USB pass through into qemu-system-ppc running Mac OS 9.2

[Howard Spoelstra]

Hi Gerd,

(Inadvertently I sent an earlier copy of this message without subject line...)
https://lists.nongnu.org/archive/html/qemu-devel/2021-03/msg02639.html
As gmail will not accept zip in zip, I also now attach two zip files with logs.

I'd like to report an issue with USB pass through of a Roland MT880
midi device into qemu-system-ppc running Mac OS 9.2. Based on an
earlier conversation in which we found that passing through e.g. a
real USB 2.0 memory stick into Mac OS 9.2 can work when connected
through a real USB 1.1 hub, attempts to pass through a real device
were done via such a hub in a old Apple keyboard.

1. Passing through a Roland MT880 midi device into Mac OS 9.2
The device can work as "original" when reporting device type 0x0014
using its own driver, or use the standard USB classes when reporting
device type 0x0015. "Original" was tested as this would provide the
desired functionality. Logging was created by connecting the device to
a real G4 Cube and by passing through into Qemu. Qemu host was macOS
Catalina.

Logs are attached in the zipped folder "Roland Logs" which contains
logs for the G4 Cube case and the Qemu case. Both sets of logs are
numbered correspondingly so they can be compared "easily" between
cases. All Cube logs are from inside MacOS 9.2

Some things I noticed in the logs:

1: In Qemu log: 1.Qemu-9.2_usb1_log_from_system_profiler.txt
10.221: [   75577309]: Fatal Error: ModemDriver: Couldn't start read
polling, -6979, 0

2: In Qemu log: 2.Qemu-9.2_usb1_0x0014-pcap.pcap
Frames 36 and 38 report a Remote I/O error (-EREMOTEIO) (-121)

3: In Qemu log 3.Qemu-9.2_usb1_0x0014-trace.log I see the last mention
of port 3.1 ep 1 in line 22222 as:
usb_packet_state_change bus 0, port 3.1, ep 1, packet 0x7fd0440371f8,
state setup -> complete
usb_ohci_td_packet_status status=-3
usb_ohci_td_stall

4: In Cube log: 5.Name Registry Output.txt a slight difference in the
drvrOut name:
drvrOut
0000: 0f 2e 55 53 42 33 42 55 4d 2d 38 38 30 4f 75 74 ..USB3BUM-880Out
In Qemu log: 5.Name Registry Output.txt
drvrOut
0000: 0f 2e 55 53 42 33 38 55 4d 2d 38 38 30 4f 75 74 ..USB38UM-880Out

These logs also show a difference in the number of UM880 devices found
in the name registry (2 vs 4).

5: In Qemu log 6.OHCI list dump.txt I see 66 counts of "halted!" while
this does not show up the corresponding Cube log.

6: In Qemu log 7.USB Bus Devices.txt, line 321 I see the status
"Endpoint 1 (Bulk): Halted", while the corresponding Cube log shows
"idle" in line 508.

I hope this provides sufficient information for an analysis (and
perhaps solution?) of this issue.

Best regards,
Howard

Roland Logs.zip
Description: Zip archive

3.Qemu-9.2_usb1_0x0014-trace.log.zip
Description: Zip archive