Re: [PATCH] Re: [Bug #10713] ehci splatter in 2.6.26-rc2

From: Alan Stern
Date: Tue May 20 2008 - 10:05:18 EST

On Tue, 20 May 2008, Lennert Buytenhek wrote:

> On Tue, May 20, 2008 at 02:18:24AM +0200, Lennert Buytenhek wrote:
>
> > > > This message has been generated automatically as a part of a report
> > > > of recent regressions.
> > > >
> > > > The following bug entry is on the current list of known regressions
> > > > from 2.6.25. Please verify if it still should be listed.
> > > >
> > > >
> > > > Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=10713
> > > > Subject : ehci splatter in 2.6.26-rc2
> > > > Submitter : Lennert Buytenhek <buytenh@xxxxxxxxxxxxxx>
> > > > Date : 2008-05-14 11:24 (5 days old)
> > > > References : http://marc.info/?l=linux-kernel&amp;m=121076435420129&amp;w=4
> > > > Handled-By : David Brownell <david-b@xxxxxxxxxxx>
> > >
> > > Just re-checked, and this device (a USB to dual PS/2 cable) works
> > > fine on 2.6.25 on this hardware:
> > >
> > > usb 1-1: new low speed USB device using orion-ehci and address 2
> > > usb 1-1: configuration #1 chosen from 1 choice
> > > input: PS2 to USB AdapterA3 as /class/input/input0
> > > input: USB HID v1.10 Keyboard [PS2 to USB AdapterA3] on usb-orion-ehci.0-1
> > > input: PS2 to USB AdapterA3 as /class/input/input1
> > > input: USB HID v1.10 Mouse [PS2 to USB AdapterA3] on usb-orion-ehci.0-1
> > >
> > > But on 2.6.26-rc3 it just gives me:
> > >
> > > usb 1-1: new low speed USB device using orion-ehci and address 2
> > > Unable to handle kernel NULL pointer dereference at virtual address 00000000
> > > pgd = c0004000
> > > [00000000] *pgd=00000000
> > > Internal error: Oops: 5 [#1] PREEMPT
> > > Modules linked in:
> > > CPU: 0 Not tainted (2.6.26-rc3 #347)
> > > PC is at qh_append_tds+0x24c/0x47c
> > > LR is at ehci_qtd_alloc+0x30/0x5c
> > > pc : [<c0242374>] lr : [<c0241d48>] psr: 00000093
> > > sp : c7cfdcb0 ip : c7e3c4c0 fp : c7cfdcfc
> > > r10: ffc83080 r9 : 00000008 r8 : c7d12240
> > > r7 : 80000080 r6 : 00000080 r5 : 00000000 r4 : 00000002
> > > r3 : c7d3f000 r2 : 00000000 r1 : 40800000 r0 : 08085000
> > > Flags: nzcv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel
> > > Control: 0005317f Table: 00004000 DAC: 00000017
> > > Process khubd (pid: 79, stack limit = 0xc7cfc268)
> > > Stack: (0xc7cfdcb0 to 0xc7cfe000)
> > > [...]
> > > Backtrace:
> > > [<c0242128>] (qh_append_tds+0x0/0x47c) from [<c0243ca0>] (ehci_urb_enqueue+0x100/0xff4)
> > > [<c0243ba0>] (ehci_urb_enqueue+0x0/0xff4) from [<c0234c30>] (usb_hcd_submit_urb+0x824/0x91c)
> > > [<c023440c>] (usb_hcd_submit_urb+0x0/0x91c) from [<c0235098>] (usb_submit_urb+0x224/0x260)
> > > [<c0234e74>] (usb_submit_urb+0x0/0x260) from [<c0235b5c>] (usb_start_wait_urb+0x44/0xac)
> > > r6:c7d12240 r5:c7cfde80 r4:00000000
> > > [<c0235b18>] (usb_start_wait_urb+0x0/0xac) from [<c0235dac>] (usb_control_msg+0xc8/0xec)
> > > r8:00000000 r7:00000100 r6:fffffff4 r5:00000040 r4:c7e3dc20
> > > [<c0235ce4>] (usb_control_msg+0x0/0xec) from [<c0230854>] (hub_port_init+0x274/0x5e4)
> > > [<c02305e0>] (hub_port_init+0x0/0x5e4) from [<c0231bec>] (hub_thread+0x608/0xc10)
> > > [<c02315e4>] (hub_thread+0x0/0xc10) from [<c0059e20>] (kthread+0x5c/0x94)
> > > [<c0059dc4>] (kthread+0x0/0x94) from [<c0047908>] (do_exit+0x0/0x67c)
> > > r6:00000000 r5:00000000 r4:00000000
> > > Code: e51bc040 e5932000 e51c309c e1520003 (15923000)
> > > ---[ end trace ce1992535f2e8e4d ]---
> > > note: khubd[79] exited with preempt_count 1
> > >
> > > If you have no idea what might have caused this to creep in, I guess
> > > I'll have to bisect it?
> >
> > A bisect turns up this:
> >
> > 7329e211b987a493cbcfca0e98c60eb108ab42df is first bad commit
> > commit 7329e211b987a493cbcfca0e98c60eb108ab42df
> > Author: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
> > Date: Thu Apr 3 18:02:56 2008 -0400
> >
> > USB: root hubs don't lie about their number of TTs
> >
> > Currently EHCI root hubs enumerate with a bDeviceProtocol code
> > indicating that they possess a Transaction Translator. However the
> > vast majority of controllers do not; they rely on a companion
> > controller to handle full- and low-speed communications. This patch
> > (as1064) changes the root-hub device descriptor to match the actual
> > situation.
> >
> > Signed-off-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxx>
> >
> > And indeed, reverting this commit from 2.6.26-rc3 makes my system
> > stop oopsing when I plug in the USB-PS/2 adapter (a low speed device),
> > and makes it work again as it did in 2.6.25.
>
> This patch appears to fix it:
>
>
> The Orion EHCI root hub does have a built-in Transaction Translator.
>
> Signed-off-by: Lennert Buytenhek <buytenh@xxxxxxxxxxx>
>
> Index: linux-2.6.26-rc3/drivers/usb/host/ehci-orion.c
> ===================================================================
> --- linux-2.6.26-rc3.orig/drivers/usb/host/ehci-orion.c
> +++ linux-2.6.26-rc3/drivers/usb/host/ehci-orion.c
> @@ -115,6 +115,8 @@ static int ehci_orion_setup(struct usb_h
> if (retval)
> return retval;
>
> + hcd->has_tt = 1;
> +
> ehci_reset(ehci);
> ehci_port_power(ehci, 0);

Pardon me for being puzzled, but I don't see how that change could
possibly have any effect on the problem you saw.

Nor do I see how the patch you identified could have caused any
problems. That has_tt flag isn't used anywhere in ehci-hcd (it is only
ever set, never read), and the only place it is used in usbcore is for
adjusting the root hub's device descriptor.

Alan Stern

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/