USB packet loss.

[Steve-o]

People say really silly things like 'I read that the Raspberry Pi would make an ideal hi-fi media [...]

I don't get why you are getting mad at him. He posted his interest in getting usb audio fixed but nowhere in his post did he demand an instant fix. I am planning to use the rPi in the same way but I understand that fixing those problems takes time.

Also no one of us bought the Pi because someone told us it could do this or that. I am studying electrical engineering and have been using linux since I bought my first computer. There is no reason why the rPi should not be able to do anything another similar linux box can't. I think it should be in everybody's intererest that those compatibility issues are found and addressed.

So calm down and be happy that the Pi is already working with whatever you are using it for

[obcd]

You articulated my thoughts very accurately

[jamesh]

People say really silly things like 'I read that the Raspberry Pi would make an ideal hi-fi media [...]

I don't get why you are getting mad at him. He posted his interest in getting usb audio fixed but nowhere in his post did he demand an instant fix. I am planning to use the rPi in the same way but I understand that fixing those problems takes time.

Also no one of us bought the Pi because someone told us it could do this or that. I am studying electrical engineering and have been using linux since I bought my first computer. There is no reason why the rPi should not be able to do anything another similar linux box can't. I think it should be in everybody's intererest that those compatibility issues are found and addressed.

So calm down and be happy that the Pi is already working with whatever you are using it for

Please remember that this issue is being actively worked on. It's not like it's hidden under the carpet in the hope it goes away.

[fredjam]

Yes it is being worked on and progress is being made and I'm perfectly calm. However I'm
also amazed at the general air of ingratitude that comes with some posts and the absurd
expectations of some people who bought the pi. I suppose press hype is responsible to
some extent., that and peoples' gullibility. There is a reason why the pi costs $35 and a quad
core Intel processor costs $350 or more. If you want to play your 24bit flac files on a pi then
be prepared for a long wait or learn to do it yourself. The pi already plays 16 bit flac files
perfectly and extensive listening tests show that people who claim they can hear the
difference between 16bits and 24bits can't hear the difference.
..

[naplam]

There is a reason why the pi costs $35 and a quad
core Intel processor costs $350 or more...

People's expectations are absolutely not absurd. The reason de Pi costs $35 is because it's non-profit. Not because it's kind of expected to be crippled. It should have worked perfectly, if it was expected not to work properly nobody at the foundation would be trying to fix the issues it has. So please guys stop once and for all that argument that we should be happy with a somewhat crippled device just because it's cheap. I'm grateful to the foundation for their work and it's ok, they don't owe me what the pi lacks but it's certainly lacking and I hope we all work together to fix the issues there are, instead of letting feelings get in the way and lead to these ridiculous irrational discussions.

[hippy]

It should have worked perfectly, if it was expected not to work properly nobody at the foundation would be trying to fix the issues it has.

Not at all true. The R-Pi is a developer board and the Foundation offers no guarantee of any particular fitness of purpose or any claim it would work properly or at all in every case.

That doesn't mean the Foundation and others won't strive to make it as perfect as it could be but no one offers any absolute guarantee or promise that the developer boards will be.

Some expectations are wrong set against what the Foundation has in the current R-Pi product and what expectations that should have created. I've commented on that disconnect in other posts and won't repeat those here.

[naplam]

It should have worked perfectly, if it was expected not to work properly nobody at the foundation would be trying to fix the issues it has.

Not at all true. The R-Pi is a developer board and the Foundation offers no guarantee of any particular fitness of purpose or any claim it would work properly or at all in every case.

That doesn't mean the Foundation and others won't strive to make it as perfect as it could be but no one offers any absolute guarantee or promise that the developer boards will be.

Some expectations are wrong set against what the Foundation has in the current R-Pi product and what expectations that should have created. I've commented on that disconnect in other posts and won't repeat those here.

Now that's a disconnect with reality. You're basically telling people they should be happy it even boots. That's completely nuts.

[Burngate]

It boots. It does a lot of things. Some things it won't do. Some of the things it won't do are things that some people wanted it to do, and some of those people expected it to do things it won't.

People's happiness depends on many things, including but not limited to how well reality meets their expectations. Their happiness is not my responsibility.

The expectations of some people did not match reality, in part because of their limited understanding of other parts of reality, such as what other members of the implicit set "Computer" do. As in, analogeously, the implicit set "car" - a F1 car is a member of "car" and does 200mph, and my Lada is a member of "car" so I expect it to do 200mph

Should a Lada owner be upset? Expand.

[naplam]

No, I'm not going to expand. I made my point very clear. If you don't want to understand, that's your problem. USB working properly is not an unreasonable expectation, the SD driver not slowing down the pi to a halt is not an unreasonable expectation, and so on.

But please let's stop polluting this thread with this nonsense (and outright trolling) and let's allow anyone who wants to report issues or help fix them work together.

[Burngate]

OK, so as to be not-troll I'll expand.

The Lada owner knows from other parts of his life that his car is different from a F1 car, but a Pi owner may not be aware of how complicated the USB system is, or the very limited power of the Pi compared to his Windows box.

So his experience of his windows box might lead him to believe USB just works and be disappointed. Or he might take it as a lesson in what he gets for what he pays. Those will colour his approach to the Pi, and therefore his posting here.

[hippy]

Now that's a disconnect with reality. You're basically telling people they should be happy it even boots. That's completely nuts.

Well that's pretty much how it is as I see it. The Foundation delivered a developer board and never guaranteed anything more than that.

We should be happy it even boots, should be happy that it does an awful lot more even if it doesn't do everything. The Foundation has not delivered anything less than promised.

USB working properly is not an unreasonable expectation

Not for a finished product perhaps but that's not the R-Pi as it is now; it's a developer board.

It's specifically a developer board because it was anticipated that it might not do all that some may expect in all cases. It's saying, don't have those expectations. That may not have been understood by all purchasers and IMO has not always been explicitly or clearly explained by the sellers as I've previously said.

[thexman]

Let me just make a couple of points as its my thread

1. I fully understand USB v1 and v1.1 also v2.0 having developed devices for a number of years all these devices have worked under windows 98se and upwards including red hat Linux mac os windows CE several other versions of Linux all on systems from a pentium 90 right up to the i7 processors and boards

2. The rpi implementation of USB has left a lot to be desired for a standard connection on the board lets face it to connect any thing to this development board requires either USB or sd card or gpio which other people developed quite successfully

3. Having developed my software under Linux I wanted to use the very cheap Pi as I didn't need a windows box and a 3ghz quad core processor I was more than happy with 700mhz and 256mb of memory to play with the hdmi was a bonus as was the ip connection and sound I won't be using sound my only requirement was USB and a processor capable of running linux

Ive said it before this is the first board of its type but it's also the first board to fail to deliver standard USB traffic transfer they just completely missed the mark there now trying to make amends and I believe there trying hard ill keep waiting

[hippy]

there now trying to make amends and I believe there trying hard ill keep waiting

Don't forget that, while it's a developer board now, the intent is to produce a finished product, get an educational release out there, so the Foundation plus the community will be working towards that goal.

Ultimately the R-Pi should be no more problematic than any other PC out there, the developer status simply recognised it might not be quite there yet.

[fredjam]

People talk about USB working properly but nobody is prepared to say what this means.
How many things can you hang off one usb hub on a low speed computer like the pi.
Can you play 24 bit flac files from a usb HDD in 7.1 surround sound through your external
sound card. Is that a reasonable expectation? Even if it is, is Alsa up to the job of playing
these files in real time without interruption.

Like other people here I have used Linux computers for many years but most of them
were more powerful than a pi and had multiple usb hubs. For high quality sound we
used the jack sound server with its configurable latency and buffers. Of course these
computers also had much more memory than the pi. People who post saying Linux can
do anything forget that a Turing machine can do anything that any other computer can
do but it may do it too slowly to be of practical use.

Don't imagine that when the Usb driver is made as stable as it can be that this will make
the pi like a computer with an Intel Ivybridge processor running the latest version of Ubuntu,
it won't.

[yoctopuce]

People talk about USB working properly but nobody is prepared to say what this means. How many things can you hang off one usb hub on a low speed computer like the pi.

Look back at the beginning of this thread: a single USB 1.1 HID device can already cause the Pi to drop packets. We have sent to the foundation a trivial device that just sends numbered packets and reproduces the issue consistently.

Don't imagine that when the Usb driver is made as stable as it can be that this will make the pi like a computer with an Intel Ivybridge processor running the latest version of Ubuntu,
it won't.

We don't expect it to be fast. We expect it to follow the specs and not drop packets that have been acknowledged on the USB bus as transmitted. That will do.

[gsh]

Actually I'd just like to mention it isn't the Pi dropping the packets it's the hub, mostly because we don't respond to the hub in time... Whoever thought that was a good design decision for a hub should be shot!

@Yoctopuce, have you tried again recently with the fixes to latency? I now have trouble reproducing the error and have to flood ping my box to make it happen! (Unfortunately it still happens!)

Gordon

[obcd]

If the problem is the hub dropping the packets, than why does it still occur on the model A Pi?
And what about the FTDI or prolific devices sending those huge number of NAK's?
Isn't an usb microframe 125 uS? So, why do we get NAK's every 16 - 20 uS?

[Burngate]

Is the $35 Pi defective? Or is the free software defective?

[jamesh]

If the problem is the hub dropping the packets, than why does it still occur on the model A Pi?
And what about the FTDI or prolific devices sending those huge number of NAK's?
Isn't an usb microframe 125 uS? So, why do we get NAK's every 16 - 20 uS?

Has someone got a Model A to test that?

The excessive NAK's are the fault/bad design of the device I believe - fast desktops can cope with the demand, slower devices have more trouble would be my guess.

As has been acknowledged, there is a problem in the Raspi USB software, but it is *exacerbated* by badly designed devices and hubs to the point where the Raspi will work completely normally, until attached to these devices/hubs. At least, that is my understanding.

[naplam]

Is the $35 Pi defective? Or is the free software defective?

[Moderator
Removed entire post - please calm down Napalm, and stop swearing. You have your opinion, please don't keep spouting it off here, along with the bad language. I will keep deleting your posts if they continue in the same vein. We are aware of the USB problems.
]

[obcd]

Gordon has a model A, and another member on this forum removed the usb hub / lan switch from his model B board, which basically makes it a model A with a bunch of unused connectors.

We all agree the Pi misses usb packets (at least I hope so).
It could be the hub dropping them?
What should a hub do with an usb packet that isn't handled if another one arrives?

I have a very old creative webcam (from the time computers were like 200Mhz),
and the Pi can't produce a proper picture with it. (it's very low resolution)
It looks like the cpu speed is used as an excuse for what we are seeing.
I have seen no solid prove so far for it, and no response either upon the questions where I could follow what's being worked on.

If there are limits what can be used and what can't (at the moment), that's understandable.
The issue that's bugging me is the fact there is no clear line what is possible.
Even your own distributor Element 14 is selling usb hub's as recommended items for the Pi that don't work and backfeed power to the upstream usb port.
They are selling usb2ps2 adapters that lot of people are having trouble with...
o it's also understandable people become dissapointed about all this.

If the idea of releasing the Pi as a development board is to find and fix issues, than at least the foundation should play open card with it's (beta) testers.

[naplam]

[Moderator

This post has also been removed. Napalm, one of the forum rules is that you do NOT post PM's in public. You would do well to re-read the forum rules before posting again. Last warning]

[jamesh]

Gordon has a model A, and another member on this forum removed the usb hub / lan switch from his model B board, which basically makes it a model A with a bunch of unused connectors.

We all agree the Pi misses usb packets (at least I hope so).
It could be the hub dropping them?
What should a hub do with an usb packet that isn't handled if another one arrives?

I have a very old creative webcam (from the time computers were like 200Mhz),
and the Pi can't produce a proper picture with it. (it's very low resolution)
It looks like the cpu speed is used as an excuse for what we are seeing.
I have seen no solid prove so far for it, and no response either upon the questions where I could follow what's being worked on.

If there are limits what can be used and what can't (at the moment), that's understandable.
The issue that's bugging me is the fact there is no clear line what is possible.
Even your own distributor Element 14 is selling usb hub's as recommended items for the Pi that don't work and backfeed power to the upstream usb port.
They are selling usb2ps2 adapters that lot of people are having trouble with...
o it's also understandable people become dissapointed about all this.

If the idea of releasing the Pi as a development board is to find and fix issues, than at least the foundation should play open card with it's (beta) testers.

Open Card - for goodness sake, HOW MUCH MORE OPEN CAN WE BE! The USB problem is acknowledged and being worked on. The Foundation has always said prior to the educational release that people are getting a development board.

As for limits, no-one knows what they are. Its impossible to test every USB device out there so how can the Foundation possible tell you what will and won't work? Also, if Farnell are selling equipment that doesn't work, then that is Farnell's issue - they are the ones selling them!

Just FYI, my hub works fine, all the USB devices I use work fine.

[naplam]

[Deleted again - no bad language, but continual concern trolling.]

[MaxK1]

Personally, I'm quite happy with both of mine.

What isn't entirely clear to me: Is the USB sub-system (hardware) *KNOWN* defective or is the driver
"just" a gut-wrenching SOB to debug? It's clear the S/W isn't fully there yet, but I don't think the foundation is to blame for arch/raspbian/riscos/whatever not working the way some people seem to think it should.