mass-storage-gadget: No mass-storage on CM4 / CM4S
nbuchwitz opened this issue · 9 comments
I tried to boot a CM4 (in official IO board) and a CM4S (in official IO board and also in our RevPi), but in both cases I was unable to get a mass storage device on my host system.
After some debugging I noticed that the file adapting dwc2-overlay.dtb is missing in the boot.img. With this file added and alsodtoverlay=dwc2,dr_mode=otg to dtoverlay=dwc2,dr_mode=peripheral in config.txt both devices work as expected.
This hasn't been tested on a Raspberry Pi 4 or 400, but I expect the same behavior.
If the above solution sounds sane to you, I'm happy to provide a PR with things fixed.
CC @timg236 as he implemented mass-storage-gadget in the first place (thanks!)
The expected overlay file name is dwc2.dtbo - which distro and kernel version are you running?
It seems to work for me
dwc-otg.dtbo and dwc2.dtbo are also in the boot.img overlays directory in this file https://github.com/raspberrypi/usbboot/blob/master/mass-storage-gadget/boot.img
I'm not aware of a dwc2-overlay.dtb file. I don't think it's a thing
Never mind the dwc2 thing, looks like a mess-up during debugging. The main things seems to be the dr_mode=peripheral. I just verified with mass-storage-gadget from this repository as following:
CM4 on CM4IO: works
CM4S on CM4IO: doesn't work (boots, but never exposes mass storage)
Kernel log from CM4S:
[...]
[ 0.840817] of_cfs_init
[ 0.840918] of_cfs_init: OK
[ 0.841839] uart-pl011 fe201000.serial: no DMA platform data
[ 0.869056] mmc0: SDHCI controller on fe340000.mmc [fe340000.mmc] using ADMA
[ 0.873596] Freeing unused kernel memory: 2048K
[ 0.873939] Run /init as init process
[ 0.873948] with arguments:
[ 0.873957] /init
[ 0.873965] with environment:
[ 0.873973] HOME=/
[ 0.873981] TERM=linux
[ 1.002905] mmc0: new DDR MMC card at address 0001
[ 1.003761] mmcblk0: mmc0:0001 BJTD4R 29.1 GiB
[ 1.004172] mmcblk0boot0: mmc0:0001 BJTD4R partition 1 4.00 MiB
[ 1.004581] mmcblk0boot1: mmc0:0001 BJTD4R partition 2 4.00 MiB
[ 1.004834] mmcblk0rpmb: mmc0:0001 BJTD4R partition 3 4.00 MiB, chardev (245:0)
[ 1.005717] mmcblk0: p1 p2
[ 1.019822] udevd[111]: starting version 3.2.11
[ 1.020671] random: udevd: uninitialized urandom read (16 bytes read)
[ 1.020877] random: udevd: uninitialized urandom read (16 bytes read)
[ 1.020974] random: udevd: uninitialized urandom read (16 bytes read)
[ 1.032460] udevd[112]: starting eudev-3.2.11
[ 1.157664] vc_sm_cma: module is from the staging directory, the quality is unknown, you have been warned.
[ 1.158999] bcm2835_vc_sm_cma_probe: Videocore shared memory driver
[ 1.159021] [vc_sm_connected_init]: start
[ 1.160810] [vc_sm_connected_init]: installed successfully
[ 1.189546] dwc2 fe980000.usb: supply vusb_d not found, using dummy regulator
[ 1.189772] dwc2 fe980000.usb: supply vusb_a not found, using dummy regulator
[ 1.407069] dwc2 fe980000.usb: DWC OTG Controller
[ 1.407109] dwc2 fe980000.usb: new USB bus registered, assigned bus number 1
[ 1.407152] dwc2 fe980000.usb: irq 37, io mem 0xfe980000
[ 1.407405] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002, bcdDevice= 5.10
[ 1.407420] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 1.407433] usb usb1: Product: DWC OTG Controller
[ 1.407444] usb usb1: Manufacturer: Linux 5.10.95-v7l dwc2_hsotg
[ 1.407455] usb usb1: SerialNumber: fe980000.usb
[ 1.408114] hub 1-0:1.0: USB hub found
[ 1.408164] hub 1-0:1.0: 1 port detected
[ 1.579004] Mass Storage Function, version: 2009/09/11
[ 1.579023] LUN: removable file: (no medium)
[ 31.196379] cam-dummy-reg: disabling
[ 206.956423] random: crng init done
[ 206.956436] random: 3 urandom warning(s) missed due to ratelimiting
My test systems where a Raspberry 3B+ running latest Buster image (5.10.120) and my workstation running Debian Bullseye (kernel 5.10.127). If I modify the config.txt it works on CM4 and CM4S as expected.
Hmm maybe cm4s defaults to host mode
Would make sense, as it is the only USB. Is there anything which speaks against having dr_mode=peripheral as the default?
I think that would be ok
Thanks for confirming and the debugging :)