feature(dcd_dwc2) : Added DMA cache synchronization on esp32p4

[roma-jam]

Requirements

1. To enable Buffer DMA mode, esp-tinyusb changes should be done (define CFG_TUD_DWC2_DMA and implement CFG_TUSB_MEM_SECTION), TBD: PR_LINK
2. Support Buffer DMA mode for ESP32S2/S3 (no sync required, so this could be achieved with step 1.) and ESP32P4 (sync required, changes in this PR is crucial for Buffer DMA on P4) chips.

Measurements

Measurements were made on tusb_msc example (read-write operations benchmark, SDMMC target).

ESP32S3

MSC Buffer size (bytes)	Slave Mode (R/W)	Buffer DMA (R/W)
512	463,5 kB/s / 161,1 kB/s	512,6 kB/s / 171,2 kB/s
4096	733,5 kB/s / 452,9 kB/s	878,8 kB/s / 469,9 kB/s
8192	N/A / N/A	N/A / N/A

ESP32P4

MSC Buffer size (bytes)	Slave Mode (R/W)	Buffer DMA (R/W)
512	909,6 kB/s / 217,8 kB/s	942,8 kB/s / 222,6 kB/s
4096	932,7 kB/s / 225,2 kB/s	3,8 MB/s / 1,1 MB/s
8192*	934,4 KB/s / 229,4 KB/s	5,3 MB/s / 1,8 MB/s

*Benchmark results provided

Limitations

Not working:

• tusb_ncm (S2/S3, TinyUSB stack overflow during recv_transfer_datagram_to_glue_logic())

Working examples:

• tusb_composite_msc_serial
• tusb_console
• tusb_hid
• tusb_midi
• tusb_msc
• tusb_serial_device (verified tx with len = [1..512])

Breaking change

No breaking changes

Checklist

• Pull Request name has appropriate format (for example: "fix(dcd_dwc2): Resolved address selection when several phy are present")
• Optional: README.md updated
• CI passing

Related issues

No related issues