Please use Reply-To-All; otherwise people reading the mailing list
will never benefit from what you have learned.
On Tue, 11 Dec 2007, tj wrote:
> Alan Stern wrote:
>
> >On Tue, 11 Dec 2007, tj wrote:
> >
> >
> >
> >>I have a Canon FB630U USB sc
er on a Compag system, OHCI, and it works first
> time every time. No hesitation. Works perfectly.
>
> I also have a Slackware 12 installation on the Via UCHI system and it
> exibits the exact same problem as the Ubuntu installation.
>
> So, why is it working on one syst
bug. Maybe then I'll recompile just the modules.
It could still be an endianness problem. Some OHCI hardware uses one
byte ordering for the register access and a different byte ordering for
the in-memory data structures. If you look through the Kco
ested almost all userland possibilities:
> - with/without the ohci/ehci drivers
> - old_scheme_first
> - other cable/device
>
> Hoping anybodyhas a clue, I may provide any additional
> info you may find relevant.
It sounds like the phone works perfectly on the x86 system. The mip
ossible to prevent the detach, because the detach is done by
the device itself rather than by Linux. That's why you have to unplug
the device to make it start working again.
Alan Stern
-
SF.Net email is sponsored by:
Chec
atement is completely wrong -- nothing you do in
the peripheral's firmware can cause the host to sent a poll.
> If the poll doesn't arrive, the data will still accumulate.
Why? If you don't submit the data until after the previous packet has
been sent, then the data can&#x
t
expected.
> Can I force the host to poll every time I have data?
No. The host polls when it wants to; you can't affect it.
> It polls every 1ms
> anyway on the interrupt eps. I use a Poll Demand trigger every time I
> have data available for the DMA to transfer to the USB TxF
me more about this URB timeout and
> what it's caused by? Could the USB stick be misbehaving or the kernel
> simply not up to date enough for these 2.0 sticks?
It's not possible to tell what's causing the timeou
ough to do this:
$ echo /sys/devices/pci:00/:00:1d.0/usb*
/sys/devices/pci:00/:00:1d.0/usb1
/sys/devices/pci:00/:00:1d.0/usb_host:usb_host1
This says that right now, that particular bus is labelled "usb1".
Alan Stern
--
On Mon, 3 Dec 2007, David Griffith wrote:
> On Mon, 3 Dec 2007, Alan Stern wrote:
> > These are low-level hardware errors. Maybe in the device, maybe caused
> > by the USB cable, maybe in your EHCI controller.
> >
> > Some tests to try:
> >
> >
nly and to go through a
disconnect/reconnect cycle whenever the data changes. Or else you
could have a program on the host that flushes the cached pages.
A less clumsy approach might be to avoid being a mass-storage device.
But I don't know
bug report again. It had "error -110", not
"error -71". The two are very different.
Alan Stern
-
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From t
ge to install 2.6 on that old machine, at least go to
2.4.
Alan Stern
-
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it
rnel.
The feature was added in 2.6.21. The patch for it is here:
http://marc.info/?l=linux-usb-devel&m=117089528503119&w=2
Most likely it will be easier for you to upgrade to a current kernel
than to backport the patch.
Alan Stern
/usb-storage-always-set-the-allow_restart-flag.patch
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse
hat's in
the partition sector by doing:
dd if=/dev/sdb count=1 | hexdump -C
Maybe for some reason it just needs to be repartitioned.
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenges
On Sat, 24 Nov 2007, RTE wrote:
> On Sat, 24 Nov 2007 07:39:13 +0300, Alan Stern wrote:
>
> > You might also want to test this patch. It should automatically set
> > max_sectors to 64 -- that's lower than you need but it should work
> > without errors. (The patch
On Fri, 23 Nov 2007, RTE wrote:
> On Wed, 21 Nov 2007 00:26:26 +0300, Alan Stern wrote:
> > Okay, here's a version of the patch for 2.6.23 and without the
> > debugging information.
>
> Many thanks to you for the help! :)
> In 2.6.23 patch works perfectly. Errors are
et detection isn't handled by audio or
video class drivers; it is handled by the hub driver.
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http:
make sure your gadget driver sends the correct descriptors to the
host. In the reply to a Get-Config-Descriptor request, you have to
send the configuration descriptor followed by all the interface and
endpoint descriptors for that configuration.
Alan Stern
On Tue, 20 Nov 2007, RTE wrote:
> On Mon, 19 Nov 2007 23:17:17 +0300, Alan Stern wrote:
>
> > Below is another test patch for 2.6.17-git22. It includes new code to
> > try and fix the problem.
>
> it works :)
> no errors in the copied data
Great!
Okay, here&#
On Mon, 19 Nov 2007, RTE wrote:
> On Fri, 16 Nov 2007 22:31:20 +0300, Alan Stern wrote:
>
> > Okay, that's good. Now here's a comparable patch for 2.6.17-git22
>
>
> 2.6.17-git22 with patch
> after copying 700 MB file from flash to HDD
> the copied file c
s not clear how such errors would cause
the drive to corrupt anything.
The large number of errors does indicate a serious problem somewhere.
Have you tried using the drive with a different brand of computer? Or
a different USB cable?
Alan Stern
On Fri, 16 Nov 2007, RTE wrote:
> On Fri, 12 Oct 2007 00:18:11 +0400, Alan Stern <[EMAIL PROTECTED]> wrote:
> > I'd like you to try some more experiments
>
> sorry for the long period of silence - not enough free time :(
>
> 2.6.17-git21 whith patch
> after co
t could
be a bad cable (or the equivalent).
The kernel log posted above isn't complete, that is, there are lines
missing in the middle. You'd be better off using usbmon (see
Documentation/usb/usbmon.txt).
Alan Stern
--
know. However you might be better off using a Gumstix
device (http://www.gumstix.com/) instead of a notebook. Other possible
device controllers are listed on the Gadget page at www.linux-usb.org.
Alan Stern
-
This SF.ne
On Wed, 14 Nov 2007, Diego Zuccato wrote:
> Alan Stern ha scritto:
> > On Tue, 13 Nov 2007, Steve Kelem wrote:
> >
> >> I have a USB hard drive that I'm trying to use with Linux.
> >> It does mount, but it takes 30 seconds!
> [...]
> > Possibly
ReiserFS: sda1: checking transaction log (sda1)
> ReiserFS: sda1: Using r5 hash to sort names
That's not very useful, is it? :-(
Try using the usbmon facility. Instructions are in the kernel source
file Documentation/usb
g to wait 30 seconds for an "ls" is
> horrible.
>
> Any idea of what's wrong?
Possibly some error occurs, the system waits during a 30-second
timeout, and then the disk is reset.
What shows up in the output from dmesg after you mount the filesystem?
Alan Stern
use
it, you'll have to port it yourself.
There is no documentation other than what exists in the kernel source.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find probl
program to read the
data back and see what bytes ended up in which sectors.
At the same time, use the usbmon facility (see
Documentation/usb/usbmon.txt) to record exactly what data gets sent
to/from the stick. That should at least be sufficient to determine the
problem is in the stick itself,
s uses). Any ideas why
> this might be happening? I've been digging around the firmware code, but
> I was hoping someone might have some suggestions.
Perhaps the device doesn't transmit any packets in response to the IN
packet for the interrupt endpoint. It should reply with either NAK o
;
> Can anyone please suggest me how to rectify the error.
I don't have any problem with "modprobe g_ether" in 2.6.24-rc1. Are
you sure that dummy_hcd is properly loaded?
Alan Stern
-
This SF.net email
ecause it is so obvious... Upgrading to 2.6.23
will probably help a lot. The 2.6.9 kernel is quite old. Furthermore,
nobody will be willing to help you debug the problem unless you use a
recent kernel.
Alan Stern
-
Th
if there's some subtle
> problem with the non-real-time signals that I should be aware of.
I don't know of any reason for this restriction.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still g
ept putting an OS on it,
of course).
> Can it be done? If so, how can it be done?
I have never tried to do it. Maybe other people have. I assume
nothing special is needed, but I haven't looked into the matter.
Alan Stern
very long timeout. Alternatively, when a timeout does occur don't
return to the user; instead loop back and submit another usb_bulk_msg.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through
hat come from the USBDEVFS_RELEASEINTERFACE IOCTL
> sent by libusb upon release the interface? Would you expect this to be
> resetting the host controllers data toggle to 0? I can look into the
> firmware to see what it is doing in response.
Yes and yes. See section 9.1.1.5 of the USB 2.0 s
arly fulfilling that role, the development
list would be the best place for this.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and conf
rb->status.
The comment in the header file is wrong; the signal always gets sent
provided you set signr to something other than 0.
> Or maybe there is some other approach that I should be taking.
Nope. This is the way to go.
Alan Stern
--
not sure if there is any relation between
> this
> and the recent reset loop issue.
In what respect isn't your EHCI controller working? Can you provide
dmesg logs with CONFIG_USB_DEBUG enabled?
Alan Stern
-
Th
nd I also have the
> problem outlined here:
>
> http://ubuntuforums.org/showthread.php?p=3624990#post3624990
I don't know what patches might fix your problem because I don't know
what your problem is. It would help if you build a kernel with
CONFIG_
in what
SendControlMessage() does. The most natural approach would be for
SendControlMessage() not to return until the message has been sent and
the buffer is empty.
Alan Stern
-
This SF.net email is sponsored by: Splunk I
I "standards"
> enough that they need special drivers?
No. They differ from the "standard" in that they use a
platform-specific interface instead of PCI. Otherwise they are pretty
much the same as any other OHCI controller.
> Or would any chip need a specific
> dr
On Wed, 24 Oct 2007, Phil Endecott wrote:
> Alan Stern wrote:
> > Interrupt endpoints don't pack more than one message into a single
> > packet.
>
> OK. So considering the example of a keyboard where the user has
> pressed two keys within one polling period, they
to drop events if
they occur too quickly.
BTW, in spite of the name it's a mistake to identify interrupt endpoint
behavior too closely with actual hardware interrupts. You're better
off thinking of them as bounded-latenc
r words, the device sent a larger packet
than the host expected, like a 64-byte packet when the host expected
only 18 bytes, or a 3-byte packet when the host expected only 2 bytes.
Alan Stern
-
This SF.net email is spo
ever you can do a port
reset using the program posted here:
http://marc.info/?l=linux-usb-users&m=116827193506484&w=2
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log file
ny
interrupts.
Also it might help if you build a kernel with CONFIG_USB_DEBUG turned
on. Then the dmesg log could provide more information.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping thro
On Wed, 17 Oct 2007, David Miller wrote:
> From: Alan Stern <[EMAIL PROTECTED]>
> Date: Wed, 17 Oct 2007 11:51:57 -0400 (EDT)
>
> > > + break;
> > > + }
> > > + if (limit_1 < 0) {
> > > +
x there with the
> same results.
>
> Already tryed to disable ACPI.
>
> My OS is Ubuntu 7.04 server with kernel 2.6.20-16.
> Compact Fujitsu with Intel chipset.
>
> Please help!
Do other USB devices work on the Fujitsu computer?
Alan Stern
ead lock, talk to the OHCI
hardware, and drop the read lock. Nevertheless, the interference
occurs. David B.'s interpretation is that the hardware's change of
state takes more time than the CPU uses in manipulating locks and
switching tasks. Hence his suggestion for
On Tue, 16 Oct 2007, David Miller wrote:
> From: Alan Stern <[EMAIL PROTECTED]>
> Date: Tue, 16 Oct 2007 11:23:58 -0400 (EDT)
>
> > Do you have any idea _where_ in ohci_hub_control the hang still occurs?
> > Is it the same unbounded reset loop?
>
> Yes
Do you have any idea _where_ in ohci_hub_control the hang still occurs?
Is it the same unbounded reset loop?
Does the patch below satisfy both Davids?
Alan Stern
Index: usb-2.6/drivers/usb/host/ohci-hub.c
===
--- usb-2.6.orig/drivers/usb/host
fice. So maybe someone
> has already done something similar and there is a software I could use
> as a start.
Look at
http://www.linux-usb.org/gadget/
You want to use the serial gadget. Your "linux fil
On Fri, 12 Oct 2007, Dominique Dumont wrote:
> Alan Stern <[EMAIL PROTECTED]> writes:
>
> > Okay. That narrows it down to a single patch, this one:
> >
> > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=03aba2f79594ca94d159c8bab
On Thu, 11 Oct 2007, RTE wrote:
> On Fri, 31 Aug 2007 22:39:40 +0400, Alan Stern wrote:
>
> > There's a good chance that your problem isn't caused by anything in the
> > USB stack, but rather by a change somewhere else.
>
> I did not use Git (150 MB is
mply retry, but there's also the converse
> situation where the device has already been removed (e.g. it has
> "bounced") to worry about - is there a better way to do this?
This last problem is unavoidable. A USB device can be unplugged at any
time, and nothing -- not even a k
On Tue, 9 Oct 2007, Patrik Sevallius wrote:
> On 10/9/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> > On Tue, 9 Oct 2007, Patrik Johansson wrote:
> >
> > > Hi,
> > >
> > > I have a board with a BF531 and a ISP1362. I'm getting timouts when
&g
itializes the EHCI controller
(taking control of the port), sees that the device isn't high speed,
and switches port ownership back to the companion controller, all
during the span of the companion's reset. The companion would never
know anything had happened! But then everyt
k I can
> find some Intel and Dell people who would be glad to test this out if
> you have a proposed patch.
Here is a proposed patch. I haven't tried running it, but it compiles
okay.
Alan Stern
Index: usb-2.6/drivers/usb/core/hcd.h
nly would be easy enough to make port resets mutually
exclusive with EHCI initialization. That would work on any platform,
PCI or not.
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log fi
device that times out
> on SET_ADDRESS is a Muvid MP-521.
If you can build kernels with CONFIG_USB_DEBUG set on both the board
and the desktop computer, then there might be some useful information
in the dmesg logs of the two mach
http://www.gniibe.org/ac-power-by-usb/ac-power-control.html
Alan Stern
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration f
?
If you build a kernel with CONFIG_USB_STORAGE_DEBUG enabled, the dmesg
log will contain lots of extra useful debugging information. It should
indicate where the problem is.
Alan Stern
-
This SF.net email is sponsored by
ty
> > where each is off by just a little bit, so they can work with other
> > devices but not with each other.
> >
>
> The knoppix experiment looks like they _can_ work with each other.
>
> dmesg from knoppix attached. Need more info?
There's no useful
On Fri, 5 Oct 2007, Jonas Nicolaisen wrote:
> On Fri, Oct 05, 2007 at 02:56:40PM -0400, Alan Stern wrote:
> > > What does the status -84 mean?
> >
> > It means there was a low-level USB communications error. Maybe the
> > cable connection isn't firm, or
use is working flawlessly, as well
> as a small laptop mouse.
>
> What does the status -84 mean?
It means there was a low-level USB communications error. Maybe the
cable connection isn't firm, or maybe the mouse just stops working
every few seconds.
>
uot;C Bi:5:003:2 0 512" indicates a successful URB completion (together
with the first 32 bytes of data received). The other lines at the end
of the log show what the system does when the disconnect occurs.
Alan Stern
---
. Is there something outside of the kernel that changed? Or did I
> pick up some small kernel change?
Most likely you changed some other part of the kernel configuration,
something which caused the kernel to try to read the last sector.
Let us know if the patch fixes your problem.
Al
o read sector 16007040. Evidently the camera really
has one less sector than it claims. This is a very common bug.
A blacklist entry should fix the problem. To create I will need to see
the contents of /proc/bus/usb/devices with the camera plugged in.
Alternatively, the same information is
ell the driver it can disable DMA. But uhci-hcd never
disables DMA as long as any active devices are plugged into the
controller. The hardware doesn't permit it. So even though uhci-hcd's
overall interrupt count may be lower, it forces the CPU to stay out of
the deep sleep states that eh
st packets. The adapter driver may even poll the
USB device; I don't know how it works.
> > ehci doesn't poll at all under normal circumstances, but it does use a
> > timer to disable the async schedule. As I mentioned above, uhci polls
> > at 4/second.
> >
&
n averages very close to 0.
Another aspect that's impossible to determine just from the interrupt
averages is how much time the CPU spends servicing each interrupt.
Alan Stern
-
This SF.net email is sponsored by: Splunk
number of interrupts is actually the same. The
> > difference is in the kernel polling routines - ehci_work vs
> > usb_hcd_poll_rh_status. Does ehci need to poll twice as often, or might
> > it be possible to tune or hack the polling interval?
ehci doesn't poll at all u
hub is connected
> to them, but if a device even demands 100 mA it borks (f.e. my old
> webcam). Chipset is nvidia (CK8S USB Controller).
Odd. Any bus-powered device is allowed to draw up to 100 mA. This
sounds like a
On Wed, 3 Oct 2007, Dominique Dumont wrote:
> Alan Stern <[EMAIL PROTECTED]> writes:
>
> > You may find it helps to turn off HAL before running your tests,
> > because HAL does constant probing of USB storage devices.
>
> Is this probing limited to storage devic
50, system will hang up.
> So I think this problem is a timing problem, I don't know how to fix the
> problem. Could you give me some suggestions?
Sorry, I'm out of ideas.
Alan Stern
-
This SF.net email is s
of output (because I have verbose debugging enabled in the
> kernel?). I've attached the end of the dmesg output below.
The end doesn't say very much. Please attach the entire thing.
You may find it helps to turn off HAL before running
even though it won't mount. --Paul
If you run "fdisk -l /dev/sdh" before trying to mount the camera, what
is the output?
What shows up in the "dmesg" log after you plug in the camera and try
to mount it?
Alan Stern
y plugging the card reader into a regular Linux system to
make sure it is okay.
Third, try plugging a different USB device (like a flash memory) into
your machine, instead of the card reader.
The error message means that Linux didn't receive a status reply from
the card reader after sending i
On Sun, 30 Sep 2007, The Almighty Pegasus Epsilon wrote:
> On 9/30/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> > Strictly speaking, "character" and "block" are properties of the
> > driver, not of the device. It's perfectly possible to have both a
&
Sun, 30 Sep 2007, The Almighty Pegasus Epsilon wrote:
>
> > On 9/30/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> >> "more information"?! You have provided hardly any, beyond the fact
> >> that it doesn't work!
> >
> > I'll have to di
to best buy? I
> kinda need the space, as you can probably guess, as I paid for it, and
> would seriously appreciate a quick fix, if one exists.
The output from "dmesg" would help, and so would the contents of
/proc/bus/usb/devices. It also wouldn't hurt if you mentioned whi
On Sat, 29 Sep 2007, Stefan Richter wrote:
> Alan Stern wrote:
> > On Sat, 29 Sep 2007, Stefan Richter wrote:
> >> I've got the mentioned MP3 player with 4GB built-in flash memory and an
Ah, the key point. I missed it before, obviously.
> Before the firmware update
nt
There's not much point in these directory listings. The names of the
files don't convey any useful information; what matters is their
contents.
> The initial successful tests with the old firmware and what you see
> above is with Linux 2.6.23-rc6.
>
&
On Wed, 26 Sep 2007, Gergely Buday wrote:
> Alan Stern writes:
>
> > > Thanks. Everything seems to be fine with usb in dmesg, except this one:
> > >
> > > libusual: modprobe for usb-storage succeeded, but module is not present
> > What happens if you
On Wed, 26 Sep 2007, Gergely Buday wrote:
> Alan Stern writes:
>
> > > when I plug an usb mass storage device to my fc6 box, nothing happens.
> > > What information do you need to diagnose the problem? What log files
> > > should I investigate?
> >
>
better would be
if you first rebuild the kernel with CONFIG_USB_DEBUG enabled.
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/v
the disk disconnecting. Pretty much everything that follows
is irrelevant: We don't need to know what happened _after_ the drive
disconnected -- we need to know what happened _before_ (i.e., what
caused the disconnect).
A usbmon t
g, you must edit the source file
drivers/usb/gadget/file_storage.c. Change the lines saying
#undef DEBUG
#undef VERBOSE
to
#define DEBUG
#define VERBOSE
Then rebuild the driver and run your test again. When it is done, post
the output from dmesg on the board so we can see what happe
it in terms of dollars per port, you may find that
the financial requirements are impractically high no matter how the
hubs are arranged, unforunately. Guessing at an average retail figure
like $5 per USB port plus wifi adapter, you would need a minimum
L PROTECTED]>. Don't forget to follow the instructions
in Documentation/SubmittingPatches.
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual
y Isochronous URB no
matter what host controller you use. EHCI, OHCI, UHCI -- they all work
the same. It's part of the USB spec, and people are expected to know
that already.
Alan Stern
-
This SF.net email is sponsor
, each of 800 bytes,
lasting for 10 ms total.
For other types of URBs, there is only one buffer and it corresponds to
as many transactions as needed (roughly speaking, the buffer size
divided by the endpoint's max-packet size).
Alan Stern
---
amount of work needed for a 10-buffer URB is less than 10 times the
work needed for a 1-buffer URB.
Alan Stern
-
This SF.net email is sponsored by: Microsoft
Defy all challenge
you can make Linux tell the
phone to install whichever configuration you want.
You can find out what configurations are available either by looking at
/proc/bus/usb/devices or by running "lsusb -v", while the phone is
attached to the computer.
Alan Stern
-
software as in the A780? Beats me...
> Thank you very much for your help!
You're welcome.
> Unless you are still interested in seeing my logs when connecting the
> Z6, with the changes in my Linux kernel that you proposed, I will stop
> spamming the list and keep what I ha
m correctly.
> So are you
> saying that it is a utility on the client side (the mobile phone) that
> is made as a totally independent application?
I didn't say that. I have no idea how Motorola chose to implement
their client-side USB software. It
described.
That message indicates the programs from Canon are out-of-date.
Earlier kernels would accept the error condition described by the log
message, but current ones do not.
Alan Stern
-
This SF.net email is sponsor
1 - 100 of 1978 matches
Mail list logo
