host/dwc2: fix bitfields access width

[HiFiPhile]

Describe the PR
Fix gcc-riscv issue about volatile bitfields are incorrectly accessed using an access width not appropriate to the type of its container.

Tested on espressif_s3_devkitm with device_info and cdc_msc_freertos examples.

For the host I've tested an usb key and an FTDI usb-uart.

Device 1: ID 0403:6001 SN FT1ZK9MQ
Device Descriptor:
  bLength             18
  bDescriptorType     1
  bcdUSB              0200
  bDeviceClass        0
  bDeviceSubClass     0
  bDeviceProtocol     0
  bMaxPacketSize0     8
  idVendor            0x0403
  idProduct           0x6001
  bcdDevice           0600
  iManufacturer       1     FTDI
  iProduct            2     TTL232R-3V3
  iSerialNumber       3     FT1ZK9MQ
  bNumConfigurations  1

Device 2: ID 0011:7788 SN C4D197AC
Device Descriptor:
  bLength             18
  bDescriptorType     1
  bcdUSB              0200
  bDeviceClass        0
  bDeviceSubClass     0
  bDeviceProtocol     0
  bMaxPacketSize0     64
  idVendor            0x0011
  idProduct           0x7788
  bcdDevice           0104
  iManufacturer       1     Generic
  iProduct            2     Mass Storage
  iSerialNumber       3     C4D197AC
  bNumConfigurations  1

Disconnection detection needs #2960 to work otherwise device index will go up 1,2,3,4...

Fix #3047

[hathach]

thank you, I am thinking, maybe we just remove the struct/bitfield in the register typedef i.e leaving only uint32_t. bitfield struct can be use with local variale only. Will try that when I got sometime,

[HiFiPhile]

I am thinking, maybe we just remove the struct/bitfield in the register typedef i.e leaving only uint32_t. bitfield struct can be use with local variable only

Yes we can do let this. Adding a macro makes the thing complicate and losing the interest of bitfields.

[hathach]

I am thinking, maybe we just remove the struct/bitfield in the register typedef i.e leaving only uint32_t. bitfield struct can be use with local variable only

Yes we can do let this. Adding a macro makes the thing complicate and losing the interest of bitfields.

sure, I will do this later if you are busy. I need to fix an racing issue with pio-usb when overclocking rp2350 to 264mhz first :)

[HiFiPhile]

I've updated the refactor.

[hathach]

perfect, thank you. I couldn't done it any better. I change dwc2_ep/channel_count() to inline function to avoid forcing gnu standard. The rest is mostly clean up