On Wed, 28 Aug 2013, Jürgen Liebmann wrote: > In the attachment the both usbmon traces: > > usbmon-reverted.out (camera works ok) > usbmon-74a6af256648 (camera fails) This is very strange indeed. Here's where the two traces start to get different results. In the kernel with the commit reverted: ffff88023073f9c0 742830275 S Co:2:009:0 s 01 0b 0000 0001 0000 0 ffff88023073f9c0 742830331 C Co:2:009:0 0 0 ffff88023073f9c0 742830336 S Ci:2:009:0 s a1 87 0100 0001 001a 26 < ffff88023073f9c0 742830583 C Ci:2:009:0 -32 0 ffff88023073f9c0 742830592 S Ci:2:009:0 s a1 81 0100 0001 001a 26 < ffff88023073f9c0 742831081 C Ci:2:009:0 0 26 = 01000101 15160500 00000000 00000000 00000018 0300fc03 0000 ffff88023073fc00 742831291 S Ci:2:009:0 s 80 06 0302 0409 00ff 255 < ffff88023073fc00 742831583 C Ci:2:009:0 0 52 = 34034d00 69006300 72006f00 73006f00 66007400 20004c00 69006600 65004300 This shows a successful Set-Interface request, followed by an unsuccessful class-specific request and a successful class-specific request which returns 26 data bytes. Following that, the computer reads a 52-byte string descriptor. Now here's the same thing with the commit present: ffff880233820f00 1078447196 S Co:2:008:0 s 01 0b 0000 0001 0000 0 ffff880233820f00 1078447253 C Co:2:008:0 0 0 ffff880233739cc0 1078447266 S Ci:2:008:0 s a1 87 0100 0001 001a 26 < ffff880233739cc0 1078447523 C Ci:2:008:0 -32 0 ffff880233739cc0 1078447539 S Ci:2:008:0 s a1 81 0100 0001 001a 26 < ffff880233739cc0 1078447898 C Ci:2:008:0 -32 0 ffff880233739a80 1078448123 S Ci:2:008:0 s 80 06 0302 0409 00ff 255 < ffff880233739a80 1078448542 C Ci:2:008:0 0 26 = 01000101 15160500 00000000 00000000 00000018 0300fc03 0000 ffff880233739a80 1078448556 S Ci:2:008:0 s 80 06 0302 0409 00ff 255 < ffff880233739a80 1078448907 C Ci:2:008:0 0 52 = 34034d00 69006300 72006f00 73006f00 66007400 20004c00 69006600 65004300 This starts with the same Set-Interface request and the same unsuccessful class-specific request. But the second class-specific request also fails, and the Get-String-Descriptor returns the 26 bytes of class-specific data! The Get-String-Descriptor request is repeated, and this time the device sends the correct 52-byte string descriptor. After that everything seems to return to normal. I don't understand this at all. There is no way that commit could cause the driver to malfunction like this. The only thing I can think of is some sort of hardware problem, either in the computer or (less likely) in the camera. Can you try testing this on a different computer? Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
