Re: kern/43475: MTK 747 A+ GPS recorder has a bogus descriptor

[Jasper Wallace <jasper%pointless.net@localhost>]

The following reply was made to PR kern/43475; it has been noted by GNATS.

From: Jasper Wallace <jasper%pointless.net@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: kern-bug-people%netbsd.org@localhost, gnats-admin%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost
Subject: Re: kern/43475: MTK 747 A+ GPS recorder has a bogus descriptor
Date: Tue, 15 Jun 2010 22:33:50 +0100 (BST)

 On Tue, 15 Jun 2010, Jonathan A. Kollasch wrote:
 
 > The following reply was made to PR kern/43475; it has been noted by GNATS.
 > 
 > From: "Jonathan A. Kollasch" <jakllsch%kollasch.net@localhost>
 > To: gnats-bugs%NetBSD.org@localhost
 > Cc: 
 > Subject: Re: kern/43475: MTK 747 A+ GPS recorder has a bogus descriptor
 > Date: Tue, 15 Jun 2010 21:09:45 +0000
 > 
 >  So, you (were) get(ing) a "umodem0: no pointer to data interface" error?
 
 I can't remember what the error was, but there definatly wasn't a ucom 
 attachment, i can try with an unpatched kernel if it's important.
 
 >  Taking a hint from Linux and cdce(4), it would appear something
 >  along the lines of the CDCE_NO_UNION quirk would be more appropriate.
 
 Maybe, but this device does have a cdc union, it's just hidden behind the 
 bogus descriptor length. Also fixing this without a quirk may make the fix 
 just work with a larger range of devices and save someone having to work 
 out what the problem is and patch the kernel to make the device work for 
 them.
 
 >  Also, I'd sooner see USB_PRODUCT_MEDIATEK_MT3329, as this ID appears
 >  to be used in multiple products based on the MT3329 chip, and "MTK"
 >  doesn't seem similar enough to the name listed in the USB-IF vendor
 >  number database.
 
 Whatever is appropiate, it's not particulary important.
 
 -- 
 [http://pointless.net/]                                   [0x2ECA0975]