Pinout for GPIO connectors

I’ve winkled the details of the Raspberry Pi’s GPIO pinout out of the hardware guys: no pretty pictures in this post, but a lovely information dump that those of you who solder for fun will enjoy getting your teeth into.

Top row, left to right:


Bottom row, left to right:



– all the UART, SPI and I2C pins can be reconfigured as GPIO if needed.
– some of the do-not-connect (DNC) pins will likely be replaced by
GPIO in a subsequent board revision.

We will not be making the boards with pins in place, in order to shave some cost off the build; you’ll have to solder them on yourself. I have to add the usual legal warning here that this pinout is provisional, subject to change if we decide something’s wrong or not optimal, and all that jazz. The likelihood of any changes is very small indeed, but do be aware that it’s there if you are using this information to build any add-on hardware before you get your hands on a Raspberry Pi.


DeliciousRaspberryCake avatar

Without pin leads pre-mounted on the board, how does that change the future of expansion boards?

liz avatar

People who want to use expansion boards will need to solder a pin header onto the board to connect it. It’s not hard, and it’s something we actively want to get kids in particular learning to do.

DeliciousRaspberryCake avatar

A subsequent question would be: will Raspberry Pi sell the pin headers? Ordering a single component from Farnell or Mouser would be a joke

Though, I can already predict many ruined Raspis due to inexperienced soldering. Or like, pin heads kept in place by super glue

liz avatar

I’m not sure whether we’ll be stocking them right from the start, but yes, we’ll be stocking them down the line. (Money again: we need a lot of capital to fill the shop with goodies, which we can’t raise until we’ve sold some boards!) Here in the UK there’s Maplin, or Fry’s in the US, where you wouldn’t feel weird ordering a single part, by the way; where do you live?

Richie avatar

There is also Rapid online in the UK for components – Used them many times – great service, I found they had a wider range than Maplins and free postage over £30

AllyR avatar

In the UK: CPC/Farnell will entertain extremely small orders (at a price), and also carry more esoteric parts than Maplin. I use both CPC and Maplin for both business and personal electronics needs.

hamjudo avatar

What are the electrical drive specs for the 8 generic GPIO pins, GPIO0 through GPIO7? How many milliamps can they source and sink?

tntexplosivesltd avatar

I assume not a lot, probably ~20mA

BlackLotus89 avatar

Is this a joke?^^ I wonder how much money can you save by not soldering them from the beginning?Anyway thanks or the layout

shiftyphil avatar

Adding a through-hole soldering step to a board where everything else is surface mount probably does add significantly to the cost, especially when the overall cost is so low.

jacklang avatar

Its not primarily cost saving, but a request from the community.
Some wanted the pins on top of the board, others underneath. Some wanted straight pins, others right angled, and a few want to solder wire direct to the board. The holes are through hole plated.

Zing avatar

Several grand, if you include the cost of placing and soldering, over 10k units. It might only change the unit cost by a dollar, but that’s a dollar they can’t spend on a better power regulator or drilling some mounting holes.

By quibbling over some pins instead of wondering about the DNC pins, I can tell you probably don’t even really care about this. Soldering is nothing, but if they don’t bring out pins then odds are there’s little we can do aftermarket to bring them out from a BGA.

Personally, I’m more curious if those pins aren’t there because they couldn’t route them all the way. Hopefully, if that is the case, maybe the pins get brought out to points elsewhere and we can greenwire them later.

Cookie Monster avatar

This (no pins) is actually the “more better way.” Several other development boards come this way (now). Even those that did not at the start have begun to leave them off in later revisions, and the last reason is cost.

Pins are OK if you are going to connect a flat cable from here to there, semi-permanently. (Assemble it and close it up in the box.) If you are prototyping, pins are actually harder to work with and easier to accidentally short, sockets work better. Also, if you are assembling a device and the thing it plugs into, such as an Arduino and a “shield” where you will mess with it later (not zipped up in a box), it is better to put the *socket* on the electrically “hot” device (Arduino or R-Pi) and put the pins on expansion board. No bare pins with volts on them.

Back to R-PI design. So what do you mount, pins or sockets? Top or bottom? What about the people who (vibration?) do not trust connectors (unless they lock) and so they want soldered in wires? No matter which you choose, you tick off someone. Leave them off and allow freedom of choice.

Chris avatar

Terribly exciting to see I2C connectors! That means you can plug in all sorts of sensor boards – like accelerometers, gyros, magnetometers, etc… This board might be very useful for all sorts of physical projects (eg data logger for model planes, controller for model helicopters, etc).
Cool! Can’t wait

MmmPi avatar

“some of the do-not-connect (DNC) pins will likely be replaced by
GPIO in a subsequent board revision.”
:( Pre-revision boards will then feel second class to me. Have to fight my perfectionist tendencies.

juan avatar

Considering the cost, and the possibilities, I’ll gladly plonk the money for V1.0 board, and find a use for it when I get a V2.0 board.

[…] to that post from RasPi here: Pinout for GPIO connectors Scoreboard ProjectPermalink ← Project […]

electrodacus avatar

Why DNC (do-not-connect) pins, are they connected to something?
Or are just simple NC cot connected pins.

liz avatar

No – they’re DNC, not NC – which is why it says DNC, not NC! They’re connected to a combination of power and ground at the moment (in later revs they’ll be GPIO), and if you connect them you will fry your board, so don’t.

Anthony avatar

Will any of the gpio pins be usable for analog input? And I assume that it is running on 5v logic since there is 5v on the header, right?

jamesh avatar

No ADC on the board. Gert’s expansion board does have ADC’s on so that would be a good option.

Sander avatar

Gert has an expansion board with ADC’s? Handy: I want to measure how my thermostat is talking to my boiler. And after knowing that, I want the RasPi to be a thermostat, that can remotely be checked and controlled.

So, Gert, is the ADS expansion board going to be in the RasPi shop? Alsjeblieft …

Gert avatar

Probably, board comes back next week.
But I was not told about the DNC so I have to do a re-spin :-(

liz avatar

Sorry Gert – I only found out about it myself last night. Go and give Eben a good kick when you get a moment!

Sander avatar


Do you connect the ADC’s via the I2C-connector?
Will the RasPi-distro’s come with a driver for the I2C-interface?

Cookie Monster avatar

No, the IO is 3.3v (not 5v).

JustACat avatar

GPIOs instead of DNC – will be nice!

herebus avatar

This is good news. This opens up the world to other peripherals such as memory expansion (i2c), 1-Wire (Clock) and Zigbee, with out the burden of jumping through the USB hoop. Plus, we get to use the header of the GPIO (after soldering) to mount the board and have access to the 5Volt rail.


Sander avatar

Hi Liz,

Just checking: can you just solder a standard strip like the ones you see on a computer motherboard? If so, that doesn’t sound to difficult.
And then you can put a small daughter board on it, or a connector with a cable to a daughter board?

Geraint Rowlands avatar

That’s the idea, yes. (Although it seems some people intend to use different types of connectors.)

liz avatar

Yes, absolutely.

eggn1n3 avatar

This is excellent; both 3V3 and 5V0 pins! Well done!

eggn1n3 avatar

Forget to mention: Good idea not to solder any pins for the GPIO. I might use different pin headers for each of my Raspberri Pi projects anyway.

Is there already something known about the GPIO addresses?

Daniel avatar

Is the position of 5V0/3V3 some kind of standard? Would it be too hard to swap 3V3 vs. the first DNC in the top row? (Its better to use DNC/SDA0/SCL0 instead of 5V0/3V3/GND than currently 5V0/… when you want 3V3/… – not only kids get this kind of things wrong)

[…] Pinout for GPIO connectors | Raspberry Pi We've just published the final GPIO pinouts for the Raspberry Pi – see for details. Source: […]

Lawrence avatar

Are they on a 0.1″ spacing?

I can see the first useful cable would be a breakout to a plugin breadboard.

jacklang avatar

Yes 2.54mm or 0.1inch spacing

Bakul avatar

Thanks for the GPIO pinout info!

Will the JTAG header be connected?

Sancho avatar

I assumed that’s what tne RXD TXD for.

morpheus avatar

Sorry for noob question.
What is GPIO? It’s means i can use RasPi as PIC?

hamster avatar

GPIO = General purpose I/O.

Yes, it can be used as either an input or an output, much like the ports on the PICs.

And much like on PICs, although some of these pins are also connected to on-chip interfaces such as a UART they can also be configured to act as GPIOs too.

Stefan avatar

Loving this news – happy fuzzy feelings! May I ask – is the left-to-right given looking at the top side of the board or the bottom (top being where the SoC is, in my mind). Also, would I be correct in saying that “top row, bottom row” translates as “row closest the edge of the board, row closest the middle?”

Looking forward to shipping day.


liz avatar

Yes, the top side. The orientation is the same as it’s been in all the pictures of the board so far, if you go and have a look at the schematic pics and the diagrams on earlier posts.

Stefan avatar

Thanks Liz – I thought so, but I’d rather ask and look silly than get myself confused!

liz avatar

No probs – as soon as I read your post it occurred to me that for some, the natural place to look for them would be on the back!

Jan avatar

I’d like to see multiple power pins (Vcc, GND) on the header, ideally one pair per interface on the DNC pins but if there will be more GPIO instead it’s OK :)

YT2095 avatar

I have a few ideas for DNC pins, an A0 and A1 would be a neat idea, effectively giving you 4×8 GPIOs, 2 can be used for a 16 bit Address , one for Data and the other for control line, making EPROM programming easy, or even the ability to add a second processor etc…
even if they were just settable strobes, LE and OE could be used with simple buffer chips like the 74244/245 or the 74374.
perhaps an Interval timer would be handy as well, and or an A to D / D to A converter.

Just a thought ;)

jamesh avatar

Note to all…as I understand it….

DNC – means Do Not Connect. That means they are not connected to anything. It also means that to connect them to anything requires the board to be redesigned. That would add at least 6-8 weeks before you can have you Pi.

So be careful what you ask for.

We might be able to enable more GPIO pins in a later revision, or we might not. At the moment the board is the board – it’s not changing in the near future.

Burngate avatar

Just a thought from what’s been said before … that they’re connected but not activated, rather than not connected.
I’d have thought if they weren’t connected they’d be called NC rather than DNC.
If you’re wrong & I’m right, they could be activated in the next revision of the blob, which means a first-10k board could be updated.
As I say, just a thought!

liz avatar

Yeah – Burngate’s right. They’re connected to power and ground in a potentially board-frying fashion – which is what the D in DNC is all about!

Burngate avatar
Stefan Hennig avatar

Are there any other tricks this pony can do? Like, you know, timers, PWM,…? Or is this supposed to happen in software? How about interrupts?

jacklang avatar

Each GPIO can interrupt, high/low/rise/fall/change

David Driscoll avatar

Do you know if the UART are TTL or Serial Level?

BTW Thanks for all your guys work.

liz avatar

They’re 3V3 TTL.

Ales Hakl avatar

LVTTL or 5V-tolerant LVTTL?

liz avatar

There are lifetime issues with 5V – you might want to add a diode to drop the voltage a bit to 4V3. 5V won’t fry the chip, but it will decrease its lifetime noticeably.

nelson avatar

So… all pins 3V3 or just the UART? Are there 5V tolerant GPIO?

BTW thanks for the great job and help-full support to the community, lets just hope you don’t get problems with idio… less informed people complaining about not being able to run x game on a 25$ computer

liz avatar

They’re all 3V3 – see my reply just above this one for reasons why!

Warringer avatar


Is it possible to supply the 5V directly through the pinheader and not the USB port?

liz avatar

Yes. :)

joep avatar

So, theoretically, there is no reason a smal board couldn’t be created with a regulator that broke out the power from PoE, for example, and powered the RP?

liz avatar

Not at all – get hacking!

Alex Gibson avatar

Cool, I just started building exactly that, a simple 9v battery carrier board with 5v regulator and passthrough for GPIO. I wanted to power the PI only via the 5v GPIO pin – is that AOK Liz?

Alex Gibson avatar

Works :) Quick and dirty TS7805 based circuit. 5v regulator runs hot and battery life is poor, but it’s all solar charged = zero carbon pi for Earth Day :)

Alex Gibson avatar

[url=]Raspberry Pi 3[/url] by [url=]alexgibsondotnet[/url], on Flickr

Rosia Nicolae avatar

What kernel drivers are already written for the SoC ?
Do we have GPIO? SPI? I2C?
Are the drivers in mainline?

Rosia Nicolae avatar

PWM? Tickless kernel? HRTimers?

Geeus avatar

This my seem the stupidest question in the world but here it goes
Are the pads throughhole drilled and plated or will they require drilling and plating before pins can be soldered?

Just working on an expansion shield the the raspi sits in. Considering calling it a Pi Kase with some spi or i2c io extenders and pwm generators.

liz avatar

Yes, they’re through hole drilled and plated to make your soldering life easier. :) The pins are at 2.54mm/0.1inch spacing.

Tom Anderson avatar

Pi Krust!

Bill Cumming avatar

I Don’t trust my soldering skills as far as i can throw the now lump of solder that was my old hub >_<

Might see if I can get a small clamp and build a grid of small solder pads to connect to the GPIO pads.

Might try and get the clamp to double as a holder for the board during testing, if it's not encased in plastic ^_^

Jacques avatar

Use the Google: Pogo Pins

Julian Grammer avatar

I fully appreciate your reservations regarding soldering. “Dem joints, dem joints, dem dry joints…” and all that jazz. I’m thinking screw-in pins and conductive paste (sparingly, obviously).

D Hardman avatar

Thank you for what seems to be an excellent design.

OK there are some warts on the I/O connector but I guess that we shall all have to live with them. With only 6 GPI/O pins ( not counting the reassignment of SPI pins) there is not much chance for an 8 bit parallel I/O. Guess you could do 4 bit nibble with 2 control lines and a strobe. However you can do a lot with SPI particularly if it is 16 bit SPI. Certainly not worth holding up production for a re-layout as I guess that MarkII (if it ever appears, in my experience, probably never – if it works do not fix it!!) will have a significant number of reworks in other areas

Thank you for choosing 2.54mm pitch and for placing them convenient to the edge of the PCB.

I do worry about novice soldering I am sure that there will be some glum faces but leaving the pins off the board is (from my view ) a good choice.

On the question of DNC. I would recommend that you supply a schematic of the pin out with the DNC pins defined in terms of what they are connected to. We are all (hopefully) consenting adults so we should be able to make our own minds up as what to do with them. I, for one, would simply get my trusty pocket DMM out and find out where the pins go just to avoid problems. If you say DO NOT CONNECT then that is fine but at least one should know where they go.

The key question is for an up to-date schematic of the board with a full data (and I mean full!!) set of the chips used on the board.

This board should give a lot of scope for hardware ingenuity and I am sure it will certainly sort out ” … the men from the boys” if you will excuse my sexist remark. Certainly a good number of hardware comments on this forum shows some naivety but there again it is up to the rest of us to help out.

Once again many thanks for an excellent job well done (well done so far……!)


David Hardman

Photon Peddler avatar

> With only 6 GPI/O pins ( not counting the reassignment of SPI pins) there is not much chance for an 8 bit parallel I/O.

Huh? I’m counting full eight GPIO pins, UART, SPI with two chip selects and an I2C bus. You could repurpose one of the SPI chip select pins as an 8-bit bus handshake/strobe, for example.

jamesh avatar

There should be plenty of GPIO – not limited to 6. Also DNC? According to Liz above (I made a mistake earlier) it’s just power and ground but the D really means D.

You won’t get a full datasheet for the SoC. There may be a cut down version.

Bill P. Godfrey avatar

I’m looking forward to being able to solder simple switches and LEDs to the board and controlling them with code.

Imagine the perennial lift controller exercise, but then attach it to an actual lift.

ricky roberson avatar

Does DNC mean “don’t even solder an unused pin here”? If so, using a single header strip is out and that’s a problem. In that case, I would suggest moving (if possible, of course) all non-DNC holes into their own smaller, contiguous rectanguar array pattern that would allow use of a single header strip.

In any event, well done. I think these GPIO holes mean that the primary use of Rpis five years from now is something none of us can yet see – and I predict it will dwarf the installed base of purely educational units. Here’s hoping.

Burngate avatar

DNC means “don’t connect any electronics here”. Pins will be ok, and probably ribbon cable attached to those (capacitance in a long cable might be iffy). If I’m right, the SoC is connected to them, and from what has been said, each connector on the Soc is fed from a multiplexer of some sort which allows the blob to choose what the connector does. At the moment, that’s “nothing”. But it could change.

Will Godfrey avatar

As you have both 3.3V and 5V on the header, I would say the obvious thing to do is have a plug-in daughter board with 5V>3.3V buffer chips on. Much better to fry them with an ‘oops’ that the R-Pi, and they are as cheap as, er, chips :)

Michael avatar

Many thanks for the info!

Please can we have details of the pinout for the JTAG header, and an enumerated list of the connecor names according to the silkscreen?

Also, do all the GPIO pins support PWM, or just some?

liz avatar

The VC JTAG connector will be disabled in OTP, so it’s not very useful for you! Not all of the pins support PWM (and those that do are shared with analogue audio, so if you use them for that you lose analogue audio).

Michael avatar

Are all 15 pins of the connector used for JTAG and disabled in OTP – nothing else useful there, eg GND or voltage pins, extra GPIOs?

liz avatar

Just asked Eben, who says there’s really nothing much useful over there – move along, nothing to see!

Martin avatar

Why disable JTAG D:?

jamesh avatar

Because it’s the JTAG on to the GPU and therefore of no use to anyone (except Broadcom people with the right software and equipment)

Andrew Scheller avatar

So no access to the JTAG of the ARM core then? :-(

Mikkz avatar

How much current can the board source max through the 5V and 3V3 Pin? Are the usb ports rated for 100 or 500mA? thx

Vernier Benoît avatar

How will those connectors be accessible via the software? Do you know where I can find documents explaining how to use these ports in my C software?

Potenzial avatar

More GND and VCC Pins gives a better electrical feeling, remember bad connections at the supply and GND Pins.

I would give 5V and 3,3 V pins at least 2 connections and the GND Pins should be 2 pins at *both* ends of the connector. That summs to 6 pins used, means that 20 pins are left over for usefull stuff.

Maybe the 3,3 V supplys should be avoided. That makes a low-drop regulator mandantory, but gives a much better electrical behaviour.

Jacques avatar

Give us a buildable exercise program for this port, and I’ll have to find something very nice to do for you folks. Thanks for the excellent work.

scologic avatar

So to clarify the TX RX ports does that equate to a RS232 connection? if so which RS232 port would it be under linux?

Sco Logic

Cookie Monster avatar

The TX/RX are 3.3v TTL serial not RS-232. You could use that to talk to a similar device, a 3.3v gps (Trimble Copernicus, GlobalTop 3229, etc.) or other peripheral, or use a TTL serial to RS-232 adapter.

While some clone PCs played with out-of-spec-but-sometimes-worked “Serial Ports”, “Real” RS-232 is +/-5v “out” minimum (could be +/- 9, +/-12, +/-15v), (when designing the input, be aware that open circuit can be up to +/-25v) and for RS-232 the bit sense is reversed. (a “1” goes out as -v, a “0” goes out as +v). If you connect that to an R-Pi (1) it won’t work because of the reverse bit sense and (2) you will blow up the Pi.

scologic avatar

ALso is there access to a USB port from GPIO that way I dont need to use a USB port for an internal component or add on?

jamesh avatar

No I am afraid not.

scologic avatar

but does the tx rx equate to rs232… txrx can also be usb with 5v and 0v..
can anyone clarify?

is the hdmi part of the gpio so we can use headers than the hdmi port..

jamesh avatar

Don’t know about the RS232, but I don’t think the HDMI is pushed out to the GPIO header.

Cookie Monster avatar

The TX/RX is 3.3v TTL serial. It is not RS-232, but it is “like” USB in that it is low voltage, positive only, but that is pretty much all it has in common; USB has a much higher level protocol defined.

You’ve got your hardware and interface specifications all whopperjod. What we’re talking about here is the bare UART only.

Morten avatar

Hi there. Interesting with the pinout published. Since the UART can be reconfigured to GPIO, I assume that it is a different UART than the one used for booting console. But I’d feel better if I ask, so I’ll ask. :)

Is the UART being lead off board a different UART than is being used for boot console?

BTW: There IS a boot console, right?


ricky roberson avatar

Speaking from sad experience here, what is the expected free-space diameter of the GPIO holes after they’ve been plated through? Generally speaking this needs to be at least 0.04 inches (1.02 mm or greater) for the standard non-metric headers to fit. If you make this clearance an even millimeter or less, it’s a tight fit at best and a no-go at worst. You’ve been nice enough to go for tenth-of-an-inch spacing, don’t assume you’re home free yet.

For the record, Digikey stocks 19 matching male pin headers for mail-order delivery that will fit the Rpi GPIO area, the cheapest of which is $1.02 for one and all of which can be purchased in quantity one. To find PDF data sheets with dimensions on all of these (and many other) parts, go the Digikey search page at

and drill down to the part type you’re interested in.

waiter avatar

the good technical news! thanks!

David Riley avatar

Quick question which I don’t see answered above: Is slave SPI supported, or is the SPI port master-only? I imagine that would be tricky on a driver level (I don’t know if Linux supports it at all), but it would be nice to know if it’s available on a hardware level. It would be useful (though not absolutely essential) for my application.

Same question goes for I2C, I guess; it’s a multi-master bus, but does the SoC support slave mode reception? Or is the fastest way to talk directly to this board (as a target) going to be through the UART pins?

– Dave

Willakan avatar

No I2S? Aww…there goes my high-res music streamer!

txf avatar

There ARE other arm boards that are low power with i2s. (some are even cheaper than the Rpi). I would think that just for playing audio the RPi is waaay overpowered; unless you’re doing high bitrate audio there are more efficient ways of making an audio streamer.

jamesh avatar

Could you ping back with which those cheaper boards are – would like to take a look. TIA.

bobdvb avatar

Olimex has a few boards which do I2S but they are more expensive:

Willakan avatar

It would be hi-res: that’s the problem. What other boards would you suggest?

Ferry Hattawidian / Indonesia avatar

Yess!!!… I could connect my Arduino and use it as ADC…
Thanks guys… you have done a great jobs..
I hope I have a chance to buy this magical thing to Indonesia and may be develop something here..

RITRedbeard avatar

Just curious, I realize TTL is 3V3; however is it possible to configure UART as PS/2 (PS/2 at 3V3, but still)?

GCL avatar

And this is one reason out of two thousand why I want one. So when will these little darlings be made available in the US? (And not via Frys.)

jamesh avatar

Few weeks time. Keep an eye on the website for release date, and its from there you buy it too. Worldwide.

Ryan Hunt avatar

I’m a total n00b to this kind of technology, now we have these pinouts does this mean it’s possible to use it as a SATA/high-speed data connector? (To connect hard drives and use the Pi as a file server)

jamesh avatar

No. Although you can use a USB harddrives to give higher storage capacities.

Jared Hudson avatar

I would imagine you could create a SPI to IDE converter like this and use the SPI lines on the Raspberry-Pi to connect it to a hard drive. A USB hard drive connection would probable be faster though.

Fabio Grande avatar

Silly question about GPIO and, actually, about the board…
I know that I will be able to run some kind of Linux distro… I guess I’ll be able to develop some small program with C, isn’t it ?
I’m just wondering if I’ll be able to find some documentation on playing with GPIO from C code…
I’m also wondering if I’ll find docs on how to prepare a SD Card for running Linux/Android on Raspi…

jamesh avatar

The answers to all your questions can be found in the forum. C is fine, there will be docs on all things Raspberry, some from the foundation, some from the wild world.

Sean Dean avatar

Do you have a realise date for the raspberry pi

John McRobert avatar

Will a hardware spec with schematics be provided for those of us who know how to expand on the hardware?

Billy avatar

What about Analogue I/O?

… [Trackback]…

[…] Informations on that Topic: […]…

walt avatar

I’m not sure how fast the SPI can run, but there really should be a signal return path (GND) adjacent to SPI_SCLK. The return path will help maintain impedance and improve signal integrity on SPI_SCLK.

Vincas avatar

Does GPIO has analog input capability?

Sergii Kolisnyk avatar

Can I use SPI NOR or parallel NAND for boot instead of SD/MMC?

JamesH avatar


Sergii Kolisnyk avatar

And what about SD/MMC on GPIO connector?

Sergii Kolisnyk avatar

Studied the datasheet, only single SD interface unfortunately. :(
Waiting for version with POP flash ;)

Brian avatar

Just an example device for anybody wanting true RS232 levels.

TMH avatar

Are there any uptadtes of the pinout?

bernd böttinger avatar

I have a problem with a kernel-Oops (system crash). It happens, when i use the
udevadm in the following command line:
udevadm info –attribute-walk –name ttyUSB0
the problem depends on exploration of usb-devices
i sent the pendig mail to mr. youn from SYNOPSYS, because the corresponding driver was written from synopsys. Ther is no answer in sight. can You help us an delegate the problem to the right person??
tho original mail follows:
Dear Mr. YOUN,
first of all, my english is not very good, but i hope, you will understand what I mean.

We us respberrypi-computers in conjunction with usb-serialport-adapters (FTDI, Prolific) and found out a “Kernel-Oopps”.
You can generate the error condition by executing the following command line as root user at the main console of a raspberrypi computer:

udevadm info –attribute-walk –name ttyUSB0

The problem occurs also at ttyUSB and is makes no matter, if the connected device is an FTDI or Prolific one.
We use this within a script in order to generate logical links to specific adapters, which are connected to special serial bus systems.

We have executed all available updates (apt-get update …apt-get upgrade last time executed 2012-10-08 11:00 am).
The problem is still pending!!

We found out (comments in source files), that your company and especially your person take part at the development of usb-integration in the raspberrypi
Operating system “raspbian”.

One of our specalists found out the following:

the Attribute-Walk while dumping the USB-OTG-Controllers register:

> [ 186.296432] Modules linked in:
> [ 186.301819] CPU: 0 Not tainted (3.2.27+ #24)
> [ 186.308709] PC is at dwc_otg_dump_spram+0x50/0x114 [ 186.315683]
> LR is at sub_preempt_count+0x4c/0x50
> [ 186.322397] pc : [] lr : [] psr: 20000013
> [ 186.322408] sp : c7ae7e50 ip : 00000000 fp : c7ae7ecc [
> 186.337913] r10: c7a7c5e0 r9 : beac3918 r8 : c7ae7f78 [ 186.345057]
> r7 : c7088138 r6 : c7088120 r5 : c8840000 r4 : c8868000 [
> 186.353449] r3 : c88680e0 r2 : e8a34a0e r1 : 00000000 r0 : 0000002b
> [ 186.361820] Flags: nzCv IRQs on FIQs on Mode SVC_32 ISA ARM
> Segment user [ 186.370816] Control: 00c5387d Table: 07a88008 DAC:
> 00000015 [ 186.378412] Process udevadm (pid: 327, stack limit =
> 0xc7ae6268) [ 186.386337] Stack: (0xc7ae7e50 to 0xc7ae8000)
> [ 186.392717] 7e40: c06151b8 00000000 c05e7178 c0110d08
> [ 186.405156] 7e60: 00000101 c01025f4 c7ae7ecc c7ae7e78 c01025f4
> c01024bc c7ae7e8c c7ae6000 [ 186.417889] 7e80: 00000001 00000000
> c7ae7f34 c88680e0 c780fda0 000280d0 c061576c 00000000 [ 186.430726]
> 7ea0: c06151b8 c7b68000 c060bab4 c7088120 c7088138 c7ae7f78 beac3918
> c7a7c5e0 [ 186.443776] 7ec0: c7ae7ee4 c7ae7ed0 c02e3970 c02e8f48
> c02e394c c0435824 c7ae7efc c7ae7ee8 [ 186.457048] 7ee0: c02963a0
> c02e3958 c0435824 c05ec6a8 c7ae7f3c c7ae7f00 c015a5f4 c0296380 [
> 186.470607] 7f00: 00000001 c7847b00 c7b4fda0 00001000 c7806820
> c7b4fda0 00001000 beac3918 [ 186.484370] 7f20: c7ae7f78 00001000
> c7ae6000 00000000 c7ae7f6c c7ae7f40 c00f3c98 c015a560 [ 186.498403]
> 7f40: 00000003 c75ab3f8 c7ae7f94 00000000 00000000 c7b4fda0 beac3918
> 00001000 [ 186.512527] 7f60: c7ae7fa4 c7ae7f70 c00f3d84 c00f3bfc
> beac4918 00000000 00000000 00000000 [ 186.526746] 7f80: 000ea1d8
> 000ea9f0 000ea180 000ea1d8 00000003 c000e3e8 00000000 c7ae7fa8 [
> 186.541127] 7fa0: c000e140 c00f3d48 000ea9f0 000ea180 00000003
> beac3918 00001000 00000000 [ 186.555562] 7fc0: 000ea9f0 000ea180
> 000ea1d8 00000003 00000000 00000000 000ea9f0 0002a2fc [ 186.570035]
> 7fe0: 00000003 beac3910 00017124 40197c5c 60000010 00000003 07ffe821
> 07ffec21 [ 186.584555] [] (dwc_otg_dump_spram+0x50/0x114)
> from [] (spramdump_show+0x24/0x44) [ 186.600201]
> [] (spramdump_show+0x24/0x44) from []
> (dev_attr_show+0x2c/0x58) [ 186.615292] []
> (dev_attr_show+0x2c/0x58) from []
> (sysfs_read_file+0xa0/0x148) [ 186.630555] []
> (sysfs_read_file+0xa0/0x148) from [] (vfs_read+0xa8/0x14c) [
> 186.645435] [] (vfs_read+0xa8/0x14c) from []
> (sys_read+0x48/0x80) [ 186.659584] [] (sys_read+0x48/0x80)
> from [] (ret_fast_syscall+0x0/0x48) [ 186.674239] Code:
> e28330e0 e1530004 e50b3040 9a00002b (e5d42000)

The function dwc_otg_dump_spram() is containd in the file drivers/usb/host/dwc_otg/dwc_otg_cil.c definiert. the dwc_otg-driver is written by Synopsys, but there is no contact address in the header of the file..

> * Synopsys HS OTG Linux Software Driver and documentation
> (hereinafter,
> * “Software”) is an Unsupported proprietary work of Synopsys, Inc.
> unless
> * otherwise expressly agreed to in writing between Synopsys and you.

(sources from:
This driver is not part of the Standard-Kernel )

The problem is also discussed in the internet (see link,
but we did not found any solution or work around to solve this problem.
Can You help as, or do You know any person at your company who can solve the problem or give us some assistance?

Our current kernel is: 3.2.27+ #244 PREEMPT Sat Oct 6 17:26:38 BST 2012 armv6l GNU/Linux
Thanks a lot
Dipl-Math. Bernd Böttinger
Leiter Software-Entwicklung ——————————————–
Kirchhoff Datensysteme Services GmbH & Co. KG
Vollbrachtstraße 17
D-99086 Erfurt
Fon : +49 361 6646110
Fax : +49 361 6646118
E-Mail :

Comments are closed