Re: [PATCH v1] usb: xhci: Check return value of wait for TRB_TRANSFER event

[Marek Vasut]

On 10/19/23 04:46, Minda Chen wrote:

[...]
000000004024a376 <abort_td>:
{
      4024a376:   7179                    addi    sp,sp,-48
      4024a378:   f406                    sd      ra,40(sp)
      4024a37a:   f022                    sd      s0,32(sp)
      4024a37c:   ec26                    sd      s1,24(sp)
      4024a37e:   e84a                    sd      s2,16(sp)
      4024a380:   e44e                    sd      s3,8(sp)
      4024a382:   e052                    sd      s4,0(sp)
      4024a384:   89ae                    mv      s3,a1
      4024a386:   84aa                    mv      s1,a0
          struct xhci_ctrl *ctrl = xhci_get_ctrl(udev);
      4024a388:   8c4fe0ef                jal     ra,4024844c <xhci_get_ctrl>
          struct xhci_ring *ring =  
ctrl->devs[udev->slot_id]->eps[ep_index].ring;
      4024a38c:   6785                    lui     a5,0x1
      4024a38e:   94be                    add     s1,s1,a5
      4024a390:   9444a603                lw      a2,-1724(s1)
      4024a394:   00198713                addi    a4,s3,1
      4024a398:   0712                    slli    a4,a4,0x4
      4024a39a:   02061793                slli    a5,a2,0x20
      4024a39e:   9381                    srli    a5,a5,0x20
      4024a3a0:   07c9                    addi    a5,a5,18
      4024a3a2:   078e                    slli    a5,a5,0x3
      4024a3a4:   97aa                    add     a5,a5,a0
      4024a3a6:   679c                    ld      a5,8(a5)
          xhci_queue_command(ctrl, NULL, udev->slot_id, ep_index, 
TRB_STOP_RING);
      4024a3a8:   2981                    sext.w  s3,s3
      4024a3aa:   86ce                    mv      a3,s3
          struct xhci_ring *ring =  
ctrl->devs[udev->slot_id]->eps[ep_index].ring;
      4024a3ac:   97ba                    add     a5,a5,a4
          xhci_queue_command(ctrl, NULL, udev->slot_id, ep_index, 
TRB_STOP_RING);
      4024a3ae:   4581                    li      a1,0
      4024a3b0:   473d                    li      a4,15
          struct xhci_ring *ring =  
ctrl->devs[udev->slot_id]->eps[ep_index].ring;
      4024a3b2:   0087ba03                ld      s4,8(a5) # 1008 
<_start-0x401feff8>
          struct xhci_ctrl *ctrl = xhci_get_ctrl(udev);
      4024a3b6:   842a                    mv      s0,a0
          xhci_queue_command(ctrl, NULL, udev->slot_id, ep_index, 
TRB_STOP_RING);
      4024a3b8:   d75ff0ef                jal     ra,4024a12c 
<xhci_queue_command>
          event = xhci_wait_for_event(ctrl, TRB_TRANSFER);
      4024a3bc:   02000593                li      a1,32
      4024a3c0:   8522                    mv      a0,s0
      4024a3c2:   ebdff0ef                jal     ra,4024a27e 
<xhci_wait_for_event>
          field = le32_to_cpu(event->trans_event.flags);
epc-> 4024a3c6:   455c                    lw      a5,12(a0)

So the fault occurs when reading the controller register(s), do I understand it 
right ?

I think it is right. Actually this error occur in error path, control tx 
transfer TRB_TRANSFER error occur and jump to error path.
sending TRB_TRANSFER again.
Could it be the problem is rather some clock, which are turned off after a 
fault ?
I think not. Just this udisk can reproduce this issue.

Can you take a closer look into this ? Is there maybe some hardware 
debug tool which can clarify what is going on better ?

It seems weird that controller register access would trigger this kind 
of bus fault (it is a bus fault, right ?)