https://bugzilla.kernel.org/show_bug.cgi?id=218890 --- Comment #15 from Alan Stern (stern@xxxxxxxxxxxxxxxxxxx) --- Here's the relevant part of the usbmon trace: ffff97af45927780 378815216 S Bo:5:004:2 -115 31 = 55534243 08000000 00020000 80000a28 00000000 00000001 0000000 000000 ffff97af45927780 378815233 C Bo:5:004:2 0 31 > ffff97af4255c900 378815246 S Bi:5:004:1 -115 512 < ffff97af4255c900 410624206 C Bi:5:004:1 -104 0 There's still a timeout and cancellation, but at least now the number of bytes transferred is reported as 0 and not 512. So the patch did fix this bug, but it didn't prevent the main problem from occurring. It looks like the main difference between the working and non-working kernels is that the non-working one issues a READ(10) command (added by Martin Petersen's commit) whereas the working one uses only READ(16) commands. I'll ask Martin about updating his commit to change this. -- You may reply to this email to add a comment. You are receiving this mail because: You are watching the assignee of the bug.
