Scenario: usbtest test-14 failure: MUSBHDRC + Netchip2280

> -----Original Message-----
> From: David Brownell [mailto:[email protected]]
> Sent: Monday, March 26, 2007 11:06 PM
> To: Alan Stern
> Cc: Pandita, Vikram; [email protected]; linux-usb-
> [email protected]
> Subject: Re: usbtest test-14 failure: MUSBHDRC + Netchip2280
> 
> On Monday 26 March 2007 9:19 am, Alan Stern wrote:
> > On Mon, 26 Mar 2007, Pandita, Vikram wrote:
> >
> > If the MUSBHDRC really did send an extra OUT token, then the bug is
in
> the
> > MUSBHDRC.  If it really sent an IN token but the Netchip set the
flag
> for
> > an OUT token, then the bug is in the Netchip.
> 
> Vikram, it would help a lot if you could describe the specific request
> that's
> triggering this problem.  Read or write?  How many bytes of data?
Where
> in
> the ep0 message does this appear?  What does a USB packet sniffer have
to
> say?
> 

Scenario:
MUSBHDRC is acting as Host. Netchip2280 is acting as device.

On MUSBHDRC run the usbtest 14 and see the error -32.

Host-MUSBHDRC sends (OUT-TOKEN) 9 bytes to Control end point. 
CATC trace attached to this mail (as jpg) shows that the Netchip gets
the 9 bytes. 

Host-MUSBHDRC sends IN-token to netchip.

Netchip instead of ACK/NACK, does a STALL to the IN token, which is not
right.


Following is the mon output for the scenario ( if it helps ):
c7d7b200 1031576715 S Co:002:00 s 01 0b 0000 0000 0000 0
c7d7b200 1031576861 C Co:002:00 0 0
c7d7b200 1031643239 S Co:002:00 s 40 5b 0000 0000 0100 256 = 00010203
04050607 08090a0b 0c0d0e0f 10111213 14151617 18191a1b 1c1d1e1f
c7d7b200 1031643405 C Co:002:00 0 256 >
c7d7b200 1031646198 S Ci:002:00 s c0 5c 0000 0000 0100 256 <
c7d7b200 1031646324 C Ci:002:00 0 256 = 00010203 04050607 08090a0b
0c0d0e0f 10111213 14151617 18191a1b 1c1d1e1f
c7d7b200 1031646423 S Co:002:00 s 40 5b 0000 0000 0001 1 = 01
c7d7b200 1031646489 C Co:002:00 0 1 >
c7d7b200 1031646510 S Ci:002:00 s c0 5c 0000 0000 0001 1 <
c7d7b200 1031646567 C Ci:002:00 0 1 = 01
c7d7b200 1031646586 S Co:002:00 s 40 5b 0000 0000 0002 2 = 0203
c7d7b200 1031646638 C Co:002:00 0 2 >
c7d7b200 1031650150 S Ci:002:00 s c0 5c 0000 0000 0002 2 <
c7d7b200 1031650236 C Ci:002:00 0 2 = 0203
c7d7b200 1031650263 S Co:002:00 s 40 5b 0000 0000 0003 3 = 030405
c7d7b200 1031650303 C Co:002:00 0 3 >
c7d7b200 1031650321 S Ci:002:00 s c0 5c 0000 0000 0003 3 <
c7d7b200 1031650417 C Ci:002:00 0 3 = 030405
c7d7b200 1031650436 S Co:002:00 s 40 5b 0000 0000 0004 4 = 04050607
c7d7b200 1031650484 C Co:002:00 0 4 >
c7d7b200 1031650501 S Ci:002:00 s c0 5c 0000 0000 0004 4 <
c7d7b200 1031650563 C Ci:002:00 0 4 = 04050607
c7d7b200 1031650587 S Co:002:00 s 40 5b 0000 0000 0005 5 = 05060708 09
c7d7b200 1031650621 C Co:002:00 0 5 >
c7d7b200 1031650634 S Ci:002:00 s c0 5c 0000 0000 0005 5 <
c7d7b200 1031650723 C Ci:002:00 0 5 = 05060708 09
c7d7b200 1031650741 S Co:002:00 s 40 5b 0000 0000 0006 6 = 06070809 0a0b
c7d7b200 1031650778 C Co:002:00 0 6 >
c7d7b200 1031650795 S Ci:002:00 s c0 5c 0000 0000 0006 6 <
c7d7b200 1031650855 C Ci:002:00 0 6 = 06070809 0a0b
c7d7b200 1031650873 S Co:002:00 s 40 5b 0000 0000 0007 7 = 0708090a
0b0c0d
c7d7b200 1031650948 C Co:002:00 0 7 >
c7d7b200 1031650965 S Ci:002:00 s c0 5c 0000 0000 0007 7 <
c7d7b200 1031651014 C Ci:002:00 0 7 = 0708090a 0b0c0d
c7d7b200 1031651033 S Co:002:00 s 40 5b 0000 0000 0008 8 = 08090a0b
0c0d0e0f
c7d7b200 1031651064 C Co:002:00 0 8 >
c7d7b200 1031651078 S Ci:002:00 s c0 5c 0000 0000 0008 8 <
c7d7b200 1031651160 C Ci:002:00 0 8 = 08090a0b 0c0d0e0f

c7d7b200 1031651178 S Co:002:00 s 40 5b 0000 0000 0009 9 = 090a0b0c
0d0e0f10 11
c7d7b200 1031656583 C Co:002:00 -32 9 >







> I've certainly seen silicon issues in both the HDRC and the net2280,
so I
> could believe either one's at fault here.  Most of the net2280 issues
have
> workarounds, at least for the main chip versions ... so I'm more
> suspicious
> of the HDRC code, since it's been much more of a pain-in-the-butt and
is
> in
> any case is much newer code.
> 
> - Dave

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel