| From | Subject | Date |
|---|---|---|
| Andrew Morton | Re: [Bugme-new] [Bug 13700] New: usb error flood in dmesg, m...
(switched to email. Please respond via emailed reply-to-all, not via the
bugzilla web interface).
It's a post-2.6.30 regression.
--
| Jul 3, 5:33 pm 2009 |
| Takashi Iwai | Re: [Bugme-new] [Bug 13700] New: usb error flood in dmesg, m...
At Fri, 3 Jul 2009 14:33:25 -0700,
Indeed.
Looking at the kernel messages, it seems relevant only with the usb stack.
Maybe bisecting with limited path (drivers/usb) would be easier.
Takashi
--
| Jul 3, 5:45 pm 2009 |
| Jiri Slaby | [PATCH 1/1] USB: usb-serial, remove unused variables
There are some unused variables in serial_do_down. Remove them.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
---
drivers/usb/serial/usb-serial.c | 6 ------
1 files changed, 0 insertions(+), 6 deletions(-)
diff --git a/drivers/usb/serial/usb-serial.c b/drivers/usb/serial/usb-serial.c
index 9e6027b..0d9bd45 100644
--- a/drivers/usb/serial/usb-serial.c
+++ b/drivers/usb/serial/usb-serial.c
@@ -281,8 +281,6 @@ bailout_serial_put:
static void serial_do_down(struct usb_serial_port *po...
| Jul 3, 5:09 pm 2009 |
| Randy Dunlap | [PATCH -next] gadget audio: select SND_PCM
From: Randy Dunlap <randy.dunlap@oracle.com>
Fix USB gadget audio: select SND_PCM, like many other sound
drivers do, to fix build errors:
drivers/built-in.o: In function `f_audio_playback_work':
audio.c:(.text+0x15a3e7): undefined reference to `snd_pcm_kernel_ioctl'
audio.c:(.text+0x15a471): undefined reference to `snd_pcm_lib_write'
drivers/built-in.o: In function `_snd_pcm_hw_param_set':
audio.c:(.text+0x15aca7): undefined reference to `snd_interval_refine'
drivers/built-in.o: In functio...
| Jul 3, 4:26 pm 2009 |
| David Brownell | Jul 3, 5:22 pm 2009 | |
| Ramya Desai | Creating URB in completion callback
Dear Experts,
I am trying to create a new URB in the completion call back.
Here I am giving the code that I am using.
void my_usb_stor_blocking_completion(struct urb *urb)
{
struct urb *rr_urb ;
//
//
//
rr_urb = (struct urb*)kmalloc(sizeof(struct urb), GFP_KERNEL);
if( NULL == rr_urb)
return -ENOMEM;
memcpy(rr_urb, urb, sizeof(struct urb));
usb_fill_bulk_urb(rr_urb,
current_urb_context->pusb_dev, current_urb_cont...
| Jul 3, 2:37 pm 2009 |
| Alan Stern | Re: Creating URB in completion callback
Don't use kmalloc to allocate URBs; use usb_alloc_urb.
You must not use GFP_KERNEL in a completion routine. Use GFP_ATOMIC
Fix those mistakes. No other special action is needed.
In fact, you might not have to create a new URB at all. You might be
able to re-use the URB that was passed to
my_usb_stor_blocking_completion().
Alan Stern
--
| Jul 3, 3:50 pm 2009 |
| Ramya Desai | Re: Creating URB in completion callback
Good. Now urb creation is working with these changes.
Thanks for help.
Regards,
RD
--
| Jul 3, 4:19 pm 2009 |
| Pierre Ossman | USB device not found on boot, only on insertion
I'm having some issues with a Philips IR Transceiver misbehaving if
it's inserted on boot. Removing it and reinserting it always gets it
running, but never when it is present during startup. It doesn't matter
if its a hot or cold boot or even a reboot.
I get this during bootup:
usb 2-2: new full speed USB device using ohci_hcd and address 2
usb 2-2: device descriptor read/64, error -62
usb 2-2: device descriptor read/64, error -62
usb 2-2: new full speed USB device using ohci_hcd and address 3
...
| Jul 3, 2:18 pm 2009 |
| David Brownell | Re: USB device not found on boot, only on insertion
As Alan noted -- not really. Maybe hook it up to an external
hub; most of those seem to claim they support power switching.
I did once have a system with an OHCI root hub which supported
power switching. Which was handy because one UPS needed to
start its enumeration from link-power-off, not unlike what you
are observing.
That system made a good regression test for the feature in the
Linux-USB stack whereby it tried to force root hub enumeration
to start from power-off. Since that system di...
| Jul 3, 4:01 pm 2009 |
| Alan Stern | Re: USB device not found on boot, only on insertion
If you have any external hubs that support port-power switching, they
should make equally good test cases. After all, Linux doesn't make
much distinction between root hub and external hub enumeration.
Alan Stern
--
| Jul 3, 5:19 pm 2009 |
| David Brownell | Re: USB device not found on boot, only on insertion
Not for OHCI root hub power switching support, though ... :)
--
| Jul 3, 5:34 pm 2009 |
| Alan Stern | Re: USB device not found on boot, only on insertion
Not on normal desktop and laptop PCs; the USB hardware doesn't provide
power control. However some hubs (not all!) do allow port power
control. See
http://www.gniibe.org/ac-power-by-usb/ac-power-control.html
Alan Stern
--
| Jul 3, 3:45 pm 2009 |
| Pierre Ossman | Re: USB device not found on boot, only on insertion
On Fri, 3 Jul 2009 15:45:07 -0400 (EDT)
Nice. None of the built-in hubs supported this though, and this
particular "feature" got me extra disappointed:
No overcurrent protection
And here I thought everything on a modern motherboard hade overcurrent
protection. :/
Luckily both two hubs I had lying around supported power switching
though (confirmed it using a optical mouse).
Now for the fun part. Once I put an extra hub between the host and the
IR device, it started working nicely on bo...
| Jul 3, 4:46 pm 2009 |
| Alan Stern | Re: USB device not found on boot, only on insertion
That isn't a reliable confirmation. When you tell a hub to turn off
power to a port, it will stop relaying packets through that port --
even if it leaves the power on. Without packets arriving every few
milliseconds, the optical mouse will go into suspend mode and turn off
Okay, good!
Alan Stern
--
| Jul 3, 5:04 pm 2009 |
| Oliver Neukum | [patch]stv680: kfree called before usb_kill_urb
The irq handler will touch memory. Even in the error case some URBs may
complete. Thus no memory must be kfreed before all URBs are killed.
Signed-off-by: Oliver Neukum <oliver@neukum.org>
--
commit e91d238d2b6f83f9b64b57b570ee150b1cd008e7
Author: Oliver Neukum <oneukum@linux-d698.(none)>
Date: Fri Jul 3 18:18:26 2009 +0200
stv680: fix access to freed memory in error case
in the error case some URBs may be active and access memory
URBs must be killed before a...
| Jul 3, 12:48 pm 2009 |
| Magnus Damm | [PATCH] usb: allow sh7724 to enable on-chip r8a66597
From: Magnus Damm <damm@igel.co.jp>
The sh7724 processor has two on-chip r8a66597 blocks, so add
it to the list of processors for SUPERH_ON_CHIP_R8A66597.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
---
Tested on a sh7724 Solution Engine.
And yes, to avoid this kind of patches in the future this Kconfig
option should be replaced with platform data. Will fix.
drivers/usb/host/Kconfig | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
--- 0001/drivers/usb/host/...
| Jul 3, 5:12 am 2009 |
| Ajay Kumar Gupta | [PATCH] musb: cleanup the nop registration for OMAP3EVM
OMAP3EVM uses ISP1504 phy which doesn't require any
programming and thus has to use NOP otg transceiver.
Cleanups being done:
- Remove unwanted code in usb-musb.c file
- Register nop in OMAP3EVM board file using
usb_nop_xceiv_register().
- Select NOP_USB_XCEIV for OMAP3EVM boards.
Signed-off-by: Ajay Kumar Gupta <ajay.gupta@ti.com>
---
This patch is refreshed based on David's recommendations at
[1] and [2].
[1] http://marc.info/?l=linux-omap&m=123904265231360&w=2
[2] ht...
| Jul 3, 5:00 am 2009 |
| David Brownell | Re: [PATCH] musb: cleanup the nop registration for OMAP3EVM
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
--
| Jul 3, 11:30 am 2009 |
| Michał Nazarewicz | USB Mass Storage + Ethernet Composite Gadget problems
Hello everyone,
I've been working on a USB Composite Gadget driver taken from Palm Pre
source code[1]. I managed to somehow port it to kernel 2.6.30 and it
works great if only mass storage is enabled. However, if I change it to
support both mass storage and RNDIS (which is our goal for now) Windows
has hard time detecting the device (ie. missing drivers). Linux on the
other hand, handles everything quite fine.
So now, first of all, does anyone know about any newer implementations of ...
| Jul 3, 3:54 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
Windows has built-in USB mass storage device driver. So it should just
work. You need an INF file for the RNDIS part.
Example:
http://www.mjmwired.net/kernel/Documentation/usb/linux.inf
But this is a bit different from the official RNDIS template inf file from
Microsoft.
http://msdn.microsoft.com/en-us/library/bb500930.aspx
I do not know why the above linux.inf file has section
related to legacy Windows 98/ME. And I do not think Windows NT
will support USB RNDIS either since it does not eve...
| Jul 3, 4:45 am 2009 |
| Michał Nazarewicz | Re: USB Mass Storage + Ethernet Composite Gadget problems
On Fri, 03 Jul 2009 10:45:55 +0200, Xiaofan Chen <xiaofanc@gmail.com>
Actually, I have a RNDIS driver installed. If I compile the device with
Ethernet Gadget for instance, I can communicate with Windows via the RNDIS
protocol. Also, If I compile the mass storage I can treat the device as
mass storage. Problem is, I cannot seem to make Windows handle a
composite with RNDIS and mass storage.
My guess is, neither mass storage nor RNDIS driver matches the device (as
it has both...
| Jul 3, 5:16 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
You may want to read this.
http://msdn.microsoft.com/en-us/library/ms793564.aspx
USB common class generic parent driver (usbccgp.sys) will
handle the enumeration for your composite device. If the interface
have windows built-in drivers (like HID, USB mass storage, etc),
they will work automatically. For the other interface, you need
the inf file to point to that interface (eg: USB\VID_0403&PID_6001&MI_00
means Interface 0 of device VID_0403&PID_6001;
VID_0403&PID_6001&MI_01 w...
| Jul 3, 5:23 am 2009 |
| Michał Nazarewicz | Re: USB Mass Storage + Ethernet Composite Gadget problems
On Fri, 03 Jul 2009 11:23:31 +0200, Xiaofan Chen <xiaofanc@gmail.com>
Currently, only "USB Mass Storage Device" is present. Properties window
shows a "This device cannot start. (Code 10)" error message. No volume
(either in My Computer or Computer Management->Disk Management) is shown.
No entries regarding the RNDIS functionality or USB Composite. Do I need
to install the "USB Common Class Generic Parent Driver"? I'll try to read
the link you've send -- maybe it'll help.
...
| Jul 3, 5:54 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
In that case, your device firmware is not working. You should not need to
install the parent driver as Windows should be able to handle it
automatically.
What is your descriptor? Please post the output of "lsusb -vvv" under
Linux if your device works under Linux.
I can not remember exactly whether IAD is necessary for USB composite device
involving RNDIS . But you may want to use IAD (interface association
descriptor).
At least IAD is the recommended method to get USB composite device involvin...
| Jul 3, 7:12 am 2009 |
| Michał Nazarewicz | Re: USB Mass Storage + Ethernet Composite Gadget problems
On Fri, 03 Jul 2009 13:12:41 +0200, Xiaofan Chen <xiaofanc@gmail.com>
Thanks again for useful links.
Bus 002 Device 066: ID 0830:0102 Palm, Inc.
Device Descriptor:
bLength 18
bDescriptorType 1
bcdUSB 2.00
bDeviceClass 0 (Defined at Interface level)
bDeviceSubClass 0
bDeviceProtocol 0
bMaxPacketSize0 64
idVendor 0x0830 Palm, Inc.
idProduct 0x0102
bcdDevice ...
| Jul 3, 8:04 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
So it is already using IAD.
Do you have the real Palm Pre and what is the Windows driver for it?
You may want to check that out. Are they using the USB RNDIS driver
or custom driver?
By the way, to debug Code 10 error, you can try to check the
log file setupapi.log.
http://www.microsoft.com/whdc/driver/install/setupapilog.mspx
http://www.microsoft.com/whdc/driver/install/diagnose.mspx
--
Xiaofan http://mcuee.blogspot.com
--
| Jul 3, 8:52 am 2009 |
| Marek Szyprowski | RE: USB Mass Storage + Ethernet Composite Gadget problems
Hello,
Michal is working with me on that composite gadget, he already went home,
so let me answer these questions.
Unfortunately no - we don't have real Palm and original Windows drivers for it.
We only adapted Palm open source kernel usb gadget driver for our ARM S3C6410
They are using standard RNDIS driver, it is well detected under Linux and
works there fine with 'rndis_host' driver.
Best regards
--
Marek Szyprowski
Samsung Poland R&D Center
--
| Jul 3, 9:26 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
On Fri, Jul 3, 2009 at 9:26 PM, Marek
Hmm, you should at least get the driver packages of Palm Pre and try that.
Anyway, could you post the inf file you are using for the RNDIS part?
You may need slightly different inf file for the rndis part when it is
part of the composite device than when it is a single interface RNDIS
device.
From what I read, there are three mode for the Palm Pre USB. One is
the "USB drive mode", the other is the USB charger mode, the last one
is the media sync mode (iPod...
| Jul 3, 10:00 am 2009 |
| Marek Szyprowski | RE: USB Mass Storage + Ethernet Composite Gadget problems
Hello,
I just did one more experiment. I've selected to install the driver manually in
Windows wizard and I seleted the standard linux.inf for Linux RNDIS Ethernet gadget
(from Documentation/usb/linux.inf, we used it earlier for the standard g_ether usb
gadget module in RNDIS mode, it is available for example here:
http://beagleboard.googlecode.com/files/linux.inf ).
I was really surprised, the driver installed correctly and RNDIS connection started
working! So the real problem is to create...
| Jul 3, 10:33 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
On Fri, Jul 3, 2009 at 10:33 PM, Marek
So it works, right? Glad to hear that.
Take note you need an INF file for the RNDIS device. Windows has the built-in
core driver files for the USB RNDIS device, but you need to provide the
INF file. The same is for the CDC-ACM device, Windows has the built-in
usbser.sys driver, but you still need an INF file to install the driver.
By the way, I believe that INF file is flawed. Firstly it will not
work for 64bit
Windows.
You may want to try out the inf ...
| Jul 3, 10:49 am 2009 |
| Xiaofan Chen | Re: USB Mass Storage + Ethernet Composite Gadget problems
From this page, it does not seem to use RNDIS at all under Windows.
http://kb.palm.com/wps/portal/kb/common/article/50331_en.html
"Windows: **If the Found New Hardware wizard opens, click Cancel. **
When prompted, select Open folder to view files using Windows Explorer."
It seems that it only uses the USB mass storage mode.
Not so sure if this helps. It seems to me that you need to enable
usbnet by yourself.
http://predev.wikidot.com/usbnet-setup
But the best is to get the real USB driver an...
| Jul 3, 10:26 am 2009 |
| Felipe Balbi | Re: USB Mass Storage + Ethernet Composite Gadget problems
Hi,
I guess you just one of those *.inf file telling windows which drivers
you need to tell the driver which backing storage to use:
# modprobe g_file_storage file=/dev/anything-goes-here
should do it.
--
balbi
--
| Jul 3, 4:09 am 2009 |
| Michał Nazarewicz | Re: USB Mass Storage + Ethernet Composite Gadget problems
Exactly, however, as we are trying to create a custom made composite
gadget which will support several functions, we probably won't be able to
use any existing drivers. This is why, I'm wondering whether creating a
Windows driver (ie. inf file) for a multi-function device is an easy
(copy-and-paste preferable) task. Note that we are not developing any new
protocol or anything -- just
We tried to use some vendor and product IDs that the Palm composite gadget
had hardcoded hoping Wi...
| Jul 3, 4:48 am 2009 |
| Ravi, Babu | [PATCH] usb: otg: fix module reinsert issue
From: Ajay Kumar Gupta <ajay.gupta@ti.com>
Platform_device instance (pd) is not set to NULL in
usb_nop_xceiv_unregister() causing usb_nop_xceiv_register()
to fail during module reinsert.
Signed-off-by: Ravi, Babu <ravibabu@ti.com>
---
Submitting on behalf of Ravi Babu.
drivers/usb/otg/nop-usb-xceiv.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/usb/otg/nop-usb-xceiv.c b/drivers/usb/otg/nop-usb-xceiv.c
index 9ed5ea5..af456b4 100644
--- a/driv...
| Jul 3, 3:48 am 2009 |
| David Brownell | Jul 3, 11:27 am 2009 | |
| Yang Fei-AFY095 | using cacheable buffer for descriptor
Hello All,
I'm working on a TI OMAP based device running linux kernel 2.6.29. I
encountered a problem with the USB host driver and would like to get
some advice on a safe solution.
The problem happens within the following code of function "int
usb_get_configuration(struct usb_device *dev)" in
drivers/usb/core/config.c
------ snip ------
/* Now that we know the length, get the whole thing */
bigbuffer = kmalloc(length, GFP_KERNEL);
if (!bigbuffer) {
result = -ENOMEM;
goto ...
| Jul 3, 12:43 am 2009 |
| Alan Stern | Re: using cacheable buffer for descriptor
For both IN and OUT transfers, cache consistency is handled by
map_urb_for_dma and unmap_urb_for_dma. Those two functions are in
You can disable DMA for ep0 if you want. It shouldn't make much
difference.
But if you're getting these errors already, maybe you'll get similar
errors with transfers to other endpoints. So it might be better to fix
the problem than to try working around it.
Alan Stern
--
| Jul 3, 5:16 pm 2009 |
| Yang Fei-AFY095 | RE: using cacheable buffer for descriptor
In drivers/usb/core/hcd.c, usb_hcd_submit_urb -> map_urb_for_dma ->
dma_map_single -> dma_cache_maint.
dma_map_single is defined in arch/arm/include/asm/dma-mapping.h
This is called when we submit urb, for OUT data, I think it's sufficient
for cache consistency. But what I need to know is how we guarantee the
cache consistency for IN data, because the DMA is going to update the
physical memory when receiving data from device, something has to be
done upon completion of the urb.
Re...
| Jul 3, 6:06 pm 2009 |
| Alan Stern | RE: using cacheable buffer for descriptor
I would have said that dma_map_single, not usb_submit_urb, calls
Obviously you looked at map_urb_for_dma. Have you also looked at
unmap_urb_for_dma?
Alan Stern
--
| Jul 3, 7:19 pm 2009 |
| Jamie Lokier | Re: 2.6.27- Sending uevent from a driver
Ah yes, the undocumented ioctl... I knew there was one, just couldn't
find it in the man page just now.
-- Jamie
--
| Jul 3, 11:36 am 2009 |
| David Brownell | Jul 2, 8:27 pm 2009 | |
| Alan Stern | Re: patch usb-cdc-acm-work-around-some-broken-devices.patch ...
Arseniy, shouldn't you simplify this?
ep = acm->dev->ep_in[usb_pipeendpoint(acm->rx_endpoint)];
After all, acm->rx_endpoint has to be an IN endpoint.
Alan Stern
--
| Jul 3, 5:07 pm 2009 |
| Arseniy Lartsev | Re: patch usb-cdc-acm-work-around-some-broken-devices.patch ...
I've assumed that such changes should be made by someone more experienced t=
han=20
me. I don't know how USB works and what an "endpoint" is.
| Jul 3, 5:28 pm 2009 |
| Alan Stern | Re: patch usb-cdc-acm-work-around-some-broken-devices.patch ...
Okay. Oliver, do you want to make this change? Or should we ask Greg
to just modify the existing patch?
Alan Stern
--
| Jul 3, 5:34 pm 2009 |
| Oliver Neukum | Re: patch usb-cdc-acm-work-around-some-broken-devices.patch ...
I'll make it.
Regards
Oliver
--
| Jul 3, 6:04 pm 2009 |
| Steve Calfee | Re: possible memleak in devio.c::processcompl
Hi,
Yes, I encountered this too. I hit it on SLED 11. This is fixed in
later kernels. search this list about this leak, it is for iso urbs. I
found http://lkml.indiana.edu/hypermail/linux/kernel/0903.2/02471.html
It is fixed in the current kernel and Novell has patches for SLED 11.
Regards, Steve
--
| Jul 2, 9:41 pm 2009 |
| Steve Calfee | Re: possible memleak in devio.c::processcompl
One other thing. uvcvideo (cheese) does not continue to leak but my
userspace stuff did. I believe that is because uvcvideo uses an urb
pool, so reuses the urbs. I could not, but maybe you can in your app
and that may be a work-around for this leak consuming all memory. This
might be useful if you cannot upgrade your kernel beyond 2.6.27....
Regards, Steve
--
| Jul 2, 11:54 pm 2009 |
| Yoshihiro Shimoda | Re: [PATCH] usb: m66592-udc buffer management update
I'm sorry. I forgot to write the Acked-by:
Acked-by: Yoshihiro Shimoda <shimoda.yoshihiro@renesas.com>
Thanks,
Yoshihiro Shimoda
--
| Jul 3, 4:34 am 2009 |
| Paul Mundt | Re: [PATCH] usb: m66592-udc buffer management update
Greg, any preferences? I was planning on just rolling this in to my
2.6.32 queue unless you have any objections.
--
| Jul 3, 4:37 am 2009 |
| previous day | today | next day |
|---|---|---|
| July 2, 2009 | July 3, 2009 | July 4, 2009 |
| Greg Kroah-Hartman | [PATCH 001/196] Chinese: Add the known_regression URI to the HOWTO |
| Linus Torvalds | Re: Dual-Licensing Linux Kernel with GPL V2 and GPL V3 |
| David Newall | Re: Slow DOWN, please!!! |
| Ian Campbell | Re: [PATCH] x86: Construct 32 bit boot time page tables in native format. |
| Matthias Scheler | Re: HEADS UP: timecounters (branch simonb-timecounters) merged into -current |
| Greg Troxel | Re: Interface to change NFS exports |
| Thor Lancelot Simon | metadata cache and memory fragmentation |
| YAMAMOTO Takashi | amap memory allocation |
git: | |
| Jarek Poplawski | [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
| Gerrit Renker | [PATCH 0/37] dccp: Feature negotiation - last call for comments |
| David Miller | [GIT]: Networking |
| Dushan Tcholich | Re: ksoftirqd high cpu load on kernels 2.6.24 to 2.6.27-rc1-mm1 |
