dabusb/usb modems conflict issue
Hi hotplug folks,
I'm writing here in the name of the developer team of the ECI USB ADSL driver, and I'd like to report there a possible conflict issue, or at least a few questions.
Well.. our driver support a lot of GlobeSpan chipset-based USB ADSL modems, and in most cases, hotplug loads dabusb module at startup to add support for the (modem) devices.
We used to disable dabusb module load by removing the module file or re-configuring the kernel to skip dabusb support, and with latest versions of hotplug (depending on the GNU/Linux distro), we recommend our users to add dabusb to the blacklist file.
You will find related info about how we manage this issue there:
the project page..
http://eciadsl.flashtux.org/index.php?lang=en
the supported modems list (including vid/pid)..
http://eciadsl.flashtux.org/modems.php?lang=en
our Q/A in the FAQ related to this dabusb issue..
http://eciadsl.flashtux.org/faq.php?lang=en#2.1
My questions:
1/
Do we use the right method to avoid dabusb loading for non-audio-dabusb devices (modems).
2/
Is this really a conflict? What's your thought about this?
3/
Do you have any piece of advice related to this (generally)?
Thanx in advance,
regards
Signature · 11 lines
wwp ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net [email protected] https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
Hi,
Quoted 3 lines
Well.. our driver support a lot of GlobeSpan chipset-based USB ADSL modems, and in most cases, hotplug loads dabusb module at startup to add support for the (modem) devices.
The URLs you mentioned don't seem available to me just now, so I can't see any more information than is found in your email message ...
EVERY version of "dabusb" I've ever seen has a bug in its MODULE_DEVICE_TABLE entries: they claim the default EZ-USB product/vendor IDs. Those IDs are not supposed to be used by _any_ product except Cypress devkits. (Which means any developers working with Cypress EZ-USB and Linux trip over the bug...)
In fact, the current drivers (2.4, 2.5) seem to have only two IDs, both with Cypress vendor IDs. One of them is a bug, and needs to be removed. I don't know about the other one; it could be legit.
Assuming this isn't a hotplug bug, there are exactly three possibilities:
- This GlobeSpan driver is also trying to bind to default EZ-USB devices. **BOTH** drivers have bugs, since no production devices may use those IDs.
- Or the globespan driver is also trying to bind to the other (pid 9999) device entry. Assuming that ID was allocated legally only one of the drivers can possibly be correct.
- Both drivers are using VID/PID entries they aren't allowed to use, for the second entry (the one that's not the EZ-USB default). For example, that ID may not have been allocated to either one by Cypress.
Do you know which is the situation?
Quoted 23 lines
We used to disable dabusb module load by removing the module file or re-configuring the kernel to skip dabusb support, and with latest versions of hotplug (depending on the GNU/Linux distro), we recommend our users to add dabusb to the blacklist file.
You will find related info about how we manage this issue there:
the project page..
http://eciadsl.flashtux.org/index.php?lang=enthe supported modems list (including vid/pid)..
http://eciadsl.flashtux.org/modems.php?lang=enour Q/A in the FAQ related to this dabusb issue..
http://eciadsl.flashtux.org/faq.php?lang=en#2.1My questions:
1/
Do we use the right method to avoid dabusb loading for non-audio-dabusb devices (modems).
Quoted 5 lines
2/
Is this really a conflict? What's your thought about this?3/
Do you have any piece of advice related to this (generally)?
The blacklist exists basically to cope with driver bugs, you should find out what the situation is. Does your modem driver include product/vendor IDs for pre-production units? If so, just remove them.
- Dave
Quoted 4 lines
Thanx in advance,
regards
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
Hi David Brownell,
On Mon, 05 May 2003 11:53:01 -0700 David Brownell <[email protected]> wrote:
[snip]
Quoted 3 lines
The URLs you mentioned don't seem available to me just now, so I can't see any more information than is found in your email message ...
Ah? The site maybe be long to reach, but should be online anyway.
[snip]
Quoted 12 lines
Assuming this isn't a hotplug bug, there are exactly three possibilities:
- This GlobeSpan driver is also trying to bind to default EZ-USB devices. **BOTH** drivers have bugs, since no production devices may use those IDs.
- Or the globespan driver is also trying to bind to the other (pid 9999) device entry. Assuming that ID was allocated legally only one of the drivers can possibly be correct.
Quoted 6 lines
- Both drivers are using VID/PID entries they aren't allowed to use, for the second entry (the one that's not the EZ-USB default). For example, that ID may not have been allocated to either one by Cypress.
Do you know which is the situation?
Arg. I feel I don't know what to answer, maybe I don't really understand this. All what I can say is that our driver (not officially supported by any brand) inits the modems (with vid1/pid1, and so get vid2/pid2) then loads the firmware sniffed under windows. Then it's able to use a .bin sniffed under windows too in order to get the ADSL synch, that's it.
But for this we need to disable dabusb of course..
Regards,
Signature · 11 lines
wwp ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net [email protected] https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
Quoted 4 lines · David Brownell
In fact, the current drivers (2.4, 2.5) seem to have only two IDs, both with Cypress vendor IDs. One of them is a bug, and needs to be removed. I don't know about the other one; it could be legit.
The current driver is available from
http://www1.baycom.de/download/dabusb/dabusb-linux-i386-2.24a.tar.gz,
the one in the kernel doesn't even work (it crashes on plugin of the device).
The current driver contains the Terratec production VID/PIDs, but also still has the Anchor VID/PIDs that were used in engineering models. These PID's should probably be disabled by default.
Tom
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
Hi Tom,
Quoted 6 lines
Quoted 4 lines
In fact, the current drivers (2.4, 2.5) seem to have only two IDs, both with Cypress vendor IDs. One of them is a bug, and needs to be removed. I don't know about the other one; it could be legit.
Quoted 8 lines
The current driver is available from
http://www1.baycom.de/download/dabusb/dabusb-linux-i386-2.24a.tar.gz,
the one in the kernel doesn't even work (it crashes on plugin of the device).The current driver contains the Terratec production VID/PIDs, but also still has the Anchor VID/PIDs that were used in engineering models. These PID's should probably be disabled by default.
So someone would make a number of folk happy by contributing patches to update both 2.4 and 2.5 kernels ... :)
- Dave
-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara The only event dedicated to issues related to Linux enterprise solutions www.enterpriselinuxforum.com
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
Hi Thomas Sailer,
Quoted 7 lines · On 06 May 2003 15:31:02 +0200 Thomas Sailer <[email protected]> wrote:
Quoted 4 lines · David Brownell
In fact, the current drivers (2.4, 2.5) seem to have only two IDs, both with Cypress vendor IDs. One of them is a bug, and needs to be removed. I don't know about the other one; it could be legit.
Quoted 8 lines
The current driver is available from
http://www1.baycom.de/download/dabusb/dabusb-linux-i386-2.24a.tar.gz,
the one in the kernel doesn't even work (it crashes on plugin of the device).The current driver contains the Terratec production VID/PIDs, but also still has the Anchor VID/PIDs that were used in engineering models. These PID's should probably be disabled by default.
Do you know if fixed dabusb (regarding to the previous topic) has been merged to recent 2.4 or 2.6-pre?
Regards,
Signature · 11 lines
wwp ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net [email protected] https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel