Re: [PATCH v6 06/10] hw/fsi: Aspeed APB2OPB interface

[Ninad Palsule <ninad@linux.ibm.com>]

Hello Cedric & Andrew,


On 10/24/23 10:21, Cédric Le Goater wrote:
> On 10/24/23 17:00, Ninad Palsule wrote:
>> Hello Cedric,
>>
>> On 10/24/23 02:46, Cédric Le Goater wrote:
>>> On 10/21/23 23:17, Ninad Palsule wrote:
>>>> This is a part of patchset where IBM's Flexible Service Interface is
>>>> introduced.
>>>>
>>>> An APB-to-OPB bridge enabling access to the OPB from the ARM core in
>>>> the AST2600. Hardware limitations prevent the OPB from being directly
>>>> mapped into APB, so all accesses are indirect through the bridge.
>>>>
>>>> Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
>>>> Signed-off-by: Ninad Palsule <ninad@linux.ibm.com>
>>>> ---
>>>> v2:
>>>> - Incorporated review comments by Joel
>>>> v3:
>>>> - Incorporated review comments by Thomas Huth
>>>> v4:
>>>>    - Compile FSI with ASPEED_SOC only.
>>>> v5:
>>>> - Incorporated review comments by Cedric.
>>>> v6:
>>>> - Incorporated review comments by Cedric.
>>>> ---
>>>>   include/hw/fsi/aspeed-apb2opb.h |  33 ++++
>>>>   hw/fsi/aspeed-apb2opb.c         | 280 
>>>> ++++++++++++++++++++++++++++++++
>>>>   hw/arm/Kconfig                  |   1 +
>>>>   hw/fsi/Kconfig                  |   4 +
>>>>   hw/fsi/meson.build              |   1 +
>>>>   hw/fsi/trace-events             |   2 +
>>>>   6 files changed, 321 insertions(+)
>>>>   create mode 100644 include/hw/fsi/aspeed-apb2opb.h
>>>>   create mode 100644 hw/fsi/aspeed-apb2opb.c
>>>>
>>>> diff --git a/include/hw/fsi/aspeed-apb2opb.h 
>>>> b/include/hw/fsi/aspeed-apb2opb.h
>>>> new file mode 100644
>>>> index 0000000000..a81ae67023
>>>> --- /dev/null
>>>> +++ b/include/hw/fsi/aspeed-apb2opb.h
>>>> @@ -0,0 +1,33 @@
>>>> +/*
>>>> + * SPDX-License-Identifier: GPL-2.0-or-later
>>>> + * Copyright (C) 2023 IBM Corp.
>>>> + *
>>>> + * ASPEED APB2OPB Bridge
>>>> + */
>>>> +#ifndef FSI_ASPEED_APB2OPB_H
>>>> +#define FSI_ASPEED_APB2OPB_H
>>>> +
>>>> +#include "hw/sysbus.h"
>>>> +#include "hw/fsi/opb.h"
>>>> +
>>>> +#define TYPE_ASPEED_APB2OPB "aspeed.apb2opb"
>>>> +OBJECT_DECLARE_SIMPLE_TYPE(AspeedAPB2OPBState, ASPEED_APB2OPB)
>>>> +
>>>> +#define ASPEED_APB2OPB_NR_REGS ((0xe8 >> 2) + 1)
>>>> +
>>>> +#define ASPEED_FSI_NUM 2
>>>> +
>>>> +typedef struct AspeedAPB2OPBState {
>>>> +    /*< private >*/
>>>> +    SysBusDevice parent_obj;
>>>> +
>>>> +    /*< public >*/
>>>> +    MemoryRegion iomem;
>>>> +
>>>> +    uint32_t regs[ASPEED_APB2OPB_NR_REGS];
>>>> +    qemu_irq irq;
>>>> +
>>>> +    OPBus opb[ASPEED_FSI_NUM];
>>>> +} AspeedAPB2OPBState;
>>>> +
>>>> +#endif /* FSI_ASPEED_APB2OPB_H */
>>>> diff --git a/hw/fsi/aspeed-apb2opb.c b/hw/fsi/aspeed-apb2opb.c
>>>> new file mode 100644
>>>> index 0000000000..6f97a6bc7d
>>>> --- /dev/null
>>>> +++ b/hw/fsi/aspeed-apb2opb.c
>>>> @@ -0,0 +1,280 @@
>>>> +/*
>>>> + * SPDX-License-Identifier: GPL-2.0-or-later
>>>> + * Copyright (C) 2023 IBM Corp.
>>>> + *
>>>> + * ASPEED APB-OPB FSI interface
>>>> + */
>>>> +
>>>> +#include "qemu/osdep.h"
>>>> +#include "qemu/log.h"
>>>> +#include "qom/object.h"
>>>> +#include "qapi/error.h"
>>>> +#include "trace.h"
>>>> +
>>>> +#include "hw/fsi/aspeed-apb2opb.h"
>>>> +#include "hw/qdev-core.h"
>>>> +
>>>> +#define TO_REG(x) (x >> 2)
>>>> +#define GENMASK(t, b) (((1ULL << ((t) + 1)) - 1) & ~((1ULL << (b)) 
>>>> - 1))
>>>> +
>>>> +#define APB2OPB_VERSION                    TO_REG(0x00)
>>>> +#define   APB2OPB_VERSION_VER              GENMASK(7, 0)
>>>> +
>>>> +#define APB2OPB_TRIGGER                    TO_REG(0x04)
>>>> +#define   APB2OPB_TRIGGER_EN               BIT(0)
>>>> +
>>>> +#define APB2OPB_CONTROL                    TO_REG(0x08)
>>>> +#define   APB2OPB_CONTROL_OFF              GENMASK(31, 13)
>>>> +
>>>> +#define APB2OPB_OPB2FSI                    TO_REG(0x0c)
>>>> +#define   APB2OPB_OPB2FSI_OFF              GENMASK(31, 22)
>>>> +
>>>> +#define APB2OPB_OPB0_SEL                   TO_REG(0x10)
>>>> +#define APB2OPB_OPB1_SEL                   TO_REG(0x28)
>>>> +#define   APB2OPB_OPB_SEL_EN               BIT(0)
>>>> +
>>>> +#define APB2OPB_OPB0_MODE                  TO_REG(0x14)
>>>> +#define APB2OPB_OPB1_MODE                  TO_REG(0x2c)
>>>> +#define   APB2OPB_OPB_MODE_RD              BIT(0)
>>>> +
>>>> +#define APB2OPB_OPB0_XFER                  TO_REG(0x18)
>>>> +#define APB2OPB_OPB1_XFER                  TO_REG(0x30)
>>>> +#define   APB2OPB_OPB_XFER_FULL            BIT(1)
>>>> +#define   APB2OPB_OPB_XFER_HALF            BIT(0)
>>>> +
>>>> +#define APB2OPB_OPB0_ADDR                  TO_REG(0x1c)
>>>> +#define APB2OPB_OPB0_WRITE_DATA            TO_REG(0x20)
>>>> +
>>>> +#define APB2OPB_OPB1_ADDR                  TO_REG(0x34)
>>>> +#define APB2OPB_OPB1_WRITE_DATA TO_REG(0x38)
>>>> +
>>>> +#define APB2OPB_IRQ_STS                    TO_REG(0x48)
>>>> +#define   APB2OPB_IRQ_STS_OPB1_TX_ACK      BIT(17)
>>>> +#define   APB2OPB_IRQ_STS_OPB0_TX_ACK      BIT(16)
>>>> +
>>>> +#define APB2OPB_OPB0_WRITE_WORD_ENDIAN     TO_REG(0x4c)
>>>> +#define   APB2OPB_OPB0_WRITE_WORD_ENDIAN_BE 0x0011101b
>>>> +#define APB2OPB_OPB0_WRITE_BYTE_ENDIAN     TO_REG(0x50)
>>>> +#define   APB2OPB_OPB0_WRITE_BYTE_ENDIAN_BE 0x0c330f3f
>>>> +#define APB2OPB_OPB1_WRITE_WORD_ENDIAN     TO_REG(0x54)
>>>> +#define APB2OPB_OPB1_WRITE_BYTE_ENDIAN     TO_REG(0x58)
>>>> +#define APB2OPB_OPB0_READ_BYTE_ENDIAN      TO_REG(0x5c)
>>>> +#define APB2OPB_OPB1_READ_BYTE_ENDIAN      TO_REG(0x60)
>>>> +#define   APB2OPB_OPB0_READ_WORD_ENDIAN_BE  0x00030b1b
>>>> +
>>>> +#define APB2OPB_OPB0_READ_DATA         TO_REG(0x84)
>>>> +#define APB2OPB_OPB1_READ_DATA         TO_REG(0x90)
>>>> +
>>>> +/*
>>>> + * The following magic values came from AST2600 data sheet
>>>> + * The register values are defined under section "FSI controller"
>>>> + * as initial values.
>>>> + */
>>>> +static const uint32_t aspeed_apb2opb_reset[ASPEED_APB2OPB_NR_REGS] 
>>>> = {
>>>> +     [APB2OPB_VERSION]                = 0x000000a1,
>>>> +     [APB2OPB_OPB0_WRITE_WORD_ENDIAN] = 0x0044eee4,
>>>> +     [APB2OPB_OPB0_WRITE_BYTE_ENDIAN] = 0x0055aaff,
>>>> +     [APB2OPB_OPB1_WRITE_WORD_ENDIAN] = 0x00117717,
>>>> +     [APB2OPB_OPB1_WRITE_BYTE_ENDIAN] = 0xffaa5500,
>>>> +     [APB2OPB_OPB0_READ_BYTE_ENDIAN]  = 0x0044eee4,
>>>> +     [APB2OPB_OPB1_READ_BYTE_ENDIAN]  = 0x00117717
>>>> +};
>>>> +
>>>> +static uint64_t fsi_aspeed_apb2opb_read(void *opaque, hwaddr addr,
>>>> +                                        unsigned size)
>>>> +{
>>>> +    AspeedAPB2OPBState *s = ASPEED_APB2OPB(opaque);
>>>> +
>>>> +    trace_fsi_aspeed_apb2opb_read(addr, size);
>>>> +
>>>> +    if (addr + size > sizeof(s->regs)) {
>>>
>>>
>>> hmm, the parameter 'size' is a memop transaction size not an address 
>>> offset.
>> OK, Changed it to validate the register (index) instead of addr + size.
>>>
>>>> + qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                      "%s: Out of bounds read: 0x%"HWADDR_PRIx" 
>>>> for %u\n",
>>>> +                      __func__, addr, size);
>>>> +        return 0;
>>>> +    }
>>>> +
>>>> +    return s->regs[TO_REG(addr)];
>>>> +}
>>>> +
>>>> +static void fsi_aspeed_apb2opb_write(void *opaque, hwaddr addr, 
>>>> uint64_t data,
>>>> +                                     unsigned size)
>>>> +{
>>>> +    AspeedAPB2OPBState *s = ASPEED_APB2OPB(opaque);
>>>> +
>>>> +    trace_fsi_aspeed_apb2opb_write(addr, size, data);
>>>> +
>>>> +    if (addr + size > sizeof(s->regs)) {
>>>
>>> same comment.
>> Fixed it same as above.
>>>
>>>
>>>> + qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                      "%s: Out of bounds write: %"HWADDR_PRIx" for 
>>>> %u\n",
>>>> +                      __func__, addr, size);
>>>> +        return;
>>>> +    }
>>>> +
>>>> +    switch (TO_REG(addr)) {
>>>> +    case APB2OPB_CONTROL:
>>>> +        fsi_opb_fsi_master_address(&s->opb[0], data & 
>>>> APB2OPB_CONTROL_OFF);
>>>
>>> fsi_opb_fsi_master_address() should statically defined in this file
>> We have separation of OPB bus implementation and APB2OPB interface. 
>> If we move this function here then we will be exposing OPB 
>> implementation here.
>>>
>>>> +        break;
>>>> +    case APB2OPB_OPB2FSI:
>>>> +        fsi_opb_opb2fsi_address(&s->opb[0], data & 
>>>> APB2OPB_OPB2FSI_OFF);
>>>
>>>
>>> same for fsi_opb_opb2fsi_address()
>> Same as above.
>>>
>>>> +        break;
>>>> +    case APB2OPB_OPB0_WRITE_WORD_ENDIAN:
>>>> +        if (data != APB2OPB_OPB0_WRITE_WORD_ENDIAN_BE) {
>>>> +            qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                          "%s: Bridge needs to be driven as BE 
>>>> (0x%x)\n",
>>>> +                          __func__, 
>>>> APB2OPB_OPB0_WRITE_WORD_ENDIAN_BE);
>>>> +        }
>>>> +        break;
>>>> +    case APB2OPB_OPB0_WRITE_BYTE_ENDIAN:
>>>> +        if (data != APB2OPB_OPB0_WRITE_BYTE_ENDIAN_BE) {
>>>> +            qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                          "%s: Bridge needs to be driven as BE 
>>>> (0x%x)\n",
>>>> +                          __func__, 
>>>> APB2OPB_OPB0_WRITE_BYTE_ENDIAN_BE);
>>>> +        }
>>>> +        break;
>>>> +    case APB2OPB_OPB0_READ_BYTE_ENDIAN:
>>>> +        if (data != APB2OPB_OPB0_READ_WORD_ENDIAN_BE) {
>>>> +            qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                          "%s: Bridge needs to be driven as BE 
>>>> (0x%x)\n",
>>>> +                          __func__, 
>>>> APB2OPB_OPB0_READ_WORD_ENDIAN_BE);
>>>> +        }
>>>> +        break;
>>>> +    case APB2OPB_TRIGGER:
>>>> +    {
>>>> +        uint32_t opb, op_mode, op_size, op_addr, op_data;
>>>> +
>>>> +        assert((s->regs[APB2OPB_OPB0_SEL] & APB2OPB_OPB_SEL_EN) ^
>>>> +               (s->regs[APB2OPB_OPB1_SEL] & APB2OPB_OPB_SEL_EN));
>>>> +
>>>> +        if (s->regs[APB2OPB_OPB0_SEL] & APB2OPB_OPB_SEL_EN) {
>>>> +            opb = 0;
>>>> +            op_mode = s->regs[APB2OPB_OPB0_MODE];
>>>> +            op_size = s->regs[APB2OPB_OPB0_XFER];
>>>> +            op_addr = s->regs[APB2OPB_OPB0_ADDR];
>>>> +            op_data = s->regs[APB2OPB_OPB0_WRITE_DATA];
>>>> +        } else if (s->regs[APB2OPB_OPB1_SEL] & APB2OPB_OPB_SEL_EN) {
>>>> +            opb = 1;
>>>> +            op_mode = s->regs[APB2OPB_OPB1_MODE];
>>>> +            op_size = s->regs[APB2OPB_OPB1_XFER];
>>>> +            op_addr = s->regs[APB2OPB_OPB1_ADDR];
>>>> +            op_data = s->regs[APB2OPB_OPB1_WRITE_DATA];
>>>> +        } else {
>>>> +            qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                          "%s: Invalid operation: 0x%"HWADDR_PRIx" 
>>>> for %u\n",
>>>> +                          __func__, addr, size);
>>>> +            return;
>>>> +        }
>>>> +
>>>> +        if (op_size & ~(APB2OPB_OPB_XFER_HALF | 
>>>> APB2OPB_OPB_XFER_FULL)) {
>>>> +            qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                          "OPB transaction failed: Unrecognised 
>>>> access width: %d\n",
>>>
>>> Unrecognized
>> Fixed
>>>
>>>> +                          op_size);
>>>> +            return;
>>>> +        }
>>>> +
>>>> +        op_size += 1;
>>>> +
>>>> +        if (op_mode & APB2OPB_OPB_MODE_RD) {
>>>> +            int index = opb ? APB2OPB_OPB1_READ_DATA
>>>> +                : APB2OPB_OPB0_READ_DATA;
>>>> +
>>>> +            switch (op_size) {
>>>> +            case 1:
>>>> +                s->regs[index] = fsi_opb_read8(&s->opb[opb], 
>>>> op_addr);
>>>> +                break;
>>>> +            case 2:
>>>> +                s->regs[index] = fsi_opb_read16(&s->opb[opb], 
>>>> op_addr);
>>>> +                break;
>>>> +            case 4:
>>>> +                s->regs[index] = fsi_opb_read32(&s->opb[opb], 
>>>> op_addr);
>>>> +                break;
>>>> +            default:
>>>> +                qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                              "%s: Size not supported: %u\n",
>>>> +                              __func__, size);
>>>
>>> this should use op_size and not size and seems redudant with
>>> the unrecognized test above.
>> true, Keeping it in case bits meaning change in future.
>>>
>>>
>>>> +                return;
>>>> +            }
>>>> +        } else {
>>>> +            /* FIXME: Endian swizzling */
>>>> +            switch (op_size) {
>>>> +            case 1:
>>>> +                fsi_opb_write8(&s->opb[opb], op_addr, op_data);
>>>> +                break;
>>>> +            case 2:
>>>> +                fsi_opb_write16(&s->opb[opb], op_addr, op_data);
>>>> +                break;
>>>> +            case 4:
>>>> +                fsi_opb_write32(&s->opb[opb], op_addr, op_data);
>>>> +                break;
>>>> +            default:
>>>> +                qemu_log_mask(LOG_GUEST_ERROR,
>>>> +                              "%s: Size not supported: %u\n",
>>>> +                              __func__, op_size);
>>>> +                return;
>>>> +            }
>>>> +        }
>>>
>>>
>>> The above is equivalent to :
>>>
>>>         MemTxResult result;
>>>         bool is_write = !(op_mode & APB2OPB_OPB_MODE_RD);
>>>         int index = opb ? APB2OPB_OPB1_READ_DATA : 
>>> APB2OPB_OPB0_READ_DATA;
>>>         AddressSpace *as = &s->opb[opb].as;
>>>
>>>         result = address_space_rw(as, op_addr, MEMTXATTRS_UNSPECIFIED,
>>>                                   &op_data, op_size, is_write);
>>>         if (result != MEMTX_OK) {
>>>             qemu_log_mask(LOG_GUEST_ERROR, "%s: OPB %s failed @%08x\n",
>>>                           __func__, is_write ? "write" : "read", 
>>> op_addr);
>>>             return;
>>>         }
>>>
>>>         if (!is_write) {
>>>             s->regs[index] = op_data;
>>>         }
>>>
>>> and the fsi_opb_* routines are useless to me.
>> We are trying to keep the separation between OPB implementation and 
>> interface hence we have all those fsi_opb_*. I feel that we should 
>> keep as it is so that future extensions will be easier. Please let me 
>> know.
>
> Well, I can't really tell because I don't know enough about FSI :/
>
> The models look fragile and I have spent already a lot of time trying
> to untangle what they are trying to do. Please ask your teammates or
> let's see in the next QEMU cycle.

Thanks for helping me to get in this patch.

On chip peripherals are connected to OPB bus for easier access. All FSI 
masters (including cascaded) will be attached to OPB bus but we may 
never implement it. So the current model is only instantiating bus and 
having wrapper around mmio. Are you suggesting to simplify it by not 
having this indirection as its not providing much? If that's the case 
then I can change as per your suggestion.

Andrew, Do you have any thoughts about this?

>
> Thanks,
>
> C.
>
>