t still, you don't want to go
around losing keystrokes whenever the CPU decides to change speed. And
you can't fix that by changing usbhid.
Alan Stern
-
Take Surveys. Earn Cash. Influence the Future of IT
Join Source
On Thu, 15 Feb 2007, Oliver Neukum wrote:
> Am Donnerstag, 15. Februar 2007 22:06 schrieb Alan Stern:
> > On Thu, 15 Feb 2007, Oliver Neukum wrote:
> >
> > > The trouble with HID is that it does input and output. I choose to queue
> > > output if the device is su
p file:
echo disabled >/sys/bus/usb/devices/usb2/power/wakeup
echo -n 0 >/sys/bus/usb/devices/usb2/power/state
Note that you will need to keep the CONFIG_PM_SYSFS_DEPRECATED option for
that second line to work properly.
Alan Stern
--
b(). The callback should also
determine whether the touchpad is being pressed, use that information to
decide how long the next timer interval will be, and restart the timer.
Alan Stern
-
Take Surveys. Earn Cash. Influence
IMO this doesn't matter very much.
The other disadvantage is that there will be timer callbacks occurring
(at widely-spaced intervals, to be sure) as long as the device is in use.
Again, this shouldn't matter very much -- since the device is in
n program is sending a
message to the device which causes it to crash. Are you certain that the
same programs are running on both machines and are sending the same
messages?
Suggestion: Use usbmon to see the data being sent (instructions in the
kernel source file Documentation/usb/usbmon.txt
elatively long (more than a few seconds) and where the device
gets used quite heavily at times (like a mass storage device). You would
end up with one timer callback every minute -- or whatever the delay is --
instead of hundreds of mod_timer() calls every second. Would this help
with
How can you prevent the completion handler from being called? There are
two ways: (1) Stop resubmitting the URB, or (2) Tell the device not to
send any data.
> Another possibility (for me) would be to make the interval (in the untouched
> times)
> longer. Is this possible?
The interval do
he HC messed
up on this URB, please resubmit it". For one thing, you'd have to change
every single USB driver to make them recognize the new code.
Far better to have the HCD handle things all by itself.
Alan Stern
-
ou can try. Turn on CONFIG_PM_SYSFS_DEPRECATED and
keep your "any_connected" change. Then do
echo -n 2 >/sys/bus/usb/devices/usb2/power/state
and see what happens when you plug in or remove a device.
Alan Stern
---
that, and the USB
core does try to send such a message when a reset occurs.
Talk to the maintainer of the usb audio driver -- that's where support
needs to be added for the suspend, resume, pre_reset, and post_reset
methods.
Alan Stern
-
On Tue, 13 Feb 2007, Jon Smirl wrote:
> On 2/13/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> > It's possible that after resuming, the device is still functional but in
> > some weird state. Perhaps sending it the right sort of HID messages would
> > get it fully
re means nothing.
> Detection messages:
> ohci_hcd: 2006 August 04 USB 1.1 'Open' Host Controller (OHCI) Driver (PCI)
> ohci_hcd :00:01.4: OHCI Host Controller
> ohci_hcd :00:01.4: new USB bus registered, assigned bus n
UNLOCKED.
Use the driver-model macros dev_err(), dev_info(), etc. instead of err()
and info().
Do you really want to start a read as soon as the driver is registered?
Why not wait until a user process opens the device file? That way you
could leave the device suspe
oes not wait (as the first
> time) to restart the transfer. Is it possible to let the USB device restart
> the transfer after I touched the pad again?
Yes. :-)
Your question isn't clear. It looks like you're asking how to restart the
transfer. T
0301, change 0001, 1.5 Mb/s
> hub 1-0:1.0: debounce: port 1: total 100ms stable 100ms status 0x301
> usb usb1: suspend_rh (auto-stop)
This means that the dongle isn't working at all. Make sure the USB
connections aren't loose.
Alan Stern
--
esn't do anything when CONFIG_USB_SUSPEND isn't set.
> Do you think that's normal ?
Yes.
Alan Stern
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you&#
_SUSP | USBPORTSC_RD)))
> + if (!(inw(port_addr) &
> + (USBPORTSC_RD | USBPORTSC_RD)))
You typed USBPORTSC_RD twice instead of USBPORTSC_SUSP. Unfortunately,
that was the whole point of the new patch. So
On Tue, 13 Feb 2007, Jon Smirl wrote:
> On 2/13/07, Jon Smirl <[EMAIL PROTECTED]> wrote:
> > On 2/13/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> > > It's possible that after resuming, the device is still functional but in
> > > some weird state. Perha
529.80] usb 5-2: usb auto-resume
> Feb 12 22:51:30 frida kernel: [ 3529.832000] uhci_hcd :00:1d.3:
> port 2 resume iterations 0:0
> Feb 12 22:51:30 frida kernel: [ 3529.832000] uhci_hcd :00:1d.3:
> port 2 portsc 1095,01
Yep, that's the problem. It's good to hav
d delay by altering the value of
USB_AUTOSUSPEND_DELAY in the kernel source file drivers/usb/core/usb.h.
For example, setting it to (HZ*60) will cause the suspend to happen only
after 60 seconds of inactivity.
Or you can prevent USB devic
a remote wakeup.
>
> I'd appreciate some advice about that. Is it possible that this command
> is what's needed to make remote wakeup work?
Probably not.
> Should I try to put the
> above call at some other place in the suspend p
after resuming, the device is still functional but in
some weird state. Perhaps sending it the right sort of HID messages would
get it fully working again. But I don't want to mess around with
experimenting and writing all sorts of special-purpose programs to te
This patch (as850b) disables remote wakeup (and everything else!) on
all EHCI ports when the shutdown() method is called. If remote wakeup
is left active then some systems will reboot instead of powering off.
This fixes Bugzilla #7828.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
Cc:
those
> drivers necessary? My ohci-s3c2410.c patch seems to work fine without
> it.
No, it isn't necessary. IMO it should be removed from those other
drivers. On the other hand, David Brownell said that it should be left in
since it does no real harm.
My advice is not to add in new unn
volved in adding
autosuspend support to usb-skeleton. Nothing special is needed, but
getting the locking just right requires a good deal of care; you have to
worry about the BKL in addition to the driver's own io_mutex.
Alan Stern
-
"driver" might indeed be NULL. It wouldn't be hard
to add synchronization, if you really think it is needed.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff do
d prevent using bitwise operations.
2. It might prevent (depending on how clever the implementation was)
certain optimizations on little-endian CPUs.
3. It would essentially force all accesses to be treated as if they
were unalig
int status;
> };
>
> struct urb;
Do we also want to update the corresponding userspace-visible declaration
in usbdevice_fs.h?
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, securi
o claim.
> How can I best properly receive the remote keystrokes without having to parse
> the usbmon output ?
Find out what's wrong with the input parsing and fix it, so that you don't
need a special driver.
Alan Stern
---
id, and the
hub driver. (Maybe usblp does too; ISTR it was updated recently.) The
hub driver is a little more complicated than the other two because it also
supports autosuspend.
Alan Stern
-
Using Tomcat but need
to resume devices from user
> space. Therefore the delay interface should resume the device.
I don't follow your argument. Why does the "delay" interface (more
properly, the "autosuspend" attribute) need to be able to resume devices?
There are other w
r and may be changed in the future.
Clearly the declaration is wrong; the status field should be signed.
Please feel free to fix it.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
t; FX2 status registers I find that the 2nd OUT packet never arrived at the
> FX2's EP.
Nope, the second OUT packet definitely arrived. I know for certain,
because the host received an ACK in response -- it's in y
On Fri, 9 Feb 2007, Jon Smirl wrote:
> On 2/9/07, Alan Stern <[EMAIL PROTECTED]> wrote:
> > Are you certain? The mute button is on the top front edge, but the light
> > appears about 2/3 of the way up the front side. (There may be another
> > light right on the m
On Fri, 9 Feb 2007, Oliver Neukum wrote:
> Am Freitag, 9. Februar 2007 23:12 schrieb Alan Stern:
>
> Firstly, cool.
> I'll look at it closer again, but just some first comments.
>
> > +static ssize_t
> > +set_autosuspend(struct device *de
Get-Report as part of its resume
processing.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier.
Download IBM
in it working on a hub, but not on a host?
I haven't tried that yet. Too short on time...
> > This looks like a prime candidate for a quirks entry. I will create one
> > later on. And I'll mail the device back early next week.
>
&g
hink usb-storage has any way of knowing.
> And strictly
> speaking I am not suspending devices. I am managing vetos.
Call it what you will; the effect is the same.
Alan Stern
-
Using Tomcat but need to do more? Need to
ly does appear to be a bug in the
device. Maybe it wants some special command before it will wake up
fully... but a reset works just as well.
This looks like a prime candidate for a quirks entry. I will create one
later on. And I'll mail the device back early next week.
Alan Stern
--
as853, posted a few minutes ago.
Alan Stern
Index: usb-2.6/fs/sysfs/file.c
===
--- usb-2.6.orig/fs/sysfs/file.c
+++ usb-2.6/fs/sysfs/file.c
@@ -516,6 +516,30 @@ int sysfs_create_file(struct kobject * k
/**
+ * sysfs_add_file_to_
.
Improve some comment texts.
Don't log the quirks value if it is 0.
Make sysfs display the quirks value in hexadecimal, which
will be more informative.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
Index: usb-2.6/drivers/usb/cor
t;
> Any insights would be very helpful.
Try using the usbmon facility (instructions in the kernel source file
Documentation/usb/usbmon.txt) to see what's really happening.
[And please instruct your email client to wrap lines at 74 columns or
thereabouts.]
Alan Stern
In
fact, the only multi-interface device I have is a Bluetooth receiver,
which I haven't ever tried to use.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickl
lbacks for a single URB.
>From what you wrote, it looked as though you may have misunderstood this.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-i
On Fri, 9 Feb 2007, Sergei Organov wrote:
> Alan Stern <[EMAIL PROTECTED]> writes:
>
> > On Thu, 8 Feb 2007, Sergei Organov wrote:
> >
> >> IMHO aligned(N) should be used in portable code only to *increase*
> >> alignment (this effectively excludes usage
On Fri, 9 Feb 2007, Oliver Neukum wrote:
> Am Donnerstag, 8. Februar 2007 17:09 schrieb Alan Stern:
> > You shouldn't call usb_autopm_get_interface(). The changes it makes are
> > lost when the driver fails to bind to the interface. You should call
> > usb_auto
oller. However you
can test the performance of a host controller + USB device combination.
One way is to use "dd" to transfer a large amount of data to or from the
device, and see what the throughput is.
Alan Stern
---
, you could remove the unnecessary code from ether.c.
It would be enough to have the line say:
req->zero = value < wLength;
as in zero.c.
Alan Stern
-
Using Tomcat but need to do more? Need to suppor
ore testing seems to indicate that multiple outstanding URBs does make
> a very small (10%?) difference in performance, and staring at usbmon, it
> appears that sometimes I do in fact get multiple callbacks from the same
> frame. But it's
pecial-case handling like this, there is no way
to unconfigure these non-compliant devices.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
Index: usb-2.6/drivers/usb/core/devio.c
===
--- usb-2.6.orig/drivers/usb/core/de
t's really
happening?
> I think I more-or-less decided it wouldn't be such a good
> thing for a host to send a zillion IN tokens only to have the device
> send a zillion NAKs: it'd suck up bandwidth that could be useful
On Thu, 8 Feb 2007, Phil Endecott wrote:
> Alan Stern wrote:
> > On Thu, 8 Feb 2007, Phil Endecott wrote:
> >> > From a
> >> > higher-level (source code) point of view, the alignment doesn't affect
> >> > the
> >> > set of operation
t;
> > All this in addition to the standard device-wide autosuspend delay, which
> > will also be settable via sysfs. The initial timeout value should be set
> > to the maximum of the device-wide delay and the interface delays.
> >
> > Is that right?
>
> Yes
's not like trying to take the address of a bitfield member. That
really can't be done.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integr
r question. Simple?
If you don't consider the alignment attribute to be part of the type, then
you can indeed express "16-bit-little-endian-unsigned-integer-unaligned"
as a C type. In fact __le16 expresses it perfectly all except for the
"unaligned" part, whi
ta size to be set in an ISO URB, right?
You mean, urb->transfer_buffer_length? Yes, that has to be set in an ISO
URB. It has to be long enough to encompass all the packets in the
descriptor list (although usbcore does
hey aren't performed in process context at all;
they are done in interrupt context.
The callbacks can't occur out of order because the host controller driver
processes the URBs in the order they were submitted. They can't occur
concurrently on
This patch (as851) fixes a bug introduced when the USB ID matching
code was refactored. If any of the device-level matches fail we
should reject right away. Instead, when none of the device-level
matches fail the current code accepts right away.
Signed-off-by: Alan Stern <[EMAIL PROTEC
.
All this in addition to the standard device-wide autosuspend delay, which
will also be settable via sysfs. The initial timeout value should be set
to the maximum of the device-wide delay and the interface delays.
Is that right?
Alan Stern
--
ite callbacks. However read callbacks could be
interspersed more or less arbitrarily among write callbacks -- although
again, not concurrently.
For more information, read the kerneldoc in the source file
drivers/usb/core/urb.c.
Alan Stern
-
e my as845 patch,
http://marc.theaimsgroup.com/?l=linux-usb-devel&m=117019834316197&w=2
which is nearly identical to Andrew's.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get st
gth = value;
> - req->zero = value < w_length;
> + req->zero = value < w_length &&
> + (value % gadget->ep0->maxpacket) == 0;
> value = usb_ep_queue (gadget->ep0, req, GFP_ATOMIC);
>
tes-little-endian-signed-integer" -- although you could easily make
up your own.)
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrate
_get_interface(intf);
> return -ENODEV;
> }
You shouldn't call usb_autopm_get_interface(). The changes it makes are
lost when the driver fails to bind to the interface. You should call
usb_autopm_resume_device() instead -- unfortunately it isn't EXPORTed.
d with a USB
mass-storage device. They are meant to be used only with devices which
support the special bulk-echo testing protocol.
> That induces an endpoint error at our
> hostcontroller layer.I think this may be the reason if i am not
> wrong.Please guide me in this.
You are
different interfaces on the same device have
radically differing autosuspend delay values? My feeling is that
autosuspend is of relatively low importance, and it should not be made any
more complicated than necessary.
Alan Stern
--
On Wed, 7 Feb 2007, Alan Stern wrote:
> This patch (as850) disables remote wakeup on EHCI ports when the
> shutdown() method is called. If it is left active then some systems
> will reboot instead of powering off. This solves Bugzilla #7828.
I'm having second thoughts about this
ing BKL means that the pointer to the device may become invalid.
> Here's the fix.
A good example of the dangers of relying on BKL for anything serious.
Alan Stern
-
Using Tomcat but need to do more? Need to support w
On Wed, 7 Feb 2007, Pete Zaitcev wrote:
> Alan, do you happen to know how much usb_iso_packet_descriptor thingies
> we allocate typically? One? Three? Tens? A hundred? In usbfs we allow 128.
> If we routinely get lots of them per URB then I need to rethink the line
> format and all other things. I
This patch (as850) disables remote wakeup on EHCI ports when the
shutdown() method is called. If it is left active then some systems
will reboot instead of powering off. This solves Bugzilla #7828.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
Index: usb-2.6/drivers/usb/host/ehci
like this.
Let someone else do it. (If any of you decides to propose this API more
widely, please remember to CC: me!)
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quick
On Wed, 7 Feb 2007, Oliver Neukum wrote:
> Am Mittwoch, 7. Februar 2007 18:40 schrieb Alan Stern:
> > On Wed, 7 Feb 2007, Oliver Neukum wrote:
>
> > > So the answer would not be the largest delay, but the largest delay
> > > currently requested to be active.
> &g
On Wed, 7 Feb 2007, Greg KH wrote:
> On Wed, Feb 07, 2007 at 12:53:31PM -0500, Alan Stern wrote:
> > Greg:
> >
> > When you refactored the USB device-matching code, you may have introduced
> > a bug. Does it seem reasonable that an entry might contain both
> &
get_unaligned_be16(), get_unaligned_be32(), get_unaligned_be64(),
and the corresponding put_unaligned...() things?
But for some unclear reason people weren't in favor of it. And as a
result, the kernel is filled with roll-your-own versions o
below needed? If yes, I'll send it with a proper patch
comment and S-O-B.
Alan Stern
Index: usb-2.6/drivers/usb/core/driver.c
===
--- usb-2.6.orig/drivers/usb/core/driver.c
+++ usb-2.6/drivers/usb/core/driver.c
@@ -415,8 +
v,
"won't remote wakeup, status %d\n",
status);
> > messages. Moreover, I had verified that the sysfs wakup attribute is
> > set.
>
> If the device is not bound to a driver, that is ignored.
Not so. T
device delay. Thus non-PM-aware drivers, or drivers that don't
use the new delay value, will continue to work as before.
This way using the maximum delay value should do exactly what you want.
Right?
Alan Stern
-
U
e timer actually use? The simplest answer is to use the
maximum of the delays. Does that sound right?
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre
t because some architectures cannot do DMA to stack
memory.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier.
0
I'd really prefer to see something like the patch below added instead of
your idle timer. (Note: not even compile-tested!)
With this infrastructure in place, all you have to do is
set_bit(USB_DF_DEVICE_BUSY, &hid_to_usb_dev(hid)->dflags);
whenever an event occurs. No
ensured.
>
> Documentation/io_ordering.txt
Not to mention Documentation/memory_barriers.txt and
Documentation/atomic_ops.txt.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
Ge
Someone who knows about PCI and EHCI may be able to answer this...
It's odd that the enumeration fails partway through instead of right at
the beginning.
Alan Stern
-- Forwarded message --
Date: Sun, 04 Feb 2007 14:20:52 -0500
From: [EMAIL PROTECTED]
To: linux-usb-
because sometimes devices just disconnect
> for no good reason, e.g. if they have bad power or a cables are not shielded
> well. But I don't have better ideas.
Another possibility is a recently-discovered bug in the USB core. Here is
a patch to fix the bug; perhaps it will help solv
This patch (as849) fixes a bug in the USB hub driver. A single
pre-allocated buffer is used for all port status reads, but nothing
guarantees exclusive use of the buffer. A mutex is added to provide
this guarantee.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
This patch should g
On Sun, 4 Feb 2007, Oliver Neukum wrote:
> Am Sonntag, 4. Februar 2007 17:07 schrieb Alan Stern:
> > On Sat, 3 Feb 2007, Oliver Neukum wrote:
> >
> > > If you want simplicity, I'd suggest serialization. kmalloc for 4 bytes
> > > is overkill. I think it i
is to get in touch with the people
who maintain the jpilot and pilot-xfer programs. They may be able to
offer more help than I can.
Alan Stern
-
Using Tomcat but need to do more? Need to support web services, security?
G
On Sun, 4 Feb 2007, Prakash Punnoor wrote:
> Am Sonntag 04 Februar 2007 schrieb Alan Stern:
>
> > Prakash, here's an alternative patch. It should work just as well as the
> > previous one (don't try to apply them both!).
>
> Seems to work fine for me. Nice wo
to be caused
either by the Treo itself or by the pilot-xfer program.
Do you agree? While there may indeed be issues remaining to be fixed,
none of them are directly connected with the kernel.
Alan Stern
-
Using To
are idle almost all the
time, and it doesn't seem worthwhile to allocate permanently one URB per
hub if it will hardly ever be used.
Prakash, here's an alternative patch. It should work just as well as the
previous one (don't try to apply them both!).
Alan Stern
On Sat, 3 Feb 2007, Oliver Neukum wrote:
> Am Samstag, 3. Februar 2007 23:02 schrieb Alan Stern:
> > Obviously we can't afford to use a single dedicated buffer like this.
> > Each thread reading a hub's port status will have to provide its own
> > buffer,
lf, but you can go ahead and try it. It ought to eliminate
the problems you've been seeing.
Alan Stern
Index: usb-2.6/drivers/usb/core/hub.c
===
--- usb-2.6.orig/drivers/usb/core/hub.c
+++ usb-2.6/drivers/u
On Sat, 3 Feb 2007, Pete Zaitcev wrote:
> On Sat, 3 Feb 2007 11:43:50 -0500 (EST), Alan Stern <[EMAIL PROTECTED]> wrote:
>
> > Here's an idea for an improvement to the text interface in usbmon which
> > should be easy to implement. I'd like to know for cer
On Sat, 3 Feb 2007, Prakash Punnoor wrote:
> Am Freitag 02 Februar 2007 schrieb Alan Stern:
> > When you have a chance, please try hooking the scanner up to that bad hub
> > port again, and use the usbmon facility. Instructions are in the kernel
> > source file Documen
or more events.
The dropped events may already be listed in the "s" file, but usually
people copy only the "t" file, and in case correlating the two isn't as
easy as seeing the information right there. Having an in-line indication
of dropouts could help a
This patch (as848) adds a useful little debugging message to let us
know when ehci-hcd's bus_suspend method runs. The other HCDs have
similar messages; now ehci-hcd doesn't need to feel left out.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
Index: usb-2.6/drivers/usb
it can never succeed.
Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
---
Index: usb-2.6/drivers/usb/core/hub.c
===
--- usb-2.6.orig/drivers/usb/core/hub.c
+++ usb-2.6/drivers/usb/core/hub.c
@@ -1902,6 +1902,7 @@ static int hub_
On Fri, 2 Feb 2007, Prakash Punnoor wrote:
> Am Freitag 02 Februar 2007 schrieb Alan Stern:
> > When you have a chance, please try hooking the scanner up to that bad hub
> > port again, and use the usbmon facility. Instructions are in the kernel
> > source file Documen
rd contains only
about 1 GB, not 2 GB.
As I understand it, the standards for SD cards holding more than 1 GB are
in flux. There are combinations of cards and readers which simply don't
work. This looks like one of them.
What happens if you use the same card and reader under anothe
ry hooking the scanner up to that bad hub
port again, and use the usbmon facility. Instructions are in the kernel
source file Documentation/usb/usbmon.txt. Start acquiring the usbmon log
before running sane, and stop it after sane has failed.
901 - 1000 of 6560 matches
Mail list logo
