8
-0+ sysbus_connect_irq(sbd_slot, 0, qdev_get_gpio_in(DEVICE(sbd), i)); + } + + sysbus_init_irq(sbd, &sdhci->irq); + memory_region_init_io(&sdhci->iomem, OBJECT(sdhci), &aspeed_sdhci_ops, + sdhci, TYPE_ASPEED_SDHCI, ASPEED_SDHCI_REG_SIZE); + sysbus_init_mmio(sbd, &sdhci->iomem); + sysbus_mmio_map(sbd, 0, sdhci->ioaddr);Here also.
Quoted 87 lines
hw/sd/aspeed_sdhci.c new file +42
-0+} + +static void aspeed_sdhci_reset (DeviceState *dev) +{ + AspeedSDHCIState *sdhci = ASPEED_SDHCI(dev); + + memset(sdhci->regs, 0, ASPEED_SDHCI_REG_SIZE); + sdhci->regs[TO_REG(ASPEED_SDHCI_INFO)] = ASPEED_SDHCI_INFO_RESET; + sdhci->regs[TO_REG(ASPEED_SDHCI_DEBOUNCE)] = ASPEED_SDHCI_DEBOUNCE_RESET; +} + +static const VMStateDescription vmstate_aspeed_sdhci = { + .name = TYPE_ASPEED_SDHCI, + .version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32_ARRAY(regs, AspeedSDHCIState, ASPEED_SDHCI_NUM_REGS), + VMSTATE_END_OF_LIST(), + }, +}; + +static void aspeed_sdhci_class_init (ObjectClass *classp,void *data) +{ + DeviceClass *dc = DEVICE_CLASS(classp); + + dc->realize = aspeed_sdhci_realize; + dc->reset = aspeed_sdhci_reset; + dc->vmsd = &vmstate_aspeed_sdhci; +} + +static TypeInfo aspeed_sdhci_info = { + .name = TYPE_ASPEED_SDHCI, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(AspeedSDHCIState), + .class_init = aspeed_sdhci_class_init, +}; + +static void aspeed_sdhci_register_types (void ) +{ + type_register_static(&aspeed_sdhci_info); +} + +type_init (aspeed_sdhci_register_types)include/hw/arm/aspeed_soc.h +3
-0@@ -29,6 +29,7 @@ #include "hw/misc/aspeed_pwm.h" #include "hw/misc/aspeed_lpc.h" #include "hw/misc/aspeed_fsi.h" +#include "hw/sd/aspeed_sdhci.h" #define ASPEED_SPIS_NUM 2 #define ASPEED_WDTS_NUM 4@@ -62,6 +63,7 @@ typedef struct AspeedSoCState { AspeedPWMState pwm; AspeedLPCState lpc; AspeedFsiState fsi[2]; + AspeedSDHCIState sdhci; } AspeedSoCState; #define TYPE_ASPEED_SOC "aspeed-soc"@@ -107,6 +109,7 @@ enum { ASPEED_ADC, ASPEED_VIDEO, ASPEED_SRAM, + ASPEED_SDHCI, ASPEED_GPIO, ASPEED_RTC, ASPEED_TIMER1,include/hw/sd/aspeed_sdhci.h new file +13
-0@@ -0,0 +1,35 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#ifndef ASPEED_SDHCI_H +#define ASPEED_SDHCI_H + +#include "hw/sd/sdhci.h" +include/hw/sd/aspeed_sdhci.h line 23
Quoted 10 lines
include/hw/sd/aspeed_sdhci.h new file +10
-0+#define TYPE_ASPEED_SDHCI "aspeed.sdhci" +#define ASPEED_SDHCI (obj) OBJECT_CHECK(AspeedSDHCIState, (obj), \ + TYPE_ASPEED_SDHCI) + +#define ASPEED_SDHCI_CAPABILITIES 0x01E80080 +#define ASPEED_SDHCI_NUM_SLOTS 2 +#define ASPEED_SDHCI_NUM_REGS (ASPEED_SDHCI_REG_SIZE / sizeof(uint32_t)) +#define ASPEED_SDHCI_REG_SIZE 0x100 + +typedef struct AspeedSDHCIState {AspeedSDHost may be ? It's the SoC SD controller.
Quoted 7 lines
Quoted 4 lines
include/hw/sd/aspeed_sdhci.h line 26
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +3
-0+ SysBusDevice parent; + + SDHCIState slots[ASPEED_SDHCI_NUM_SLOTS];I think the SoC should own the SD slots.
Then it would be tricky/impossible to access the slots from the Aspeed SD specific functions? For example to connect the irqs and either mirror the regs or do some alias mapping.
yes. Forget that suggestion.
So the only thing I would change is how the mapping is done. We should be able to use memory_region_add_subregion() I think.
Thanks,
C.
Quoted 23 lines
Thanks for the quick review!
Eddie
Quoted 17 lines
include/hw/sd/aspeed_sdhci.h line 28
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +2
-0+ + hwaddr ioaddr;not needed.
include/hw/sd/aspeed_sdhci.h line 35
Quoted 8 lines
include/hw/sd/aspeed_sdhci.h new file +7
-0+ MemoryRegion iomem; + qemu_irq irq; + + uint32_t regs[ASPEED_SDHCI_NUM_REGS]; +} AspeedSDHCIState; + +#endif /* ASPEED_SDHCI_H */ Thanks,
C.
Quoted 354 lines · Cédric Le Goater
Quoted 62 lines · Eddie James
Quoted 52 lines · Cédric Le Goater
Hello Eddie,
Quoted 9 lines · Eddie James
The Aspeed SOCs have two SD/MMC controllers. Add a device that encapsulates both of these controllers and models the Aspeed-specific registers and behavior.
Tested by reading from mmcblk0 in Linux:
qemu-system-arm -machine romulus-bmc -nographic -serial mon:stdio \
-drive file=_tmp/flash-romulus,format=raw,if=mtd \
-device sd-card,drive=sd0 -drive file=_tmp/kernel,format=raw,if=sdQuoted 4 lines
Signed-off-by Eddie James <[email protected]>---
This patch applies on top of Cedric's set of recent Aspeed changes. Therefore, I'm sending as an RFC rather than a patch.yes. we can worked that out when the patch is reviewed. You can base on mainline when ready. My tree serves as an overall test base.
Quoted 5 lines
Changes since v1:
- Move slot realize code into the Aspeed SDHCI realize function
- Fix interrupt handling by creating input irqs and connecting them to the slot irqs.
- Removed card device creation codeI think all the code is here but it needs some more reshuffling :)
The raspi machine is a good source for modelling pratices.Quoted 25 lines
hw/arm/aspeed.c | 1 - hw/arm/aspeed_soc.c | 24 ++++++ hw/sd/Makefile.objs | 1 + hw/sd/aspeed_sdhci.c | 190 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/aspeed_soc.h | 3 + include/hw/sd/aspeed_sdhci.h | 35 ++++++++ 6 files changed, 253 insertions(+), 1 deletion(-)create mode 100644
hw/sd/aspeed_sdhci.ccreate mode 100644include/hw/sd/aspeed_sdhci.hhw/arm/aspeed.c +0
-1@@ -480,7 +480,6 @@ static void aspeed_machine_class_init(ObjectClass *oc, void *data) mc->desc = board->desc; mc->init = aspeed_machine_init; mc->max_cpus = ASPEED_CPUS_NUM; - mc->no_sdcard = 1; mc->no_floppy = 1; mc->no_cdrom = 1; mc->no_parallel = 1;hw/arm/aspeed_soc.c +0
-1--- a/hw/arm/aspeed_soc.cQuoted 8 lines
hw/arm/aspeed_soc.c line 25
Quoted 6 lines
+++
b/hw/arm/aspeed_soc.chw/arm/aspeed_soc.c +1
-0@@ -22,6 +22,7 @@ #include "qemu/error-report.h" #include "hw/i2c/aspeed_i2c.h" #include "net/net.h" +#include "sysemu/blockdev.h" I would expect block devices to be handled at the machine level in
aspeed.c, like the flash devices are. Something like :Quoted 1 line
OK, I did have that in v1 but Peter mentioned it was typically done at the command line?
yes, indeed. This works also and this is less code which is even better.
Quoted 1 line
I guess it can go in
aspeed.ctoo.Well, let's do without if we can.
We might be able to get rid of
aspeed_board_init_flashes()now.We should start writing a QEMU cookbook page on the OpenBMC wiki to document the Aspeed machine command line.
Quoted 193 lines
Quoted 184 lines
/* Create and plug in the SD cards */
for (i = 0; i <ASPEED_SDHCI_NUM_SLOTS; i++) { BusState *bus;
DriveInfo *di =drive_get_next(IF_SD); BlockBackend *blk = di ?blk_by_legacy_dinfo(di): NULL; DeviceState *carddev;bus =
qdev_get_child_bus(DEVICE(&s->soc), "sd-bus"); if (!bus) {
error_report("No SD bus found for SD card %d", i); exit(1);
}
carddev =qdev_create(bus, TYPE_SD_CARD);qdev_prop_set_drive(carddev, "drive", blk, &error_fatal);object_property_set_bool(OBJECT(carddev), true, "realized", &error_fatal);
}Quoted 33 lines
hw/arm/aspeed_soc.c +0
-0#define ASPEED_SOC_IOMEM_SIZE 0x00200000 @@ -62,6 +63,7 @@ static const hwaddr aspeed_soc_ast2500_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000,[ASPEED_ADC] =
0x1E6E9000,
[ASPEED_SRAM] =0x1E720000,
+ [ASPEED_SDHCI] =0x1E740000,
[ASPEED_GPIO] =0x1E780000,
[ASPEED_RTC] =0x1E781000,
[ASPEED_TIMER1] =0x1E782000,hw/arm/aspeed_soc.c +3
-0@@ -100,6 +102,7 @@ static const hwaddr aspeed_soc_ast2600_memmap[] = { [ASPEED_XDMA] = 0x1E6E7000, [ASPEED_ADC] = 0x1E6E9000, [ASPEED_VIDEO] = 0x1E700000, + [ASPEED_SDHCI] = 0x1E740000, [ASPEED_GPIO] = 0x1E780000, [ASPEED_RTC] = 0x1E781000, [ASPEED_TIMER1] = 0x1E782000,@@ -146,6 +149,7 @@ static const int aspeed_soc_ast2400_irqmap[] = { [ASPEED_ETH1] = 2, [ASPEED_ETH2] = 3, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 26, }; #define aspeed_soc_ast2500_irqmap aspeed_soc_ast2400_irqmap@@ -163,6 +167,7 @@ static const int aspeed_soc_ast2600_irqmap[] = { [ASPEED_SDMC] = 0, [ASPEED_SCU] = 12, [ASPEED_XDMA] = 6, + [ASPEED_SDHCI] = 43, [ASPEED_ADC] = 46, [ASPEED_GPIO] = 40, [ASPEED_RTC] = 13,hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 360
Quoted 6 lines
hw/arm/aspeed_soc.c +3
-0@@ -350,6 +355,15 @@ static void aspeed_soc_init(Object *obj) sysbus_init_child_obj(obj, "fsi[*]", OBJECT(&s->fsi[0]), sizeof(s->fsi[0]), TYPE_ASPEED_FSI); } + + sysbus_init_child_obj(obj, "sdhci", OBJECT(&s->sdhci), sizeof(s->sdhci), + TYPE_ASPEED_SDHCI);This is the Aspeed SD host interface. May be we should call it sdhost ?
I suppose this is our "sd-bus" device ?
hw/arm/aspeed_soc.c static void aspeed_soc_init(Object *obj) · line 363
Quoted 2 lines
hw/arm/aspeed_soc.c +2
-0+ /* Init sd card slot class here so that they're under the correct parent */ + for (i = 0; i < ASPEED_SDHCI_NUM_SLOTS; ++i) {and these are the slots, I would put them at the SoC level.
hw/arm/aspeed_soc.c static void aspeed_soc_realize(DeviceState *dev, Error **errp) · line 699
Quoted 12 lines
hw/arm/aspeed_soc.c +6
-0+ sysbus_init_child_obj(obj, "sdhci_slot[*]", OBJECT(&s->sdhci.slots[i]), + sizeof(s->sdhci.slots[i]), TYPE_SYSBUS_SDHCI); + } } /*@@ -680,6 +694,16 @@ static void aspeed_soc_realize(DeviceState *dev, Error **errp) sysbus_connect_irq(SYS_BUS_DEVICE(&s->fsi[0]), 0, aspeed_soc_get_irq(s, ASPEED_FSI1)); } + + /* SD/SDIO - set the reg address so slot memory mapping can be set up */ + s->sdhci.ioaddr = sc->info->memmap[ASPEED_SDHCI];That's weird. We do all mappings in the SoC.
I think you should be realizing the slots here also. See the other SoCs, XlnxZynqMPState for instance.
Quoted 63 lines
hw/arm/aspeed_soc.c +7
-0+ object_property_set_bool(OBJECT(&s->sdhci), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sysbus_connect_irq(SYS_BUS_DEVICE(&s->sdhci), 0, + aspeed_soc_get_irq(s, ASPEED_SDHCI)); } static Property aspeed_soc_properties[] = {
DEFINE_PROP_UINT32("num-cpus", AspeedSoCState, num_cpus, 0),hw/sd/Makefile.objs +1
-0@@ -8,3 +8,4 @@ obj-$(CONFIG_MILKYMIST) += milkymist-memcard.o obj-$(CONFIG_OMAP) += omap_mmc.o obj-$(CONFIG_PXA2XX) += pxa2xx_mmci.o obj-$(CONFIG_RASPI) += bcm2835_sdhost.o +obj-$(CONFIG_ASPEED_SOC) += aspeed_sdhci.ohw/sd/aspeed_sdhci.c new file +37
-0@@ -0,0 +1,190 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qemu/error-report.h" +#include "hw/sd/aspeed_sdhci.h" +#include "qapi/error.h" + +#define ASPEED_SDHCI_INFO 0x00 +#define ASPEED_SDHCI_INFO_RESET 0x00030000 +#define ASPEED_SDHCI_DEBOUNCE 0x04 +#define ASPEED_SDHCI_DEBOUNCE_RESET 0x00000005 +#define ASPEED_SDHCI_BUS 0x08 +#define ASPEED_SDHCI_SDIO_140 0x10 +#define ASPEED_SDHCI_SDIO_148 0x18 +#define ASPEED_SDHCI_SDIO_240 0x20 +#define ASPEED_SDHCI_SDIO_248 0x28 +#define ASPEED_SDHCI_WP_POL 0xec +#define ASPEED_SDHCI_CARD_DET 0xf0 +#define ASPEED_SDHCI_IRQ_STAT 0xfc + +#define TO_REG (addr) ((addr) / sizeof(uint32_t)) + +static uint64_t aspeed_sdhci_read (void *opaque,hwaddr addr,unsigned int size) +{ + uint32_t val = 0; + AspeedSDHCIState *sdhci = opaque; + + switch (addr) { + case ASPEED_SDHCI_SDIO_140: + val = (uint32_t)sdhci->slots[0].capareg;hw/sd/aspeed_sdhci.c line 47
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ break; + case ASPEED_SDHCI_SDIO_148: + val = (uint32_t)sdhci->slots[0].maxcurr; + break; + case ASPEED_SDHCI_SDIO_240: + val = (uint32_t)sdhci->slots[1].capareg; + break; + case ASPEED_SDHCI_SDIO_248: + val = (uint32_t)sdhci->slots[1].maxcurr; + break;We could mirror the 16bytes segment for [
SDHC_CAPAB...SDHC_MAXCURR+ 4 ] of each slot under the MMIO region of the Aspeed SD controller at offset:
(slot + 1) *0x10. If that worked, we wouldn't need these redirections.I think you need alias regions.
hw/sd/aspeed_sdhci.c line 51
Quoted 4 lines
hw/sd/aspeed_sdhci.c new file +4
-0+ default: + if (addr < ASPEED_SDHCI_REG_SIZE) { + val = sdhci->regs[TO_REG(addr)]; + }we could return some errors for not implemented registers.
Quoted 16 lines
hw/sd/aspeed_sdhci.c new file +16
-0+ } + + return (uint64_t)val; +} + +static void aspeed_sdhci_write (void *opaque,hwaddr addr,uint64_t val, + unsigned int size) +{ + AspeedSDHCIState *sdhci = opaque; + + switch (addr) { + case ASPEED_SDHCI_SDIO_140: + sdhci->slots[0].capareg = (uint64_t)(uint32_t)val; + break; + case ASPEED_SDHCI_SDIO_148: + sdhci->slots[0].maxcurr = (uint64_t)(uint32_t)val;Quoted 8 lines
hw/sd/aspeed_sdhci.c line 74
Quoted 7 lines
hw/sd/aspeed_sdhci.c new file +7
-0+ break; + case ASPEED_SDHCI_SDIO_240: + sdhci->slots[1].capareg = (uint64_t)(uint32_t)val; + break; + case ASPEED_SDHCI_SDIO_248: + sdhci->slots[1].maxcurr = (uint64_t)(uint32_t)val; + break;I think these regs are readonly.
Quoted 1 line
Well the actual regs at slot + 0x40/0x48 are indeed, but not the Aspeed-specific ones that mirror there. I think the idea is that Aspeed-specific code can set it's capabilities differently if desired. This may prevent the use of alias regions here.
ah yes. The SD controller regs can be used to HW init the slots.
This is unfortunate. So your model is fine. I don't see any other elegant ways to do these settings.
Quoted 30 lines
Quoted 30 lines
Quoted 18 lines
hw/sd/aspeed_sdhci.c new file +18
-0+ default: + if (addr < ASPEED_SDHCI_REG_SIZE) { + sdhci->regs[TO_REG(addr)] = (uint32_t)val; + } + } +} + +static const MemoryRegionOps aspeed_sdhci_ops = { + .read = aspeed_sdhci_read, + .write = aspeed_sdhci_write, + .endianness = DEVICE_NATIVE_ENDIAN, + .valid.min_access_size = 4, + .valid.max_access_size = 4, +}; + +static void aspeed_sdhci_set_irq (void *opaque,int n,int level) +{ + AspeedSDHCIState *sdhci = opaque;hw/sd/aspeed_sdhci.c line 102
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ + if (level) { + sdhci->regs[TO_REG(ASPEED_SDHCI_IRQ_STAT)] |= BIT(n); + + qemu_irq_raise(sdhci->irq); + } else { + sdhci->regs[TO_REG(ASPEED_SDHCI_IRQ_STAT)] &= ~BIT(n); + + qemu_irq_lower(sdhci->irq); + }Doesn't that work the other way around ?
Quoted 5 lines
Quoted 3 lines
The SDHCI objects trigger their IRQ which call the
irq_notify()handler in which we need to deduce the slot number to updateASPEED_SDHCI_IRQ_STATand raise/lower the Aspeed SD host IRQ. I think that's how it should work.That's exactly what's happening here. Maybe my variable/function naming is confusing?
Sorry. I was looking at the aspeed-4.1 tree which includes your previous patch with the
irq_notify()handler and I got confused. This looks good.Quoted 41 lines
Quoted 40 lines
Quoted 28 lines
hw/sd/aspeed_sdhci.c new file +27
-0+} + +static void aspeed_sdhci_realize (DeviceState *dev,Error **errp) +{ + Error *err = NULL; + SysBusDevice *sbd = SYS_BUS_DEVICE(dev); + AspeedSDHCIState *sdhci = ASPEED_SDHCI(dev); + + /* Create input irqs for the slots */ + qdev_init_gpio_in_named_with_opaque(DEVICE(sbd), aspeed_sdhci_set_irq, + sdhci, NULL, ASPEED_SDHCI_NUM_SLOTS); + + for (int i = 0; i < ASPEED_SDHCI_NUM_SLOTS; ++i) { + hwaddr slot_addr = sdhci->ioaddr + (0x100 * (i + 1)); + Object *sdhci_slot = OBJECT(&sdhci->slots[i]); + SysBusDevice *sbd_slot = SYS_BUS_DEVICE(&sdhci->slots[i]); + + object_property_set_int(sdhci_slot, 2, "sd-spec-version", &err); + if (err) { + error_propagate(errp, err); + return;5f + } + + object_property_set_uint(sdhci_slot, ASPEED_SDHCI_CAPABILITIES, + "capareg", &err); + if (err) { + error_propagate(errp, err);hw/sd/aspeed_sdhci.c line 139
Quoted 10 lines
hw/sd/aspeed_sdhci.c new file +10
-0+ return; + } + + object_property_set_bool(sdhci_slot, true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + + sysbus_mmio_map(sbd_slot, 0, slot_addr);We should do the mapping at the SoC level and I would also put the slots at the SoC level.
Quoted 7 lines
Quoted 2 lines
OK. I did that in v1 but Peter made some comments about initializing things in the Aspeed SD code so I moved it all down here...
ok. we can keep most of the code here but not the mapping.
Would it be possible to add the memory regions of the SDHCIState objects as subregions of the AspeedSDHCIState memory region and do the mapping at the SoC level.
I tried this but couldn't get it to work correctly. The generic SDHCI init/realize code uses the sysbus as the parent memory. The only way I could see it working is if I subregion'd the SDHCI SysBusDevice mmio memory region itself to the Aspeed region after the SDHCI init is done. But it was just making the code more unreadable with exactly the same result, so I didn't include this in v3.
Thanks,
Eddie
Quoted 146 lines
Quoted 110 lines
Quoted 110 lines
hw/sd/aspeed_sdhci.c line 148
Quoted 9 lines
hw/sd/aspeed_sdhci.c new file +8
-0+ sysbus_connect_irq(sbd_slot, 0, qdev_get_gpio_in(DEVICE(sbd), i)); + } + + sysbus_init_irq(sbd, &sdhci->irq); + memory_region_init_io(&sdhci->iomem, OBJECT(sdhci), &aspeed_sdhci_ops, + sdhci, TYPE_ASPEED_SDHCI, ASPEED_SDHCI_REG_SIZE); + sysbus_init_mmio(sbd, &sdhci->iomem); + sysbus_mmio_map(sbd, 0, sdhci->ioaddr);Here also.
Quoted 87 lines
hw/sd/aspeed_sdhci.c new file +42
-0+} + +static void aspeed_sdhci_reset (DeviceState *dev) +{ + AspeedSDHCIState *sdhci = ASPEED_SDHCI(dev); + + memset(sdhci->regs, 0, ASPEED_SDHCI_REG_SIZE); + sdhci->regs[TO_REG(ASPEED_SDHCI_INFO)] = ASPEED_SDHCI_INFO_RESET; + sdhci->regs[TO_REG(ASPEED_SDHCI_DEBOUNCE)] = ASPEED_SDHCI_DEBOUNCE_RESET; +} + +static const VMStateDescription vmstate_aspeed_sdhci = { + .name = TYPE_ASPEED_SDHCI, + .version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32_ARRAY(regs, AspeedSDHCIState, ASPEED_SDHCI_NUM_REGS), + VMSTATE_END_OF_LIST(), + }, +}; + +static void aspeed_sdhci_class_init (ObjectClass *classp,void *data) +{ + DeviceClass *dc = DEVICE_CLASS(classp); + + dc->realize = aspeed_sdhci_realize; + dc->reset = aspeed_sdhci_reset; + dc->vmsd = &vmstate_aspeed_sdhci; +} + +static TypeInfo aspeed_sdhci_info = { + .name = TYPE_ASPEED_SDHCI, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(AspeedSDHCIState), + .class_init = aspeed_sdhci_class_init, +}; + +static void aspeed_sdhci_register_types (void ) +{ + type_register_static(&aspeed_sdhci_info); +} + +type_init (aspeed_sdhci_register_types)include/hw/arm/aspeed_soc.h +3
-0@@ -29,6 +29,7 @@ #include "hw/misc/aspeed_pwm.h" #include "hw/misc/aspeed_lpc.h" #include "hw/misc/aspeed_fsi.h" +#include "hw/sd/aspeed_sdhci.h" #define ASPEED_SPIS_NUM 2 #define ASPEED_WDTS_NUM 4@@ -62,6 +63,7 @@ typedef struct AspeedSoCState { AspeedPWMState pwm; AspeedLPCState lpc; AspeedFsiState fsi[2]; + AspeedSDHCIState sdhci; } AspeedSoCState; #define TYPE_ASPEED_SOC "aspeed-soc"@@ -107,6 +109,7 @@ enum { ASPEED_ADC, ASPEED_VIDEO, ASPEED_SRAM, + ASPEED_SDHCI, ASPEED_GPIO, ASPEED_RTC, ASPEED_TIMER1,include/hw/sd/aspeed_sdhci.h new file +13
-0@@ -0,0 +1,35 @@ +/* +* Aspeed SD Host Controller +* Eddie James <[email protected]> +* +* Copyright (C) 2019 IBM Corp +* SPDX-License-Identifer: GPL-2.0-or-later +*/ + +#ifndef ASPEED_SDHCI_H +#define ASPEED_SDHCI_H + +#include "hw/sd/sdhci.h" +include/hw/sd/aspeed_sdhci.h line 23
Quoted 10 lines
include/hw/sd/aspeed_sdhci.h new file +10
-0+#define TYPE_ASPEED_SDHCI "aspeed.sdhci" +#define ASPEED_SDHCI (obj) OBJECT_CHECK(AspeedSDHCIState, (obj), \ + TYPE_ASPEED_SDHCI) + +#define ASPEED_SDHCI_CAPABILITIES 0x01E80080 +#define ASPEED_SDHCI_NUM_SLOTS 2 +#define ASPEED_SDHCI_NUM_REGS (ASPEED_SDHCI_REG_SIZE / sizeof(uint32_t)) +#define ASPEED_SDHCI_REG_SIZE 0x100 + +typedef struct AspeedSDHCIState {AspeedSDHost may be ? It's the SoC SD controller.
Quoted 6 lines
Quoted 4 lines
include/hw/sd/aspeed_sdhci.h line 26
Quoted 3 lines
include/hw/sd/aspeed_sdhci.h new file +3
-0+ SysBusDevice parent; + + SDHCIState slots[ASPEED_SDHCI_NUM_SLOTS];I think the SoC should own the SD slots.
Then it would be tricky/impossible to access the slots from the Aspeed SD specific functions? For example to connect the irqs and either mirror the regs or do some alias mapping.
yes. Forget that suggestion.
So the only thing I would change is how the mapping is done. We should be able to use
memory_region_add_subregion()I think.Thanks,
C.
Quoted 21 lines
Thanks for the quick review!
Eddie
Quoted 16 lines
include/hw/sd/aspeed_sdhci.h line 28
Quoted 2 lines
include/hw/sd/aspeed_sdhci.h new file +2
-0+ + hwaddr ioaddr;not needed.
include/hw/sd/aspeed_sdhci.h line 35
Quoted 8 lines
include/hw/sd/aspeed_sdhci.h new file +7
-0+ MemoryRegion iomem; + qemu_irq irq; + + uint32_t regs[ASPEED_SDHCI_NUM_REGS]; +} AspeedSDHCIState; + +#endif /* ASPEED_SDHCI_H */ Thanks,
C.