Can't get MSC to work (especially writing)

[WilheJo]

I try to directly write to an external flash (16MB).
All runs on a CH32V307 using the HS USB peripheral.
Everything is compiled with -Ofast.

The final code uses littlefs on the flash and allows the device to run in 2 modes:
Either present the flash contents to the pc (config change, store update-file,...), or run the actual application.
The read/write access using littlefs already works - but the flash access isn't really the problem, though.

After modifying and embedding the MSC example, I can read from flash:

LC_ALL=C dd if=/dev/sdb bs=512 of=/tmp/foo count=32768 oflag=sync status=progress
16237056 bytes (16 MB, 15 MiB) copied, 12 s, 1.3 MB/s
32768+0 records in
32768+0 records out
16777216 bytes (17 MB, 16 MiB) copied, 12.3488 s, 1.4 MB/s

"Writing" (ie. just return bufsize; in tud_msc_write10_cb in order to simulate the transfer) also works:

LC_ALL=C dd if=/tmp/foo of=/dev/sdb bs=512 count=32768 oflag=sync status=progress
16319488 bytes (16 MB, 16 MiB) copied, 29 s, 563 kB/s 
32768+0 records in
32768+0 records out
16777216 bytes (17 MB, 16 MiB) copied, 29.8062 s, 563 kB/s

The next step is adding some "busyness":

int32_t tud_msc_write10_cb(uint8_t lun, uint32_t lba, uint32_t offset, uint8_t* buffer, uint32_t bufsize)
{
  (void) lun;
  static int counter=0;
  counter++;
  if (counter&1)
	  return 0;
return bufsize;
}

This does not work at all!

LC_ALL=C dd if=/tmp/foo of=/dev/sdb bs=512 count=32768 oflag=sync status=progress
dd: error writing '/dev/sdb': Input/output error
1+0 records in
0+0 records out
0 bytes copied, 218.107 s, 0.0 kB/s

dmesg report:

[15924.412856] scsi 7:0:0:0: Direct-Access     Powermet Mass Storage     1.0  PQ: 0 ANSI: 2
[15924.413352] sd 7:0:0:0: Attached scsi generic sg1 type 0
[15924.413704] sd 7:0:0:0: [sdb] 4096 4096-byte logical blocks: (16.8 MB/16.0 MiB)
[15924.414015] sd 7:0:0:0: [sdb] Write Protect is off
[15924.414022] sd 7:0:0:0: [sdb] Mode Sense: 03 00 00 00
[15924.414322] sd 7:0:0:0: [sdb] No Caching mode page found
[15924.414327] sd 7:0:0:0: [sdb] Assuming drive cache: write through
[15924.419287] sd 7:0:0:0: [sdb] Attached SCSI removable disk
[15959.559399] usb 1-12: reset high-speed USB device number 19 using xhci_hcd
[15959.700300] xhci_hcd 0000:00:14.0: Canceled URB td not found on endpoint ring
[15990.286540] usb 1-12: reset high-speed USB device number 19 using xhci_hcd
[16027.827008] usb 1-12: reset high-speed USB device number 19 using xhci_hcd
[16065.370826] usb 1-12: reset high-speed USB device number 19 using xhci_hcd

What am I doing wrong here?
If I get the code right, I'm supposed to return 0, in case the storage is busy.

On my way to this minimum code snipped, I found, that tud_task does not exit until you return something other than 0.
So the storage "task" needs to execute within tud_msc_write10_cb, right?

[zx96]

@WilheJo I was having similar issues - high speed MSC device on CH32V307 reading just fine, but failing to complete writes. It seems that tud_msc_write10_cb simply wasn't getting called again when it should have been.

I'm not sure what the underlying cause is, but I noticed #2392 and found that cherry-picking the commits from that MR seems to make things work as intended for me; it's worth giving that a try!

[WilheJo]

Thanks!

Cherry-picking both commits made it work.