https://bugzilla.kernel.org/show_bug.cgi?id=213081 --- Comment #6 from Alan Stern (stern@xxxxxxxxxxxxxxxxxxx) --- Here's where the difference shows up. In the 5.11 log: ffff8881259c3540 48224904 S Bo:6:004:2 -115 31 = 55534243 03000000 60000000 80000603 00000060 00000000 00000000 000000 ffff8881259c3540 48224939 C Bo:6:004:2 0 31 > ffff888102854900 48224942 S Bi:6:004:1 -115 96 < ffff888102854900 48225040 C Bi:6:004:1 -121 18 = 70000200 0000000a 00000000 3a000000 0000 ffff8881259c3540 48225042 S Bi:6:004:1 -115 13 < ffff8881259c3540 48225112 C Bi:6:004:1 -32 0 ffff8881259c3540 48225115 S Co:6:004:0 s 02 01 0000 0081 0000 0 ffff8881259c3540 48225522 C Co:6:004:0 0 0 ffff8881259c3540 48225525 S Bi:6:004:1 -115 13 < ffff8881259c3540 48225598 C Bi:6:004:1 0 13 = 55534253 03000000 4e000000 00 In the 5.12 log: ffff88812366e6c0 273105524 S Bo:6:010:2 -115 31 = 55534243 03000000 60000000 80000603 00000060 00000000 00000000 000000 ffff88812366e6c0 273105551 C Bo:6:010:2 0 31 > ffff88815e5c5c00 273105558 S Bi:6:010:1 -115 96 < ffff88815e5c5c00 273105676 C Bi:6:010:1 -121 18 = 70000200 0000000a 00000000 3a000000 0000 ffff88812366e6c0 273105682 S Bi:6:010:1 -115 13 < ffff88812366e6c0 273105860 C Bi:6:010:1 0 0 ffff88812366e6c0 273105864 S Bi:6:010:1 -115 13 < ffff88812366e6c0 303293804 C Bi:6:010:1 -104 0 In both of these, the computer asks for a 96-bytes of sense data and the device responds with only 18 bytes. Then when the computer asks for a status code, the device is supposed to respond with a STALL (code -32 in the trace). This does happen (in the sixth line) under 5.11, but under 5.12 the STALL isn't reported. Things get worse from there. This appears to be a bug in the xhci-hcd driver. I can't tell you how to debug that driver; your best approach is probably to use bisection to find the commit between 5.11 and 5.12 responsible for introducing the bug. -- You may reply to this email to add a comment. You are receiving this mail because: You are watching the assignee of the bug.
