Still no luck with E220

viking777 · Post by **viking777** » 25 Jan 2010, 15:11

Hello, first I would like to say that I really appreciate the immense effort that the people creating this program have put in over the years and particularly recently with two versions in one week! Especially when it is more or less forced upon you by kernels that dont work properly!

However despite that heroic effort I am still not able to get my Huawei E220 working on either Mint 8 or Ubuntu 9.10. My problems are twofold first, every time I try to compile the program I get this:

Code: Select all

/o/usb-modeswitch-1.1.0# make install
gcc -Wall -l usb -o usb_modeswitch usb_modeswitch.c
usb_modeswitch.c:66:17: error: usb.h: No such file or directory
usb_modeswitch.c: In function ‘main’:
usb_modeswitch.c:343: warning: implicit declaration of function ‘usb_init’
usb_modeswitch.c:346: warning: implicit declaration of function ‘usb_set_debug’
usb_modeswitch.c:348: warning: implicit declaration of function ‘usb_find_busses’
usb_modeswitch.c:349: warning: implicit declaration of function ‘usb_find_devices’
usb_modeswitch.c:397: error: dereferencing pointer to incomplete type
usb_modeswitch.c:398: error: dereferencing pointer to incomplete type
usb_modeswitch.c:400: warning: implicit declaration of function ‘usb_open’
usb_modeswitch.c:400: warning: assignment makes pointer from integer without a cast
usb_modeswitch.c:407: error: dereferencing pointer to incomplete type
usb_modeswitch.c:409: error: dereferencing pointer to incomplete type
usb_modeswitch.c:411: error: dereferencing pointer to incomplete type
usb_modeswitch.c:561: warning: implicit declaration of function ‘usb_close’
usb_modeswitch.c: In function ‘deviceDescription’:
usb_modeswitch.c:576: error: dereferencing pointer to incomplete type
usb_modeswitch.c:577: warning: implicit declaration of function ‘usb_get_string_simple’
usb_modeswitch.c:577: error: dereferencing pointer to incomplete type
usb_modeswitch.c:586: error: dereferencing pointer to incomplete type
usb_modeswitch.c:587: error: dereferencing pointer to incomplete type
usb_modeswitch.c:596: error: dereferencing pointer to incomplete type
usb_modeswitch.c:597: error: dereferencing pointer to incomplete type
usb_modeswitch.c: In function ‘deviceInquire’:
usb_modeswitch.c:629: warning: implicit declaration of function ‘usb_claim_interface’
usb_modeswitch.c:634: warning: implicit declaration of function ‘usb_clear_halt’
usb_modeswitch.c:636: warning: implicit declaration of function ‘usb_bulk_write’
usb_modeswitch.c:642: warning: implicit declaration of function ‘usb_bulk_read’
usb_modeswitch.c:669: warning: implicit declaration of function ‘usb_release_interface’
usb_modeswitch.c: In function ‘resetUSB’:
usb_modeswitch.c:685: warning: implicit declaration of function ‘sleep’
usb_modeswitch.c:687: warning: implicit declaration of function ‘usb_reset’
usb_modeswitch.c: In function ‘switchConfiguration’:
usb_modeswitch.c:753: warning: implicit declaration of function ‘usb_set_configuration’
usb_modeswitch.c: In function ‘switchAltSetting’:
usb_modeswitch.c:769: warning: implicit declaration of function ‘usb_set_altinterface’
usb_modeswitch.c: In function ‘switchHuaweiMode’:
usb_modeswitch.c:786: warning: implicit declaration of function ‘usb_control_msg’
usb_modeswitch.c:786: error: ‘USB_TYPE_STANDARD’ undeclared (first use in this function)
usb_modeswitch.c:786: error: (Each undeclared identifier is reported only once
usb_modeswitch.c:786: error: for each function it appears in.)
usb_modeswitch.c:786: error: ‘USB_RECIP_DEVICE’ undeclared (first use in this function)
usb_modeswitch.c:786: error: ‘USB_REQ_SET_FEATURE’ undeclared (first use in this function)
usb_modeswitch.c: In function ‘switchSonyMode’:
usb_modeswitch.c:872: warning: assignment makes pointer from integer without a cast
usb_modeswitch.c: In function ‘detachDriver’:
usb_modeswitch.c:909: warning: implicit declaration of function ‘usb_get_driver_np’
usb_modeswitch.c:926: warning: implicit declaration of function ‘usb_detach_kernel_driver_np’
usb_modeswitch.c: In function ‘checkSuccess’:
usb_modeswitch.c:982: warning: assignment makes pointer from integer without a cast
usb_modeswitch.c:987: error: dereferencing pointer to incomplete type
usb_modeswitch.c:987: error: dereferencing pointer to incomplete type
usb_modeswitch.c: In function ‘search_devices’:
usb_modeswitch.c:1093: warning: implicit declaration of function ‘usb_get_busses’
usb_modeswitch.c:1093: warning: assignment makes pointer from integer without a cast
usb_modeswitch.c:1093: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1095: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1095: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1097: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1097: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1098: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1119: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1126: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1126: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1128: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1138: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1142: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1144: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1146: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1147: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1153: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1153: error: dereferencing pointer to incomplete type
usb_modeswitch.c: In function ‘find_first_bulk_output_endpoint’:
usb_modeswitch.c:1176: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1179: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1180: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1181: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1181: error: ‘USB_ENDPOINT_TYPE_MASK’ undeclared (first use in this function)
usb_modeswitch.c:1181: error: ‘USB_ENDPOINT_TYPE_BULK’ undeclared (first use in this function)
usb_modeswitch.c:1182: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1182: error: ‘USB_ENDPOINT_DIR_MASK’ undeclared (first use in this function)
usb_modeswitch.c:1183: error: dereferencing pointer to incomplete type
usb_modeswitch.c: In function ‘find_first_bulk_input_endpoint’:
usb_modeswitch.c:1194: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1197: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1198: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1199: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1199: error: ‘USB_ENDPOINT_TYPE_MASK’ undeclared (first use in this function)
usb_modeswitch.c:1199: error: ‘USB_ENDPOINT_TYPE_BULK’ undeclared (first use in this function)
usb_modeswitch.c:1200: error: dereferencing pointer to incomplete type
usb_modeswitch.c:1200: error: ‘USB_ENDPOINT_DIR_MASK’ undeclared (first use in this function)
usb_modeswitch.c:1201: error: dereferencing pointer to incomplete type
make: *** [usb_modeswitch] Error 1

This holds true for both the 1.0.7 and 1.1.0 versions. I don't know what it means, but it certainly doesn't work! I have the latest Ubuntu 'build-essential' program installed so I don't think I lack any compilers.

Secondly the .deb file for 1.0.7 on the debian package archive will not install either (I haven't found one for 1.1.0 yet). I install the usb-modeswitch-data file first -

usb-modeswitch-data_20100110-1_all.deb

then when I try to install usb-modeswitch itself it complains that the -data file conflicts with it and refuses to install! I tried a forced install but this resulted in none of my modems working (I have 3), not just the E220. Installing usb-modeswitch first just results in an 'unsatisfiable dependency' error.

Both Mint8 and Ubuntu 9.10 are using the 2.6.31-17 kernel but my E220 hasn't worked unsupported since the 2.6.28 kernel although it does work with modeswitch 1.0.7 on Sidux (2.6.32-5)

In case it is relevant, using modeswitch 1.0.2 or 1.0.5 on the .31 kernel results in a continuous loop of switching between modem mode and storage mode, about 60 seconds as a modem followed by 3 seconds as a storage device. I know 3 seconds is not long, but it is long enough to stop it functioning as a modem.

Once again thank you for your immensely valuable work and sorry I have to add to it!

sakis · Post by **sakis** » 25 Jan 2010, 15:28

When it comes into compiling from tarball found in this site, you need to install "libusb-dev" along with build-essential.

When it comes to complaining that conflicts with -data, I guess that usb-modeswitch package also provides usb-modeswitch-data. If you choose to use the debian package, uninstall -data package and only install program.

Sakis

viking777 · Post by **viking777** » 25 Jan 2010, 17:26

sakis wrote:When it comes into compiling from tarball found in this site, you need to install "libusb-dev" along with build-essential.

When it comes to complaining that conflicts with -data, I guess that usb-modeswitch package also provides usb-modeswitch-data. If you choose to use the debian package, uninstall -data package and only install program.

Sakis

Thanks Sakis, that solved one problem at least, - I was missing libusb-dev.

After installing that it compiled straight away (version 1.1.0). Sadly though it made absolutely no difference, I am still stuck in the endless loop of about 60 seconds as a modem followed by approximately three seconds as a storage device. This is a typical dmesg output:

70.328091] usb 6-2: new full speed USB device using uhci_hcd and address 5
[ 70.484405] usb 6-2: configuration #1 chosen from 1 choice
[ 70.490296] scsi6 : SCSI emulation for USB Mass Storage devices
[ 70.490548] usb-storage: device found at 5
[ 70.490550] usb-storage: waiting for device to settle before scanning
[ 70.585093] usb 6-2: USB disconnect, address 5
[ 71.320073] usb 6-2: new full speed USB device using uhci_hcd and address 6
[ 71.476871] usb 6-2: configuration #1 chosen from 1 choice
[ 71.485769] scsi9 : SCSI emulation for USB Mass Storage devices
[ 71.486837] usb-storage: device found at 6
[ 71.486839] usb-storage: waiting for device to settle before scanning
[ 71.550335] usbcore: registered new interface driver usbserial
[ 71.550351] USB Serial support registered for generic
[ 71.550391] usbcore: registered new interface driver usbserial_generic
[ 71.550393] usbserial: USB Serial Driver core
[ 71.554737] USB Serial support registered for GSM modem (1-port)
[ 71.554790] option 6-2:1.0: GSM modem (1-port) converter detected
[ 71.554871] usb 6-2: GSM modem (1-port) converter now attached to ttyUSB0
[ 71.554881] option 6-2:1.1: GSM modem (1-port) converter detected
[ 71.554927] usb 6-2: GSM modem (1-port) converter now attached to ttyUSB1
[ 71.554944] usbcore: registered new interface driver option
[ 71.554946] option: v0.7.2:USB Driver for GSM modems
[ 76.485350] usb-storage: device scan complete
[ 76.488292] scsi 9:0:0:0: CD-ROM HUAWEI Mass Storage 2.31 PQ: 0 ANSI: 2
[ 76.511293] sr1: scsi-1 drive
[ 76.511459] sr 9:0:0:0: Attached scsi CD-ROM sr1
[ 76.511547] sr 9:0:0:0: Attached scsi generic sg3 type 5
[ 88.670503] sr1: CDROM (ioctl) error, command: Get configuration 46 00 00 00 00 00 00 00 20 00
[ 88.670536] sr: Sense Key : Hardware Error [current]
[ 88.670546] sr: Add. Sense: No additional sense information
[ 88.918628] sr1: CDROM (ioctl) error, command: Get configuration 46 00 00 00 00 00 00 00 20 00
[ 88.918661] sr: Sense Key : Hardware Error [current]
[ 88.918671] sr: Add. Sense: No additional sense information
[ 89.038500] option: option_instat_callback: error -108
[ 89.038659] option1 ttyUSB0: GSM modem (1-port) converter now disconnected from ttyUSB0
[ 89.038691] option 6-2:1.0: device disconnected
[ 89.038800] option1 ttyUSB1: GSM modem (1-port) converter now disconnected from ttyUSB1
[ 89.038829] option 6-2:1.1: device disconnected
[ 89.148139] usb 6-2: reset full speed USB device using uhci_hcd and address 6
[ 89.291046] option 6-2:1.1: GSM modem (1-port) converter detected
[ 89.291220] usb 6-2: GSM modem (1-port) converter now attached to ttyUSB0
[ 89.292876] option 6-2:1.0: GSM modem (1-port) converter detected
[ 89.293022] usb 6-2: GSM modem (1-port) converter now attached to ttyUSB1
[ 89.690929] option: option_instat_callback: error -108
[ 89.692231] option1 ttyUSB1: GSM modem (1-port) converter now disconnected from ttyUSB1
[ 89.692282] option 6-2:1.0: device disconnected
[ 89.692445] option1 ttyUSB0: GSM modem (1-port) converter now disconnected from ttyUSB0
[ 89.692487] option 6-2:1.1: device disconnected
[ 89.809126] usb 6-2: reset full speed USB device using uhci_hcd and address 6
[ 89.932221] usb 6-2: device descriptor read/64, error -71
[ 90.156134] usb 6-2: device descriptor read/64, error -71
[ 90.372136] usb 6-2: reset full speed USB device using uhci_hcd and address 6
[ 90.493188] usb 6-2: device descriptor read/64, error -71
[ 90.716355] usb 6-2: device descriptor read/64, error -71
[ 90.932143] usb 6-2: reset full speed USB device using uhci_hcd and address 6
[ 91.340145] usb 6-2: device not accepting address 6, error -71
[ 91.452155] usb 6-2: reset full speed USB device using uhci_hcd and address 6
[ 91.860113] usb 6-2: device not accepting address 6, error -71
[ 91.860222] usb 6-2: USB disconnect, address 6
[ 91.976163] usb 6-2: new full speed USB device using uhci_hcd and address 7
[ 92.096131] usb 6-2: device descriptor read/64, error -71
[ 92.321177] usb 6-2: device descriptor read/64, error -71
[ 92.536127] usb 6-2: new full speed USB device using uhci_hcd and address 8
[ 92.656181] usb 6-2: device descriptor read/64, error -71
[ 92.880143] usb 6-2: device descriptor read/64, error -71
[ 93.096118] usb 6-2: new full speed USB device using uhci_hcd and address 9
[ 93.504147] usb 6-2: device not accepting address 9, error -71
[ 93.616280] usb 6-2: new full speed USB device using uhci_hcd and address 10
[ 94.028108] usb 6-2: device not accepting address 10, error -71
[ 94.028146] hub 6-0:1.0: unable to enumerate USB device on port 2

It repeats endlessly until you unplug it.

To explain about the .deb installation, if I uninstall the -data package and then try to install the usb-modeswitch package it fails due to missing dependency usb-modeswitch-data! So that doesn't work either. Maybe it has been packaged wrongly? Anyway it doesn't matter much as I have it installed now from the tarball and it still doesn't solve my problem.

Post by **Josh** » 25 Jan 2010, 21:30

There is switching code for the E220 in the kernel. Also the "option" driver knows this device when switched. You probably see an interference with the "userspace" switching attempts.

You should not need any of the packages. In the long term, the switching code will be removed from the kernel. But this will happen only if adequate switching is provided by the distributions.

viking777 · Post by **viking777** » 26 Jan 2010, 11:40

It did strike me that there might be 'interference' between modeswitch and the kernel, but believe me I have tried it every way I can - with modeswitch installed, without modeswitch installed and with modeswitch installed but not configured in the .conf file and none of it makes any difference, I am still stuck in this endless loop. The only thing that saves me is that the E220 is not the only modem I own, my ZTE 622 and my Huawei K3565 both work.

However on other forums I am trying to assist people who are not so lucky and are tearing out their hair because the E220 is the only modem they have, it used to work and now it no longer does. I know of several cases where users have downgraded their distros just to get this modem working again as it usually works on anything between .26 and .28 kernels.

There has to be an answer somewhere.

Post by **Josh** » 26 Jan 2010, 15:47

And there IS an answer. I was slow to catch up ...

This is the notorious "BAD SENSE" bug. It affects the storage part of switched devices, causing a reset and a "switch loop". See this and this.

Instant remedy is a "blacklist usb-storage" in one of modprobe's conf files. Or a very recent kernel.

viking777 · Post by **viking777** » 26 Jan 2010, 18:22

Thanks for the reply Josh - at least I know the problem is being worked on. I added my own comments to the end of the launchpad thread. I am not well up on adding kernel patches so I will chicken out on that one.

Your comment on very recent kernels may tie in with the fact that my E220 does work on Sidux (2.6.32.6 now) although it still needs usb-modeswitch to make it work (I have version 1.0.7 on Sidux).

I realise that blacklisting the usb-storage module works, and has done for a long time, but let me ask you, how many usb storage devices do you use these days and could you do without all of them?

Anyway thanks for the help and please keep up the good work. I have been using these devices for some time and 'modeswitch' has (up till now - and I know the present situation is not your fault) made my life a whole lot easier.

If you ever need anything testing, I would be more than willing to help. I have 3 modems already and I am buying a fourth tomorrow. Of course if you accept the offer you also have to accept that, in the Linux world, I very rarely know exactly what I am talking about

Post by **Josh** » 28 Jan 2010, 21:36

viking777 wrote:If you ever need anything testing, I would be more than willing to help. I have 3 modems already and I am buying a fourth tomorrow. Of course if you accept the offer you also have to accept that, in the Linux world, I very rarely know exactly what I am talking about

This is the kind of courage I sort of expect from a true viking.

My first questions are not hard to guess: what modem models do you have at your hand, and do they work with version 1.1.0 out of the box?

viking777 · Post by **viking777** » 31 Jan 2010, 20:36

Josh wrote:
viking777 wrote:If you ever need anything testing, I would be more than willing to help. I have 3 modems already and I am buying a fourth tomorrow. Of course if you accept the offer you also have to accept that, in the Linux world, I very rarely know exactly what I am talking about
This is the kind of courage I sort of expect from a true viking.

My first questions are not hard to guess: what modem models do you have at your hand, and do they work with version 1.1.0 out of the box?

Huawei E220, ZTE MF622, Huawei K3565 and now Huawei E5830.

The answer to your second question is far too difficult to answer as I have a lot of different OS's and kernels.

For the full story read this:
http://forums.linuxmint.com/viewtopic.p ... 52#p214276

Although even that is not the full story because it hasn't been updated for the E5830, but that is easy, in wifi mode it works anywhere and everywhere (in fact, if they catch on you will be out of work

). But as a plugged modem they don't work at all. Read this:
http://forums.linuxmint.com/viewtopic.p ... 54#p234993

I haven't upgraded all my distros to modeswitch version 1.1 yet.

viking777 · Post by **viking777** » 01 Feb 2010, 13:15

Josh, on the 28th Jan. you asked me about the performance of modeswitch 1.1. I wasn't able to answer it fully then as I did not have it installed everywhere. I now do have it installed, at least everywhere that needs it. The answer to your question follows these notes.

First I should say that where I use the term 'works' or any of its derivatives in regard to a device, this means that the distro concerned creates the device nodes needed for connection, not that it actually connects, this is particularly important with regard to the E5830 which connects to absolutely nothing (except Windows) in 'tethered' mode, but nearly always produces the device nodes required to make a connection.

Second I have not installed modeswitch of any variety onto PCLOS or Mint7 as the 2.6.26/28 kernel on these distros handle all switching and all the devices work.

Mint 8/Ubuntu9.10 (they both behave in the same manner) - kernel 2.6.31.17 - ZTE, K3565, E5830 work, E220 doesn't.

Sidux - kernel 2.6.32.7- All modems work.

Fedora 12 - kernel 2.6.31.5 - K3565 works, all others fail.

viking777 · Post by **viking777** » 10 Mar 2010, 11:57

As I started this thread I thought you may be interested in some further information that I have learned. The continual cycling of the E220 modem on my machine is, I believe, the fault of Gnome, more specifically, Nautilus.

I received some information that another E220 user with the same problem had cured the fault by switching to KDE. I don't like KDE4 so I decided to try Fluxbox. Firstly I installed it as a standalone distro (Mint Fluxbox CE) - my modem worked perfectly. Then I installed it as an additional window manager on top of my Mint Gnome installation. Logging in to a Fluxbox session, the modem worked perfectly. However, as soon as I started Nautilus my modem instantly stopped working and reverted to the 'cycling' behaviour. I rebooted, logged in again to a Fluxbox session then started Nautilus but this time with the '--no-desktop' option - my modem continued to be fully functional.

So I believe what is happening (although this is now more guesswork on my part) is that as soon as Nautilus starts to manage the desktop it polls for new devices and discovers the E220 storage module and lists it in the 'Places' section of the Nautilus window ( or mounts it if you have automount enabled). As soon as it does this the modem function is lost. Usb-modeswitch then kicks in and unloads the storage module allowing it to revert to modem usage, but only until the next time Nautilus polls for new devices again. Hence the cycling is 'eternal' and the modem unusable. The only thing I can't figure out is why this doesn't happen with other modems, only the E220?

So my advice to anyone having this problem with an E220 is to switch to another desktop environment or window manager and the problem will go away.

It is only a workround I guess not a solution, but the solution will be down to the Gnome/Nautilus developers.

Post by **Josh** » 10 Mar 2010, 14:27

The BAD_SENSE bug (see post and links above) is caused by only one culprit: Huawei.
They didn't adhere to the USB specification, and when the kernel became a bit stricter about the standards, the error was exposed.

As you noticed correctly, this bug (causing a reset of the modem) will be relevant as soon as the storage device is mounted. This is often done automatically by the major desktop environments; it's perfectly legal, and not their fault if the device crashes.

I would assume that Gnome or KDE do have a switch somewhere to disable the automounting feature.
Other than that, a kernel update will most likely help.

viking777 · Post by **viking777** » 10 Mar 2010, 19:38

Thanks for the reply Josh. I am very reluctant to disagree with you as I am sure you know a lot more about this subject than I do, I only look at it from a users viewpoint, not a programmers or developers, but I feel I should ask you this, if it is not the fault of any desktop environment and only Hauwei's BAD_SENSE bug, then why does that bug not affect either KDE or Fluxbox or Xfce but only Gnome?

I may have misled you in the previous post when I used the word 'automount'. I regard the automounting of usb devices as the work of the devil, and I never, ever allow it to happen on any of my distros, so that has nothing to do with it.

I have upgraded my kernel. A little while ago I took the very latest Debian Sid kernel and applied it to my Mint Gnome installation. It made no difference at all to the behaviour of the E220, the only thing that does make a difference is changing window managers. I can't tell you why, but I know it works.

Post by **Josh** » 11 Mar 2010, 09:39

Hmm, the fix was introduced in 2.6.32.2, so with all kernels above this bug should be gone ... What's your kernel version ?

Anyway, Gnome seems to do a kind of automounting, otherwise it would not be able to show the device in that "Places" list, right?
I'm not even sure if mounting is required to expose the bug, any data request from the storage device may be enough. But I'm not a hardware specialist either.

viking777 · Post by **viking777** » 18 Mar 2010, 12:40

Just installed modeswitch version 1.1.1, and as expected( by me anyway) it doesn't make any difference to my E220 behaviour, it still cycles. In answer to your previous question I am using kernel 2.6.32-13 so that is not it either.

It is a bit of a moot point at the moment, because everything works as long as I don't use the Gnome desktop, and as I am really enjoying using Fluxbox right now the problem no longer bothers me, but I thought you might like to know the results anyway.

Something else just crossed my mind which may be of interest too. I recently installed the fluxbox wm on top of my Mint gnome installation, so I have a system which uses fluxbox desktop but nautilus file manager. On this setup the E220 will not work.

However on a pure fluxbox system (mint fluxbox ce), which uses Thunar as a file manager the E220 works perfectly.

It seems to suggest a culprit to me.