USB Cleans Up Its Terminology, Continued USB4 Work For Linux 5.9

[starshipeleven]

They hand out the white privilege card once you pay your first $250,000/yr dues at the country club.

But on the other post....Gadget might be what it's called, but unlike how allowlist/denylist where changing the terms from whitelist/blacklist makes the words more clear on what they do, it isn't readily clear what role Gadget will play until one reads a tech manual.

I can agree that "gadget" is a bullshit name, but it's completely tangential here.

To actually set up and use this USB mode in your device you need to read the API documentation, so you must know what "USB gadget" means anyway because that's the API name. Before this change it was even more bullshit because the API name differs from the kernel feature name. Now at least they have the same name so you can at least find the docs faster I guess.

For the sake of additional clarity: USB gadget is not something that is enabled/used by default, and to use it you must know how to configure it and it's a bit more involved than just flipping a config switch. So assuming whoever is using this has RTFM is fair, since it has no use otherwise.

USB ports are initialized by default in host mode (the "slave owner that is whipping his slaves" mode, you know, that mode)

[zyxxel]

They could have just stuck to:
USB 3.0 (5 Gb/s)
USB 3.1 (10 Gb/s)
USB 3.2 (2x10 Gb/s)
USB 4.0 (2x20 Gb/s)

Instead of renaming everything with each new version. But now you can have a USB 3.2 port with only 3.0 speeds. We never got devices with "USB 2.0 low speed" ports.

There is not just the ones you mentioned. There is now:

Gen 1x1
Gen 1X2
Gen 1x3
Gen 2x1
Gen 2X2
Gen 2x3

With 1X2 and 2x1 having the same speed.

There were USB 2 low speed ports - the command set was USB 2.0 but it as 12 Mbit/s instead of 480 Mbit/s. An example was some Freescale PPC processors.

[LinAGKar]

There were USB 2 low speed ports - the command set was USB 2.0 but it as 12 Mbit/s instead of 480 Mbit/s. An example was some Freescale PPC processors.

Were there any marketed as USB 2.0 devices?

[zyxxel]

Were there any marketed as USB 2.0 devices?

Yes. The NXP MPC875. Somewhere down the line, they sneaked in the word "compatible" and nowadays writes "USB 2.0 full-/low-speed compatible". When originally released, they were more aggressive in marketing them as USB 2.0 devices and then specified that the device supported full-/low speed transfer rates.

Change note of datasheet for release 0.8 "Added the reference to USB 2.0 to the Features list and removed 1.1 from USB on the block diagrams."

Change note of datasheet for release 0.9 "Changed the USB description to full-/low-speed compatible."

The big players where already implementing their hardware when these datasheet versions where released. And the USB 2.0 information flowed all the way out to released end-user products.

Not too many companies actually managed to get the USB to work at all with Linux - they skipped some of the hardware and relied on the software driver to feed the hardware while keeping track of how many more bytes that could be sent before next SOF (and figure out how long the interrupt latency was, i.e. how much of the 1ms transfer window that was already lost). So they did lots of trickery with the USB in this line of processors.

But in the end - there were products marketed as USB 2.0 that only managed 12 Mbit/s, and where the claim to fame for marketing them as USB 2.0 was that the command set was USB 2.0.

[LinAGKar]

Yes. The NXP MPC875. Somewhere down the line, they sneaked in the word "compatible" and nowadays writes "USB 2.0 full-/low-speed compatible". When originally released, they were more aggressive in marketing them as USB 2.0 devices and then specified that the device supported full-/low speed transfer rates.

I see, interesting. USB 2.0 full-speed compatible sounds pretty scummy.

[zyxxel]

I see, interesting. USB 2.0 full-speed compatible sounds pretty scummy.

Since I did get access to early silicon and preliminary documentation - and the information from their application engineers - I hated them so very much when the full implications of their USB 2.0 support became obvious. But then it was too late - the product was already designed and almost ready for production. But I did see a number of other products with the same chip - and with the original Freescale information inserted in product data sheets.