-----Original Message----- From: u-boot-bounces@lists.denx.de [mailto:u-boot-bounces@lists.denx.de] On Behalf Of Sergei Trofimovich Sent: Saturday, August 07, 2010 7:10 PM To: Wolfgang Denk Cc: u-boot@lists.denx.de Subject: Re: [U-Boot] 'usb start' works very unstable
It's a tip of 'u-boot-marvell/testing' branch.
Did you try current mainline instead? It is usually not recommended to use any of the custodian trees and especially their testing branches unless you know _exactly_ what these contain and
what you are
looking after.
I was not sure sheevaplug support was merged. Pesky page [0] suggested it and I thought i's called 'mainline'.
Ok, trying 'u-boot/master' v2010.06-267-gb1f95b4 with latest commit
commit b1f95b4438a6d44e12f04e7f51412f7a711ef87b Author: Wolfgang Denk wd@denx.de Date: Sat Aug 7 00:32:50 2010 +0200
Replace CHANGELOG files by auto-generated "snapshot.commit"Marvell>> version U-Boot 2010.06-00267-gb1f95b4 (Aug 07 2010 - 15:30:45) Marvell-Sheevaplug[ 23.015860] scsi0 : usb-storage 1-1:1.0 [ 26.054268] scsi 0:0:0:0: Direct-AccessStoreJet Transcend PQ: 0 ANSI: 2 CCS
[ 28.774083] sd 0:0:0:0: [sda] 488397168 512-bytelogical blocks: (250 GB/232 GiB)
...
Device 0: Vendor: StoreJet Rev: Prod: Transcend Type: Hard Disk Capacity: 28759.9 MB = 28.0 GB (488397168 x 512)Heh, capacity is reported incorrectly, but it works am I'm able to load kernel from there.
The reason for this is that the multiplication 488397168 x 512 will overflow a 32 bit integer as used in "disk/part.c"; this shouldbe fixed...
This issue is still here:
Marvell>> usb start (Re)start USB... USB: Register 10011 NbrPorts 1 USB EHCI 1.00 scanning bus for devices... 2 USB Device(s) found scanning bus for storage devices... 1 StorageDevice(s) found Marvell>> usb storage Device 0: Vendor: StoreJet Rev: Prod: Transcend Type: Hard Disk Capacity: 28759.9 MB = 28.0 GB (488397168 x 512)
Latest change in 'disk/part.c' is:
commit 4b142febff71eabdb7ddbb125c7b583b24ddc434
Author: Heiko Schocher hs@denx.de
Date: Thu Dec 3 11:21:21 2009 +0100
Which looks older, than u-boot-marvell branch.
All these things are almost 100% reproducible.
Vendor-shipped (Marvell's) u-boot has exactly the same disease: ** MARVELL BOARD: SHEEVA PLUG LE U-Boot 1.1.4 (Mar 19 2009 - 16:06:59) Marvell version: 3.4.16
This can be anything - issues with the USB driver, missing or incorrect initializatioon of the SoC, or even hardware issues.
Can you try to debug it?
And I can't reproduce it anymore with 'u-boot/master'. Rebooted 3 times and I have stable detection \o/.
I can still reproduce another (hardware?) failure in a dirty way:
- Boot Linux kernel
- run hdparm on the root partition (not the beast idea, as
we will see): sh-4.0# hdparm /dev/sda /dev/sda: HDIO_DRIVE_CMD(identify) failed: Invalid exchange readonly = 0 (off) readahead = 256 (on) geometry = 30401/255/63, sectors = 488397168, start = 0 system losts USB, root, and ability to run binaries 3. attach with JTAG: $sudo openocd -f /usr/share/openocd/scripts/board/sheevaplug.cfg Open On-Chip Debugger 0.5.0-dev-00466-g28ddefd (2010-08-06-22:54) ... > reset
This will just reset the Kirkwood and not entire system, u-boot performs some basic initialization that may not be sufficient to bring back already sucked peripharal unless we reset them.
I think - If the things work with true system reset (i.e. re-power) that is well withing u-boot's scope and limitation.
Regards.. Prafulla . .