FAQ: [#5944] Disable anomaly 456 for bf526-ezbrd, gadget fs bulk test fails in dma mode (2010)

[#5944] Disable anomaly 456 for bf526-ezbrd, gadget fs bulk test fails in dma mode

Submitted By: Vivi Li

Open Date

2010-03-05 04:41:03

Priority:

Very Low Assignee:

Bob Liu

Status:

Open Fixed In Release:

N/A

Found In Release:

2010R1 Release:

Category:

Drivers Board:

N/A

Processor:

BF526 Silicon Revision:

Is this bug repeatable?:

Yes Resolution:

Assigned (Not Start)

Uboot version or rev.:

Toolchain version or rev.:

gcc4.3-2010_Jan_22

App binary format:

N/A

Summary: Disable anomaly 456 for bf526-ezbrd, gadget fs bulk test fails in dma mode

Details:

Disable anomaly 456 for bf526-ezbrd, gadget fs bulk test fails in dma mode.

On target:

root:/> mkdir /dev/gadget

root:/> mount -t gadgetfs gadgetfs /dev/gadget

gadgetfs: USB Gadget filesystem, version 24 Aug 2004

root:/> ls /dev/gadget/ -l

-rw------- 1 root root 0 Jan 2 13:04 musb_hdrc

root:/> ./usb -r "1.3" -v^M

gadgetfs: bound to musb_hdrc driver^M

/dev/gadget/musb_hdrc ep0 configured^M

serial="1.3"^M

gadgetfs: connected^M

** Fri Jan 2 13:04:30 1970^M

CONNECT high speed^M

gadgetfs: disconnected^M

DISCONNECT^M

gadgetfs: connected^M

read 2 ep0 events^M

CONNECT high speed^M

SETUP 80.06 v0300 i0000 255^M

gadgetfs: configuration #3^M

SETUP 80.06 v0302 i0409 255^M

SETUP 80.06 v0301 i0409 255^M

SETUP 80.06 v0303 i0409 255^M

SETUP 00.09 v0003 i0000 0^M

CONFIG #3^M

simple_source_thread start 1026 fd 6^M

SETUP 80.06 v0304 i0409 255^M

simple_sink_thread start 2051 fd 7^M

SETUP 80.06 v0305 i0409 255^M

On host:

sudo ./src/testusb -D /proc/bus/usb/005/066 -t1

unknown speed /proc/bus/usb/005/066

(can not exit)

Follow-ups

--- Cliff Cai 2010-07-06 02:30:26

Some ideas about the changes should be made to get mode 1 work under current

driver framework.

1.make sure that the actual data length in the rx fifo(RXCOUNT) should be equal

to musb_ep->packet_sz,then we can make usb work in DMA MODE 1.We can't just

enable this mode because we see that the request-> length is bigger than

musb_ep->packet_sz.

2.the current driver starts DMA transfer only when there is already a packet in

the rx fifo,this doesn't work for DMA MODE1.In other words,we should set up

mode1 before any packet arrives,or we should read out the first packet in DMA

MODE0 and set up MODE 1 transfer for the rest packets.

Cliff

--- Sonic Zhang 2010-08-24 00:09:21

No hardware fix in bf526 v0.2 for anomaly 456 is reported to

processor.support@analog.com.

Downgrade to 5.

Files

Changes

Commits

Dependencies

Duplicates

Associations