USB 2.0 devices slow down significantly in stock 2.6.10 kernel of 10.1

firebird · 03-24-2005, 08:46 AM

I'm pretty new with Linux (half year with Slack and not yet compiled the kernel

), so please bear my ignorance.

Short story first.

I upgraded to the 2.6.10 kernel shipped in 10.1 release, in hoping that USB 2.0 devices can work better (and faster). My SanDisk USB 2.0 pen drive that can not be reliably recognized by 2.4.x kernel did work great now. However, I found ALL of my USB 2.0 high speed drives now became very slow, an informal probe give me a unbearable transfer rate of 0.48 MByte/s. Most of them (except the SanDisk one) works just great under 2.4.x kernel.

I've tried solution suggested by thread 271226 (I cannot post URL yet, so I have to post like this. sorry). i.e., add the following commands in rc.local:

Code:

rmmod ohci_hcd
rmmod ehci_hcd
modprobe ehci_hcd
modprobe ohci_hcd

It does not help.

Thread 286473 mentioned that avoiding "supermount" will help. But I don't know what that mean. Is it applicable to Slackware?

Now the long story. Detail of my upgrade process first:

I upgrade to 2.6.10 kernel like this:

Code:

telinit 1
mount -o loop -o ro slackware-10.1-install-d2.iso /mnt/disk2
cd /mnt/disk2/testing/packages/linux-2.6.10
installpkg kernel-generic-2.6.10-i486-1.tgz
installpkg kernel-modules-2.6.10-i486-1.tgz
installpkg alsa-driver-1.0.8_2.6.10-i486-1.tgz

# My boot partition is reiserfs:
cd /boot
mkinitrd -c -k 2.6.10 -m reiserfs

... and add a new GRUB menu item looks like this:

Code:

title Slackware 10.1 [2.6.10]
    root (hd1,7)
    kernel /vmlinuz root=/dev/hdb9 ro vga=771
    initrd /initrd.gz
    boot

... reboot and done. I found the original "/etc/rc.d/rc.modules" is renamed to a BAK file, so I moved it to my backup directory.

Now detail about my config. I really can not find anything wrong according to all documents I've read. I wonder if someone can share his experience with USB 2.0 high speed devices on the stock 2.6.10 of 10.1. I just find a few reports regarding such problem (USB 2.0 slow down), I believe the majority of us just get it right. If it works in your case, please let me know, at least give me some hope

Please help me find out what I've done wrong. Thank you very much for your time and your kind help!

# lspci -v | grep HCI

Code:

00:07.2 USB Controller: VIA Technologies, Inc. VT6202 [USB 2.0 controller] (rev 1a) (prog-if 00 [UHCI])
00:07.3 USB Controller: VIA Technologies, Inc. VT6202 [USB 2.0 controller] (rev 1a) (prog-if 00 [UHCI])
00:0f.0 USB Controller: NEC Corporation USB (rev 41) (prog-if 10 [OHCI])
00:0f.1 USB Controller: NEC Corporation USB (rev 41) (prog-if 10 [OHCI])
00:0f.2 USB Controller: NEC Corporation USB 2.0 (rev 02) (prog-if 20 [EHCI])

# lsmod | grep hcd

Code:

ohci_hcd               18824  0
ehci_hcd               28420  0
uhci_hcd               28944  0

When connect my SanDisk USB 2.0 pen drive:
# tail -f /var/log/messages

Code:

Mar 24 21:59:30 mixer kernel: usb 1-4: new high speed USB device using ehci_hcd and address 2
Mar 24 21:59:31 mixer kernel: uba: device 2 capacity nsec 512000 bsize 512
Mar 24 21:59:31 mixer kernel: uba: device 2 capacity nsec 512000 bsize 512
Mar 24 21:59:31 mixer udev[6963]: creating device node '/dev/uba'
Mar 24 21:59:31 mixer kernel:  uba: uba1
Mar 24 21:59:31 mixer kernel: usbcore: registered new driver ub
Mar 24 21:59:31 mixer udev[6982]: configured rule in '/etc/udev/rules.d/000_local.rules' at line 22 applied, added symlink 'USB_sandisk256M'
Mar 24 21:59:31 mixer udev[6982]: creating device node '/dev/uba1'

#udevinfo -a -p `udevinfo -q path -n /dev/uba1`

Code:

  looking at the device chain at '/sys/devices/pci0000:00/0000:00:0f.2/usb1/1-4':
    BUS="usb"
    ID="1-4"
    SYSFS{bConfigurationValue}="1"
    SYSFS{bDeviceClass}="00"
    SYSFS{bDeviceProtocol}="00"
    SYSFS{bDeviceSubClass}="00"
    SYSFS{bMaxPower}="100mA"
    SYSFS{bNumConfigurations}="1"
    SYSFS{bNumInterfaces}=" 1"
    SYSFS{bcdDevice}="0104"
    SYSFS{bmAttributes}="80"
    SYSFS{detach_state}="0"
    SYSFS{devnum}="2"
    SYSFS{idProduct}="7101"
    SYSFS{idVendor}="0781"
    SYSFS{manufacturer}="SanDisk Corporation "
    SYSFS{maxchild}="0"
    SYSFS{product}="Cruzer Mini "
    SYSFS{serial}="000635722"
    SYSFS{speed}="480"
    SYSFS{version}=" 2.00"

  looking at the device chain at '/sys/devices/pci0000:00/0000:00:0f.2/usb1':
    BUS="usb"
    ID="usb1"
    SYSFS{bConfigurationValue}="1"
    SYSFS{bDeviceClass}="09"
    SYSFS{bDeviceProtocol}="01"
    SYSFS{bDeviceSubClass}="00"
    SYSFS{bMaxPower}="  0mA"
    SYSFS{bNumConfigurations}="1"
    SYSFS{bNumInterfaces}=" 1"
    SYSFS{bcdDevice}="0206"
    SYSFS{bmAttributes}="e0"
    SYSFS{detach_state}="0"
    SYSFS{devnum}="1"
    SYSFS{idProduct}="0000"
    SYSFS{idVendor}="0000"
    SYSFS{manufacturer}="Linux 2.6.10 ehci_hcd"
    SYSFS{maxchild}="5"
    SYSFS{product}="NEC Corporation USB 2.0"
    SYSFS{serial}="0000:00:0f.2"
    SYSFS{speed}="480"
    SYSFS{version}=" 2.00"

There is no error report in dmesg about USB devices. It really looks like everything is just fine.

nnsg · 03-24-2005, 10:57 AM

Hi firebird,

It seems that you are accessing the usb key with "Low Performance USB Block driver" (ubx), maybe you can try to remove the module (it's under "device drivers -> block devices" in kernel config), and use usb mass storage (sdxx) instead.

MMYoung · 03-24-2005, 01:06 PM

Quote:

Originally posted by nnsg
Hi firebird,

It seems that you are accessing the usb key with "Low Performance USB Block driver" (ubx), maybe you can try to remove the module (it's under "device drivers -> block devices" in kernel config), and use usb mass storage (sdxx) instead.

EXACTLY! When you see your USB devices set up as ub whatever you have BORKED usb mass storage (which sets USB devices as SCSI). This is another reason NOT to use any stock Slack kernels above 2.6.7 as the UB option is selected by default, or if you use the stock kernel be sure and go through the config file and remove that option (it's under Block Devices not USB). The help file for UB tells you:

Quote:

Low Performance USB Block driver

CONFIG_BLK_DEV_UB:

This driver supports certain USB attached storage devices such as flash keys.

Warning: Enabling this cripples the usb-storage driver.

If unsure, say N.

It goes into a little more detail, but suffice it to say it can cause a lot of problems when it's enabled.

Later,
MMYoung

firebird · 03-24-2005, 07:18 PM

Thank you very much for the important information, guys! I know I must get something wrong.

Any chance that I can disable the UB driver without recompiling the Kernel?

Anyway, I'll take one step further: try to compile the kernel this weekend. If I do not mess up my installation completely, I'll let you know the result

MMYoung · 03-24-2005, 07:42 PM

Quote:

Originally posted by firebird
Any chance that I can disable the UB driver without recompiling the Kernel?

You might be able to, but I don't think so. Read this thread HERE, sorry.

Quote:

Originally posted by firebird
Anyway, I'll take one step further: try to compile the kernel this weekend. If I do not mess up my installation completely, I'll let you know the result

Thing about it, once you get your kernel compiled just the way you like it, you can always use the same config file each time, just check for any options marked as (NEW).

Later,
MMYoung

firebird · 03-24-2005, 07:54 PM

Quote:

Originally posted by MMYoung
You might be able to, but I don't think so. Read this thread HERE, sorry.

I just found that thread. Yes, sounds like re-compile is a must.

Quote:

Originally posted by MMYoung
Thing about it, once you get your kernel compiled just the way you like it, you can always use the same config file each time, just check for any options marked as (NEW).
[/B]

Hey, thank you! I feel much better about re-compiling kernel now. I'll try it once I get home.

firebird · 03-24-2005, 07:56 PM

Quote:

Originally posted by MMYoung
You might be able to, but I don't think so. Read this thread (thread 305552), sorry.

I also found that thread just now. Yes, sounds like re-compile is a must.

Quote:

Originally posted by MMYoung
Thing about it, once you get your kernel compiled just the way you like it, you can always use the same config file each time, just check for any options marked as (NEW).
[/B]

Hey, thank you! I feel much better about re-compiling kernel now. I'll try it as soon as I get home.

firebird · 03-24-2005, 07:59 PM

Err ... A message flash by looks like it does not allow me to post URL infected message, so I posted again. Sorry for the duplicate post.

eelriver · 03-24-2005, 10:27 PM

I must say this was one of the most thought out and well researched questions I've seen. All the information required was there, without too much filler. Good job.

If I was a moderator I'd let Firebird post urls early.

firebird · 03-26-2005, 04:43 AM

Problem solved! Well, almost.

In short: I download 2.6.11.5 from linux kernel archive and compile, using thread 127095 and this article (bitbenderforums) as reference. I disabled CONFIG_BLK_DEV_UB, as nnsg and MMYoung said (thank you very much!). After one night and one day, lots of reading and serverl re-compile, all my high-speed USB 2.0 devices works again - except the SanDisk one as I mentioned in the original post. Well, I guess I am at least as healthy as stock 2.4.29 kernel now

.

Compile and install new kernel is much easier than I have previously assumed, thanks very much to thread 127095, and my special thank goes to MMYoung for his encouragement

However, I was caught by 'make install' at first. It really sounds like the only (legal) way to install kernel at first, because it is mentioned everywhere, especially in README of kernel source code (newbie as me will treat it like the Bible). While 'make install' actually assumes I'm using LILO so it politely invoke /sbin/lilo for me (don't hate me for using GRUB

). What's really bad is I wrongly believe /sbin/lilo only update its map file and won't touch MBR (I really hope I have learned LILO well before I install GRUB). So I just get ahead re-arrange those files in /boot (copy around, rename, create symbol link etc.), update my GRUB menu and reboot. You may already know what happened: my system cannot boot! Well, after I reboot using SystemRecuseCD (I forget where I put my Slackware installation CD) and re-install GRUB to MBR, everything is just great!

Now back to my lousy SanDisk cruzer mini 256M pen drive. It's really a headache! This guy will be shown in M$ systems as "Removable Disk", just like the way my USB 2.0 card reader will be shown. So I think I can fixed it following this guide for card reader, i.e., turn on CONFIG_SCSI_MULTI_LUN and re-compile. In fact this does not fix my problem. Maybe SanDisk just hate Linux

Here is what happened when I connect that device:

#tail -f /var/log/messages

Code:

Mar 26 17:31:53 mixer kernel: usb 1-4: new high speed USB device using ehci_hcd and address 10
Mar 26 17:31:53 mixer kernel: usb 1-4: Product: Cruzer Mini
Mar 26 17:31:53 mixer kernel: usb 1-4: Manufacturer: SanDisk Corporation
Mar 26 17:31:53 mixer kernel: usb 1-4: SerialNumber: 000635722
Mar 26 17:31:53 mixer kernel: scsi9 : SCSI emulation for USB Mass Storage devices
Mar 26 17:31:58 mixer kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 1.04
Mar 26 17:31:58 mixer kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 26 17:31:58 mixer kernel: SCSI device sda: 512000 512-byte hdwr sectors (262 MB)
Mar 26 17:33:24 mixer kernel: usb 1-4: reset high speed USB device using ehci_hcd and address 10
Mar 26 17:33:24 mixer kernel: sda: Write Protect is off
Mar 26 17:33:24 mixer kernel: SCSI device sda: 512000 512-byte hdwr sectors (262 MB)
Mar 26 17:33:25 mixer kernel: sda: Write Protect is off
Mar 26 17:33:25 mixer kernel:  sda:<7>usb-storage: queuecommand called
Mar 26 17:33:25 mixer kernel: Attached scsi removable disk sda at scsi9, channel 0, id 0, lun 0
Mar 26 17:33:25 mixer kernel: Attached scsi generic sg0 at scsi9, channel 0, id 0, lun 0,  type 0
Mar 26 17:33:25 mixer udev[5428]: creating device node '/dev/sda'
Mar 26 17:33:25 mixer udev[5444]: configured rule in '/etc/udev/rules.d/udev.rules' at line 24 applied, 'sg0' becomes '%k'
Mar 26 17:33:25 mixer udev[5444]: creating device node '/dev/sg0'
Mar 26 17:33:25 mixer udev[5449]: configured rule in '/etc/udev/rules.d/000_local.rules' at line 22 applied, added symlink 'USB_sandisk256M'
Mar 26 17:33:25 mixer udev[5449]: creating device node '/dev/sda1'
Mar 26 17:35:07 mixer kernel: usb 1-4: reset high speed USB device using ehci_hcd and address 10
Mar 26 17:37:24 mixer kernel: : -- transfer complete
Mar 26 17:38:11 mixer kernel: usb 1-4: USB disconnect, address 10
Mar 26 17:38:11 mixer udev[5495]: removing device node '/dev/sg0'
Mar 26 17:38:11 mixer udev[5497]: removing device node '/dev/sda1'
Mar 26 17:38:11 mixer udev[5506]: removing device node '/dev/sda'

Please notice the time, it takes almost one and half minute to create the /dev/sda1 node (i.e., 1'30" before I can run mount), and another three minutes for me to mount it (i.e., the 'mount' command just hang there for three minutes).

This happens every time. I have re-partition the device again and again, try to format it using various FS (FAT, W95 FAT16, W95 FAT32 etc.) and under various system (Slackware, Knoppix, Win2k, WinXP), nothing helps.

Well, after that long wait, transfer rate is actually sweet, about 6.3 Mbyte/sec, that's faster than windows (about 3 Mbyte/sec).

Here's other relevant information:

#udevinfo -a -p `udevinfo -q path -n /dev/sda1`

Code:

looking at the device chain at '/sys/devices/pci0000:00/0000:00:0f.2/usb1/1-4':
  BUS="usb"
  ID="1-4"
  SYSFS{bConfigurationValue}="1"
  SYSFS{bDeviceClass}="00"
  SYSFS{bDeviceProtocol}="00"
  SYSFS{bDeviceSubClass}="00"
  SYSFS{bMaxPower}="100mA"
  SYSFS{bNumConfigurations}="1"
  SYSFS{bNumInterfaces}=" 1"
  SYSFS{bcdDevice}="0104"
  SYSFS{bmAttributes}="80"
  SYSFS{detach_state}="0"
  SYSFS{devnum}="11"
  SYSFS{idProduct}="7101"
  SYSFS{idVendor}="0781"
  SYSFS{manufacturer}="SanDisk Corporation "
  SYSFS{maxchild}="0"
  SYSFS{product}="Cruzer Mini "
  SYSFS{serial}="000635722"
  SYSFS{speed}="480"
  SYSFS{version}=" 2.00"

looking at the device chain at '/sys/devices/pci0000:00/0000:00:0f.2/usb1':
  BUS="usb"
  ID="usb1"
  SYSFS{bConfigurationValue}="1"
  SYSFS{bDeviceClass}="09"
  SYSFS{bDeviceProtocol}="01"
  SYSFS{bDeviceSubClass}="00"
  SYSFS{bMaxPower}="  0mA"
  SYSFS{bNumConfigurations}="1"
  SYSFS{bNumInterfaces}=" 1"
  SYSFS{bcdDevice}="0206"
  SYSFS{bmAttributes}="e0"
  SYSFS{detach_state}="0"
  SYSFS{devnum}="1"
  SYSFS{idProduct}="0000"
  SYSFS{idVendor}="0000"
  SYSFS{manufacturer}="Linux 2.6.11.5 ehci_hcd"
  SYSFS{maxchild}="5"
  SYSFS{product}="NEC Corporation USB 2.0"
  SYSFS{serial}="0000:00:0f.2"
  SYSFS{speed}="480"
  SYSFS{version}=" 2.00"

My udev rule:
#cat /etc/udev/rules.d/000_local.rules | grep 'sandisk'

Code:

BUS="usb", SYSFS{manufacturer}="SanDisk*", SYSFS{product}="Cruzer Mini*", KERNEL="sd?1",SYMLINK="USB_sandisk256M"

This time ohci_hcd, ehci_hcd and uhci_hcd are build into kernel.

It will be great if anyone can help me find the problem. To tell the truth, I'm considering give that device to my windows-only friends. I'll find one that work well with Linux. Any suggestion?

MMYoung · 03-28-2005, 06:37 AM

My tail -f /var/log/messages:

Code:

Mar 28 06:14:05 darkstar kernel: usb 1-4.4.3: new high speed USB device using ehci_hcd and address 10
Mar 28 06:14:05 darkstar kernel: scsi3 : SCSI emulation for USB Mass Storage devices
Mar 28 06:14:10 darkstar kernel:   Vendor: LEXAR     Model: JUMPDRIVE SECURE  Rev: 1000
Mar 28 06:14:10 darkstar kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 28 06:14:10 darkstar kernel: SCSI device sdc: 248928 512-byte hdwr sectors (127 MB)
Mar 28 06:14:10 darkstar kernel: SCSI device sdc: 248928 512-byte hdwr sectors (127 MB)
Mar 28 06:14:11 darkstar kernel:  sdc: sdc1
Mar 28 06:14:11 darkstar kernel: Attached scsi disk sdc at scsi3, channel 0, id 0, lun 0
Mar 28 06:14:11 darkstar udev[8322]: configured rule in '/etc/udev/rules.d/10-local.rules' at line 12 applied, added symlink 'jump128'
Mar 28 06:14:11 darkstar udev[8322]: configured rule in '/etc/udev/rules.d/10-local.rules' at line 12 applied, 'sdc' becomes '%k'
Mar 28 06:14:11 darkstar udev[8322]: creating device node '/dev/sdc'
Mar 28 06:14:11 darkstar udev[8325]: configured rule in '/etc/udev/rules.d/10-local.rules' at line 12 applied, added symlink 'jump128'
Mar 28 06:14:11 darkstar udev[8325]: configured rule in '/etc/udev/rules.d/10-local.rules' at line 12 applied, 'sdc1' becomes '%k'
Mar 28 06:14:11 darkstar udev[8325]: creating device node '/dev/sdc1'
Mar 28 06:14:11 darkstar fstab-sync[8329]: added mount point /media/MMYOUNG for /dev/sdc1

Don't know why it's taking your PC to create the node, mine took less than 10 seconds.
Here is the output of "udevinfo -a -p /sys/block/sdc"

Code:

    BUS="usb"
    ID="1-4.4.3"
    SYSFS{bConfigurationValue}="1"
    SYSFS{bDeviceClass}="00"
    SYSFS{bDeviceProtocol}="00"
    SYSFS{bDeviceSubClass}="00"
    SYSFS{bMaxPower}="100mA"
    SYSFS{bNumConfigurations}="1"
    SYSFS{bNumInterfaces}=" 1"
    SYSFS{bcdDevice}="2000"
    SYSFS{bmAttributes}="80"
    SYSFS{detach_state}="0"
    SYSFS{devnum}="10"
    SYSFS{idProduct}="a400"
    SYSFS{idVendor}="05dc"
    SYSFS{manufacturer}="LEXAR MEDIA"
    SYSFS{maxchild}="0"
    SYSFS{product}="JUMPDRIVE SECURE"
    SYSFS{serial}="5DD2BC09043509161004"
    SYSFS{speed}="480"
    SYSFS{version}=" 2.00"

Which was all I really needed to write this udev rule:

Code:

BUS="usb", SYSFS{product}="JUMPDRIVE SECURE", KERNEL="sd*?", NAME="%k", SYMLINK="jump128"

Main differences I can notice "up front" is:
1.) You're Cruzer Mini is being assigned first to sda

Code:

Mar 26 17:31:58 mixer kernel: SCSI device sda: 512000 512-byte hdwr sectors (262 MB)

then it's assigned to sg0

Code:

Mar 26 17:33:25 mixer kernel: Attached scsi generic sg0 at scsi9, channel 0, id 0, lun 0,  type 0

then udev "changes" it back to sda (if I'm reading all that right).

Code:

Mar 26 17:33:25 mixer udev[5449]: creating device node '/dev/sda1'

I'm not real sure what a "Cruzer Mini" is, but isn't it just a USB "thumb" drive, similar to my Lexar JumpDrive Secure? Or is it one of those multi-card readers?

Later,
MMYoung

firebird · 03-28-2005, 08:29 AM

Yes, "cruzer mini" is a thumb drive from SanDisk.

Though it looks like a thumb drive, its behavior is nothing like a thumb drive. I have several other thumb drives, none of them behave like this. I tends to believe SanDisk use a very different hardware interface.

Your log said:

Code:

Mar 28 06:14:11 darkstar kernel: Attached scsi disk sdc at scsi3, channel 0, id 0, lun 0

While mine said:

Code:

Mar 26 17:33:25 mixer kernel: Attached scsi removable disk sda at scsi9, channel 0, id 0, lun 0

This is similar to the behavior of my card reader. This is the log for my 4-slot card reader, (slot 3 is loaded):

Code:

Mar 28 22:15:31 mixer kernel: usb 1-4: new high speed USB device using ehci_hcd and address 4
Mar 28 22:15:31 mixer kernel: usb 1-4: Product: USB 2.0 Card Reader
Mar 28 22:15:31 mixer kernel: usb 1-4: SerialNumber: 00059
Mar 28 22:15:32 mixer kernel: scsi2 : SCSI emulation for USB Mass Storage devices
Mar 28 22:15:37 mixer kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 0130
Mar 28 22:15:37 mixer kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 28 22:15:37 mixer kernel: SCSI device sda: 498176 512-byte hdwr sectors (255 MB)
Mar 28 22:15:37 mixer kernel: sda: Write Protect is off
Mar 28 22:15:37 mixer kernel: SCSI device sda: 498176 512-byte hdwr sectors (255 MB)
Mar 28 22:15:37 mixer kernel: sda: Write Protect is off
Mar 28 22:15:37 mixer kernel:  sda:<7>usb-storage: queuecommand called
Mar 28 22:15:37 mixer udev[4487]: configured rule in '/etc/udev/rules.d/udev.rules' at line 24 applied, 'sg0' becomes '%k'
Mar 28 22:15:37 mixer udev[4487]: creating device node '/dev/sg0'
Mar 28 22:15:37 mixer udev[4485]: creating device node '/dev/sda'
Mar 28 22:15:37 mixer udev[4488]: creating device node '/dev/sda1'
Mar 28 22:15:37 mixer kernel: Attached scsi removable disk sda at scsi2, channel 0, id 0, lun 0
Mar 28 22:15:37 mixer kernel: Attached scsi generic sg0 at scsi2, channel 0, id 0, lun 0,  type 0
Mar 28 22:15:37 mixer kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 0130
Mar 28 22:15:37 mixer kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 28 22:15:38 mixer udev[4514]: creating device node '/dev/sdb'
Mar 28 22:15:38 mixer udev[4516]: configured rule in '/etc/udev/rules.d/udev.rules' at line 24 applied, 'sg1' becomes '%k'
Mar 28 22:15:38 mixer udev[4516]: creating device node '/dev/sg1'
Mar 28 22:15:38 mixer kernel: Attached scsi removable disk sdb at scsi2, channel 0, id 0, lun 1
Mar 28 22:15:38 mixer kernel: Attached scsi generic sg1 at scsi2, channel 0, id 0, lun 1,  type 0
Mar 28 22:15:38 mixer kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 0130
Mar 28 22:15:38 mixer kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 28 22:15:38 mixer kernel: SCSI device sdc: 498176 512-byte hdwr sectors (255 MB)
Mar 28 22:15:38 mixer kernel: sdc: Write Protect is off
Mar 28 22:15:38 mixer kernel: SCSI device sdc: 498176 512-byte hdwr sectors (255 MB)
Mar 28 22:15:38 mixer kernel: sdc: Write Protect is off
Mar 28 22:15:38 mixer kernel:  sdc:<7>usb-storage: queuecommand called
Mar 28 22:15:38 mixer kernel: Attached scsi removable disk sdc at scsi2, channel 0, id 0, lun 2
Mar 28 22:15:38 mixer kernel: Attached scsi generic sg2 at scsi2, channel 0, id 0, lun 2,  type 0
Mar 28 22:15:38 mixer kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 0130
Mar 28 22:15:38 mixer kernel:   Type:   Direct-Access                      ANSI SCSI revision: 00
Mar 28 22:15:38 mixer udev[4564]: creating device node '/dev/sdc'
Mar 28 22:15:38 mixer udev[4566]: configured rule in '/etc/udev/rules.d/udev.rules' at line 24 applied, 'sg2' becomes '%k'
Mar 28 22:15:38 mixer udev[4568]: creating device node '/dev/sdd'
Mar 28 22:15:38 mixer udev[4566]: creating device node '/dev/sg2'
Mar 28 22:15:38 mixer udev[4567]: creating device node '/dev/sdc1'
Mar 28 22:15:38 mixer udev[4575]: configured rule in '/etc/udev/rules.d/udev.rules' at line 24 applied, 'sg3' becomes '%k'
Mar 28 22:15:38 mixer udev[4575]: creating device node '/dev/sg3'
Mar 28 22:15:38 mixer kernel: Attached scsi removable disk sdd at scsi2, channel 0, id 0, lun 3
Mar 28 22:15:38 mixer kernel: Attached scsi generic sg3 at scsi2, channel 0, id 0, lun 3,  type 0

You can see /dev/sg? nodes also created for card reader.

If I load card AFTER the card reader be connected to the computer, /dev/sdc1 will not be created. i.e., the system won't detect a memory card is loaded. this article has mentioned this problem.

So I believe SanDisk use card reader-like interface for its thumb drive, it behaves somewhat like a card reader whose memory card is loaded AFTER being conected to the computer. I think I can solve the problem for SanDisk as soon as I solve the memory card detection problem for my card reader

firebird · 02-13-2006, 08:06 PM

I'm gladly report here that I my "cruzer mini" SanDisk USB 2.0 drive worked with kernel 2.6.14.7. I don't know which kernel version it first start to work, it should be the 2.6.14.x series, I remember I checked it in 2.6.13.x, it still not work.

Thanks everyone here for help. I guess I also owe my thanks to USB driver developers.

P.S.
Sorry for bringing up an old post, just in case someone found this using search.