Malta + USB on 2.4, anyone?

Malta + USB on 2.4, anyone?

[Jun Sun]

Has anybody tried USB on malta with 2.4 kernel?  I just found that
I got 0xff IRQ number and kernel panics.

Will look further tomorrow, but want to see if anybody knows about 
it first.

Also, the kgdb seems to be flaky.  Targets can send chars too fast
so that chars get lost.  It appears that the linux status register
might be lying about "transmitter buffer empty".  

Jun

Re: Malta + USB on 2.4, anyone?

[Chris Dearman]

Jun Sun wrote:
> Has anybody tried USB on malta with 2.4 kernel?  I just found that
> I got 0xff IRQ number and kernel panics.

> Also, the kgdb seems to be flaky.  Targets can send chars too fast
> so that chars get lost.  It appears that the linux status register
> might be lying about "transmitter buffer empty".  

   I regularly use gdb @ 115200 on Malta and haven't noticed any problems.

	Chris


-- 
Chris Dearman          The Fruit Farm, Ely Road    voice +44 1223 706206
MIPS Technologies (UK) Chittering, Cambs, CB5 9PH  fax   +44 1223 706250

Re: Malta + USB on 2.4, anyone?

[Maciej W. Rozycki]

On Wed, 30 Jul 2003, Jun Sun wrote:

> Has anybody tried USB on malta with 2.4 kernel?  I just found that
> I got 0xff IRQ number and kernel panics.

 Possibly IRQ routing is broken -- the PIIX4 uses INTD for its USB
controller's interrupt.  For the Malta it should be routed to the IRQ11
input of the PIIX4's internal dual-8259A PIC.  What does `/sbin/lspci -vv
-s 00:0a.2' print?

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: Malta + USB on 2.4, anyone?

[Jun Sun]

On Thu, Jul 31, 2003 at 12:26:44PM +0200, Maciej W. Rozycki wrote:
> On Wed, 30 Jul 2003, Jun Sun wrote:
> 
> > Has anybody tried USB on malta with 2.4 kernel?  I just found that
> > I got 0xff IRQ number and kernel panics.
> 
>  Possibly IRQ routing is broken -- the PIIX4 uses INTD for its USB
> controller's interrupt.  For the Malta it should be routed to the IRQ11
> input of the PIIX4's internal dual-8259A PIC.  What does `/sbin/lspci -vv
> -s 00:0a.2' print?
>

The output seems to say the same thing:

root@10.0.18.6:~# lspci -vv -s 00:0a.2                                          
00:0a.2 USB Controller: Intel Corporation 82371AB PIIX4 USB (rev 01) (prog-if 00
 [UHCI])                                                                        
        Control: I/O+ Mem- BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Step
ping- SERR- FastB2B-                                                            
        Status: Cap- 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort
- <MAbort- >SERR- <PERR-                                                        
        Latency: 32                                                             
        Interrupt: pin D routed to IRQ 11                                       
        Region 4: I/O ports at 1220 [size=32]                                   
                                                                                
Jun

Re: Malta + USB on 2.4, anyone?

[Jun Sun]

On Thu, Jul 31, 2003 at 10:53:06AM +0100, Chris Dearman wrote:
> Jun Sun wrote:
> > Has anybody tried USB on malta with 2.4 kernel?  I just found that
> > I got 0xff IRQ number and kernel panics.
> 
> > Also, the kgdb seems to be flaky.  Targets can send chars too fast
> > so that chars get lost.  It appears that the linux status register
> > might be lying about "transmitter buffer empty".  
> 
>    I regularly use gdb @ 115200 on Malta and haven't noticed any problems.
>

Yes.  I notice that kgdb generally works, but gets flaky when I hit
this particular problem.  That makes me wonder maybe something gets
very wrong with the controller chip....

Keep debugging ...

Jun

Re: Malta + USB on 2.4, anyone?

[Jun Sun]

On Thu, Jul 31, 2003 at 10:36:29AM -0700, Jun Sun wrote:
> On Thu, Jul 31, 2003 at 12:26:44PM +0200, Maciej W. Rozycki wrote:
> > On Wed, 30 Jul 2003, Jun Sun wrote:
> > 
> > > Has anybody tried USB on malta with 2.4 kernel?  I just found that
> > > I got 0xff IRQ number and kernel panics.
> > 
> >  Possibly IRQ routing is broken -- the PIIX4 uses INTD for its USB
> > controller's interrupt.  For the Malta it should be routed to the IRQ11
> > input of the PIIX4's internal dual-8259A PIC.  What does `/sbin/lspci -vv
> > -s 00:0a.2' print?
> >
> 
> The output seems to say the same thing:
> 
> root@10.0.18.6:~# lspci -vv -s 00:0a.2                                          
> 00:0a.2 USB Controller: Intel Corporation 82371AB PIIX4 USB (rev 01) (prog-if 00
>  [UHCI])                                                                        
>         Control: I/O+ Mem- BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Step
> ping- SERR- FastB2B-                                                            
>         Status: Cap- 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort
> - <MAbort- >SERR- <PERR-                                                        
>         Latency: 32                                                             
>         Interrupt: pin D routed to IRQ 11                                       
>         Region 4: I/O ports at 1220 [size=32]                                   
> 

Using the alternative JE driver sovles the problem.  

I suspect the main UHCI driver does not get cache flushing
or bus/virt address right.


Jun

Re: Malta + USB on 2.4, anyone?

[Maciej W. Rozycki]

On Thu, 31 Jul 2003, Jun Sun wrote:

> Using the alternative JE driver sovles the problem.  
> 
> I suspect the main UHCI driver does not get cache flushing
> or bus/virt address right.

 Well, since the JE driver is the UHCI driver for 2.6, I wouldn't care.

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: Malta + USB on 2.4, anyone?

[Atsushi Nemoto]

>>>>> On Thu, 31 Jul 2003 11:15:06 -0700, Jun Sun <jsun@mvista.com> said:
jsun> I suspect the main UHCI driver does not get cache flushing or
jsun> bus/virt address right.

It seems usb_control_msg is not safe with an unaligned buffer.  Is
this patch solve your problem?

diff -u linux-2.4.21/drivers/usb/usb.c linux/drivers/usb/usb.c
--- linux-2.4.21/drivers/usb/usb.c	Tue Jul 15 13:49:34 2003
+++ linux/drivers/usb/usb.c	Tue Jul 15 21:20:44 2003
@@ -1172,9 +1172,24 @@
 {
 	struct usb_ctrlrequest *dr = kmalloc(sizeof(struct usb_ctrlrequest), GFP_KERNEL);
 	int ret;
+#ifdef __mips__ /* BUG??? */
+	void *tmpdata = NULL;
+#endif
 	
 	if (!dr)
 		return -ENOMEM;
+#ifdef __mips__ /* BUG??? */
+	if (size) {
+		tmpdata = kmalloc(size, GFP_KERNEL);
+		if (!data) {
+			kfree(dr);
+			return -ENOMEM;
+		}
+		memcpy(tmpdata, data, size);
+	} else {
+		tmpdata = data;
+	}
+#endif
 
 	dr->bRequestType = requesttype;
 	dr->bRequest = request;
@@ -1184,7 +1199,15 @@
 
 	//dbg("usb_control_msg");	
 
+#ifdef __mips__ /* BUG??? */
+	ret = usb_internal_control_msg(dev, pipe, dr, tmpdata, size, timeout);
+	if (size) {
+		memcpy(data, tmpdata, size);
+		kfree(tmpdata);
+	}
+#else
 	ret = usb_internal_control_msg(dev, pipe, dr, data, size, timeout);
+#endif
 
 	kfree(dr);
 
---
Atsushi Nemoto

Re: Malta + USB on 2.4, anyone?

[Jun Sun]

On Fri, Aug 01, 2003 at 05:26:23PM +0900, Atsushi Nemoto wrote:
> >>>>> On Thu, 31 Jul 2003 11:15:06 -0700, Jun Sun <jsun@mvista.com> said:
> jsun> I suspect the main UHCI driver does not get cache flushing or
> jsun> bus/virt address right.
> 
> It seems usb_control_msg is not safe with an unaligned buffer.  Is
> this patch solve your problem?
>

Nope.

The symptoms of erractic hw behavior seems to suggest maybe addressing
is wrong in physical space.  Just a guess.  I don't think I will spend
more time on this though. :)

Thanks for the patch.

Jun