insmod uhci causes computer to reset

insmod uhci causes computer to reset

[Oliver Henshaw]

Even with the fixes described previously, grub fails with an UHCI
controller (VIA uhci
chipset on the motherboard). When no devices are attached, module
loading completes
successfully. When a mouse is attached, module loading completes but
listing the device with
'usb' causes the machine to reset. If I try without 'set debug=uhci'
then 'insmod uhci' causes
the computer to hang.





With no devices:



grub> insmod uhci

../bus/usb/uhci.c:188: class=0x0c 0x03 interface 0x00 base=0xe400

../bus/usb/uhci.c:278: UHCI initialized

../bus/usb/uhci.c:188: class=0x0c 0x03 interface 0x00 base=0xe800

../bus/usb/uhci.c:278: UHCI initialized

../bus/usb/uhci.c:636: detect=0x48a port=0

../bus/usb/uhci.c:636: detect=0x58a port=1

../bus/usb/uhci.c:636: detect=0x48a port=0

../bus/usb/uhci.c:636: detect=0x48a port=1

../bus/usb/uhci.c:668: registered


grub>


With a mouse attached:



grub> insmod uhci

../bus/usb/uhci.c:188: class=0x0c 0x03 interface 0x00 base=0xe400

../bus/usb/uhci.c:278: UHCI initialized

../bus/usb/uhci.c:188: class=0x0c 0x03 interface 0x00 base=0xe800

../bus/usb/uhci.c:278: UHCI initialized

../bus/usb/uhci.c:636: detect=0x48a port=0

../bus/usb/uhci.c:636: detect=0x5ab port=1

../bus/usb/uhci.c:582: enable=1 port=1

../bus/usb/uhci.c:593: detect=0x5ab

../bus/usb/uhci.c:602: reset completed

../bus/usb/uhci.c:608: waiting for the port to be enabled

../bus/usb/uhci.c:613: >3detect=0x5a7

../bus/usb/usbtrans.c:43: control: reqtype=0x80 req=0x06 val=0x100 idx=0x00 size

=18

../bus/usb/uhci.c:408: transaction: endp=0, type=2, addr=0, toggle=0, size=8 dat

a=0x67ad0 td=0x2403e000

../bus/usb/uhci.c:408: transaction: endp=0, type=0, addr=0, toggle=1, size=18 da

ta=0x2403d760 td=0x2403e020

../bus/usb/uhci.c:408: transaction: endp=0, type=1, addr=0, toggle=1, size=0 dat

a=0x0 td=0x2403e040

../bus/usb/uhci.c:481: setup transaction 0

../bus/usb/uhci.c:487: initiate transaction

../bus/usb/uhci.c:498: >t status=0x440007 data=0x67ad0 td=0x2403e000

../bus/usb/uhci.c:504: t status=0x440007

../bus/usb/uhci.c:509: >>t status=0x440007

../bus/usb/uhci.c:548: transaction failed

../bus/usb/usbtrans.c:43: control: reqtype=0x00 req=0x05 val=0x01 idx=0x00 size=
0

../bus/usb/uhci.c:408: transaction: endp=0, type=2, addr=0, toggle=0, size=8 dat
a=0x67b10 td=0x2403e040

../bus/usb/uhci.c:408: transaction: endp=0, type=0, addr=0, toggle=1, size=0 dat
a=0x0 td=0x2403e020

../bus/usb/uhci.c:481: setup transaction 0

../bus/usb/uhci.c:487: initiate transaction

../bus/usb/uhci.c:498: >t status=0x440007 data=0x67b10 td=0x2403e040

../bus/usb/uhci.c:504: t status=0x440007

../bus/usb/uhci.c:509: >>t status=0x440007

../bus/usb/uhci.c:548: transaction failed

../bus/usb/uhci.c:636: detect=0x48a port=0

../bus/usb/uhci.c:636: detect=0x48a port=1

../bus/usb/uhci.c:668: registered

grub> usb

USB devices:



USB version 0.0, VendorID: 0x00, ProductID: 0x00, #conf: 0

Low speed device

Interface #0: #Endpoints: 22   ../bus/usb/usbtrans.c:43: control: reqtype=0x80 r
eq=0x06 val=0x3c3 idx=0x409 size=1


(and then the computer resets)

Re: insmod uhci causes computer to reset

[Pavel Roskin]

On Sun, 2009-06-07 at 20:03 +0100, Oliver Henshaw wrote:
> Even with the fixes described previously, grub fails with an UHCI
> controller (VIA uhci
> chipset on the motherboard). When no devices are attached, module
> loading completes
> successfully. When a mouse is attached, module loading completes but
> listing the device with
> 'usb' causes the machine to reset. If I try without 'set debug=uhci'
> then 'insmod uhci' causes
> the computer to hang.

Maybe the crash is in usb_keyboard?

I know that term/usb_keyboard.c assumes that there is exactly one USB
keyboard.  It there is no USB keyboard, the code would crash because
usbdev is NULL.  Also, the matching of the keyboard is incorrect.  It
checks the device descriptor (even that code is a hack), but not the
interface descriptor.

-- 
Regards,
Pavel Roskin

Re: insmod uhci causes computer to reset

[Oliver Henshaw]

2009/6/7 Pavel Roskin <proski@gnu.org>:
> Maybe the crash is in usb_keyboard?
I don't think that can be it. AFAICS usb_keyboard would only be called
if I manually loaded the model. And I've never done that.

Re: insmod uhci causes computer to reset

[Pavel Roskin]

On Sun, 2009-06-07 at 21:22 +0100, Oliver Henshaw wrote:
> 2009/6/7 Pavel Roskin <proski@gnu.org>:
> > Maybe the crash is in usb_keyboard?
> I don't think that can be it. AFAICS usb_keyboard would only be called
> if I manually loaded the model. And I've never done that.

I see.  Anyway, the USB code needs a lot of work, and I appreciate your
effort.  The original author of the USB code, Marco Gerards, wrote
recently that he doesn't have much time for the GRUB work.

-- 
Regards,
Pavel Roskin