HOWTO use iRiver iFP players w/UMS firmware with 2.6 kernel

[Thalion]

I thought it was a wise idea to write this 'cause the original thread where the solution was posted is quite long and cluttered with screen dumps and all that. Here is just a simple instruction which will (hopefully) make your player work without the need for ifp-line and other third-party tools.

First, grab the iFP UMS firmware from their website (http://www.iriver.com) if you haven't already. If you use ifp-line, you can perform the upgrade as follows:

Code: Select all

# ifp firmupdate FIRMWARE.HEX 

BE WARNED: if anything goes wrong, you WON'T be able to revert this change from Linux with 2.6 kernel (you simply won't be able to upload the non-UMS firmware on it), so make sure you have a PC with Windows or 2.4-based Linux nearby (or in dual boot).

Now the kernel. Most likely you already have it lying somewhere in /usr/src/linux-2.6.x, with /usr/src/linux symlinked to it. If not, then emerge it (and don't forget to make the symlink!). I've tested the patch with development-sources-2.6.1, so I'm not sure it works with patched versions (gentoo, love or whatever). So if it works for you, tell the others! =)

Here goes the patch. Copy&paste the following code and save it as /usr/src/ifp-ums-2.6.patch:

Code: Select all

--- linux/include/scsi/scsi_devinfo.h.orig	2004-01-06 01:00:29.000000000 +0100
+++ linux/include/scsi/scsi_devinfo.h	2004-01-06 20:15:50.036892568 +0100
@@ -19,4 +19,5 @@
 #define BLIST_MS_SKIP_PAGE_08	0x2000	/* do not send ms page 0x08 */
 #define BLIST_MS_SKIP_PAGE_3F	0x4000	/* do not send ms page 0x3f */
 #define BLIST_USE_10_BYTE_MS	0x8000	/* use 10 byte ms before 6 byte ms */
+#define BLIST_NORMB     	0x10000	/* Known to be not removable */
 #endif
--- linux/drivers/scsi/scsi_scan.c.orig	2004-01-06 01:00:29.000000000 +0100
+++ linux/drivers/scsi/scsi_scan.c	2004-01-06 20:10:19.731106680 +0100
@@ -536,7 +536,8 @@
 		sdev->online = FALSE;
 	}
 
-	sdev->removable = (0x80 & inq_result[1]) >> 7;
+	sdev->removable = (((0x80 & inq_result[1]) >> 7) &&
+		!(*bflags & BLIST_NORMB));
 	sdev->lockable = sdev->removable;
 	sdev->soft_reset = (inq_result[7] & 1) && ((inq_result[3] & 7) == 2);
 
--- linux/drivers/scsi/scsi_devinfo.c.orig	2004-01-06 01:00:29.000000000 +0100
+++ linux/drivers/scsi/scsi_devinfo.c	2004-01-06 20:13:46.890613648 +0100
@@ -183,6 +183,7 @@
 	{"SGI", "TP9500", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"MYLEX", "DACARMRB", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"XYRATEX", "RS", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
+	{"iRiver", "iFP Mass Driver", NULL, BLIST_NORMB},
 	{ NULL, NULL, NULL, 0 },
 };
 

Make sure the lines don't wrap. Now open the terminal, go to /usr/src, and from there run:

Code: Select all

# patch -p0 < ifp-ums-2.6.patch

It should not complain (if it does, then either you're doing something wrong or your kernel is not compatible, or maybe you're doing it on 2.6.2 and the kernel guys have already fixed it by then). Then, cd linux, and rebuild and install the kernel as usual (make clean seems to be unneeded). After reboot, you can try to mount your player. It should be the device /dev/sda (note: not /dev/sda1!) if it is the only UMS device plugged. As root, do:

Code: Select all

# mkdir /mnt/player
# mount /dev/sda /mnt/player
# ls /mnt/player

If this doesn't tell you anything about "bad superblock", than it should work from now on.

Hope this helps

[numerodix]

I first tried mounting my player without this and I got the superblock error. But the patch wouldn't work, neither with gentoo-dev-sources-2.6.1 nor development-sources-2.6.1-rc1... I always get this output:

Code: Select all

#  patch -p0 < ifp-ums-2.6.patch 
patching file linux/include/scsi/scsi_devinfo.h
Hunk #1 FAILED at 19.
1 out of 1 hunk FAILED -- saving rejects to file linux/include/scsi/scsi_devinfo.h.rej
patching file linux/drivers/scsi/scsi_scan.c
Hunk #1 FAILED at 536.
1 out of 1 hunk FAILED -- saving rejects to file linux/drivers/scsi/scsi_scan.c.rej
patching file linux/drivers/scsi/scsi_devinfo.c
Hunk #1 FAILED at 183.
1 out of 1 hunk FAILED -- saving rejects to file linux/drivers/scsi/scsi_devinfo.c.rej

[Thalion]

Are you sure you have /usr/src/linux symlinked to the kernel you 're patching?

Also, try vanilla 2.6.1 final - that's what I tested it on. As I said, I'm not sure gentoo-dev-sources will work.

[numerodix]

Yes, I did have that setup at the time of testing.

Do you know if this is something that might be included in a newer kernel release?

[funkmankey]

patch also failed here, against 2.6.2-rc1-mm3 -- dnno why, the source files look similar enough that there should not be a problem.

anyway, I modified the few lines manually, rebuilt, and now my 180T is working just fine as a usb storage device. thanks!

[pilla]

Moved from H&L

[tamran]

The patch worked on gentoo-dev-sources 2.6.2

[unfo-]

I've 2.6.5 kernel from kernel.org and it's worked fine except for

Code: Select all

bash-2.05b# patch -p0 < ifp-ums-2.6.patch
patching file linux/include/scsi/scsi_devinfo.h
Hunk #1 FAILED at 19.
1 out of 1 hunk FAILED -- saving rejects to file linux/include/scsi/scsi_devinfo.h.rej
patching file linux/drivers/scsi/scsi_scan.c
Hunk #1 FAILED at 536.
1 out of 1 hunk FAILED -- saving rejects to file linux/drivers/scsi/scsi_scan.c.rej
patching file linux/drivers/scsi/scsi_devinfo.c
Hunk #1 FAILED at 183.
1 out of 1 hunk FAILED -- saving rejects to file linux/drivers/scsi/scsi_devinfo.c.rej
bash-2.05b#

I have a iFP-190TC with 1.11 UMS.
I've tried also the trick of changing the '0x10000' to '0x12000' which was mentioned in another thread, but it failed also.

when I plug the device in tail -f /var/log/messages shows:

Code: Select all

Apr 20 23:02:52 baileys Attached scsi removable disk sda at scsi2, channel 0, id 0, lun 0
Apr 20 23:02:52 baileys Attached scsi generic sg0 at scsi2, channel 0, id 0, lun 0,  type 0
Apr 20 23:02:52 baileys usb-storage: queuecommand called
Apr 20 23:02:52 baileys usb-storage: *** thread awakened.
Apr 20 23:02:52 baileys usb-storage: Bad LUN (0:1)
Apr 20 23:02:52 baileys usb-storage: scsi cmd done, result=0x40000
Apr 20 23:02:52 baileys usb-storage: *** thread sleeping.
... [[This continues from Bad LUN (0:1)  to Bad LUN (0:7)]]
Apr 20 23:02:52 baileys USB Mass Storage device found at 4

And when I try to mount:

Code: Select all

bash-2.05b# mount -t vfat /dev/sda /mnt/usb
mount: /dev/sda: can't read superblock

I've built most SCSI-related things, which could remotely relate to USB-devices, as modules and also most of the USB-support is built into the kernel.

Any suggestions?

// unfo-

[mank]

Here's how I got the wonderful iFP-390T working as USB Mass Storage

ISSUE 1: Iriver's iFP-390T does not work with kernel-2.6 out of the box due to some issues with current SCSI/USB code in kernel 2.6

The error I got was: 'Unable to read device/partition'

To resolve this issue apply the patch below on kernel 2.6. Recompile the kernel with scsi-emulation/USB etc.

Code: Select all

--- linux/include/scsi/scsi_devinfo.h.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/include/scsi/scsi_devinfo.h	2004-01-13 04:12:19.509266640 +0100
@@ -19,4 +19,5 @@
 #define BLIST_MS_SKIP_PAGE_08	0x2000	/* do not send ms page 0x08 */
 #define BLIST_MS_SKIP_PAGE_3F	0x4000	/* do not send ms page 0x3f */
 #define BLIST_USE_10_BYTE_MS	0x8000	/* use 10 byte ms before 6 byte ms */
+#define BLIST_NORMB	            0x10000 /* Known to be not removable */
 #endif
--- linux/drivers/scsi/scsi_scan.c.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/drivers/scsi/scsi_scan.c	2004-01-13 04:13:07.728936136 +0100
@@ -536,7 +536,8 @@
 		sdev->online = FALSE;
 	}
 
-	sdev->removable = (0x80 & inq_result[1]) >> 7;
+	sdev->removable = (((0x80 & inq_result[1]) >> 7) &&
+		!(*bflags & BLIST_NORMB));
 	sdev->lockable = sdev->removable;
 	sdev->soft_reset = (inq_result[7] & 1) && ((inq_result[3] & 7) == 2);
 
--- linux/drivers/scsi/scsi_devinfo.c.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/drivers/scsi/scsi_devinfo.c	2004-01-13 04:13:40.591940200 +0100
@@ -183,6 +183,7 @@
 	{"SGI", "TP9500", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"MYLEX", "DACARMRB", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"XYRATEX", "RS", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
+	{"iRiver", "iFP Mass Driver", NULL, BLIST_NORMB},
 	{ NULL, NULL, NULL, 0 },
 };

To apply this patch, copy the above code to file 'filename.patch' and save the file in /usr/src. Then issue the following command: patch -p0 < filename.patch

Boot with the new kernel.

ISSUE 2: After I got iFP to mount properly, all I could see inside the FS was junk characters.

This is believe is due to non-standard FAT partition.
To resolve this, delete all the partitions in iFP with gnu fdisk and recreate a VFAT partition with gnu fdisk. Next, format this partition with mkfs.vfat (use -I option).

--mank

Please note that I used iFP-390T on kernel-2.6.4 and have played with iFP for only an hour.

[numerodix]

Has anyone else had noticed that sometimes when you replace your songs with new one, the new songs will occasionally switch the the old songs (in the middle of playback) that are supposed to have been erased? Apparently sometimes overwriting files isn't entirely successful and the files "underneath" still remain in part. I'm speculating on the reason here but what else could it be?

Anyway, I've been using mostly linux lately and I can't quite say if that might be it but it certainly has been trying to get the player to connect properly, it seems windows handles the fat filesystems a lot better. Speaking of fs, is it supposed to be fat or fat32?

Anyway, to remedy this fragmentation problem I tried erasing the whole memory with a simple
dd if=/dev/zero of=/dev/sda bs=1024k
which writes zero's on the drive. I then copied file onto it and so far no anomalies but I would hate to find out I have to do this everytime I want to replace my songs.

[unfo-]

To resolve this issue apply the patch below on kernel 2.6. Recompile the kernel with scsi-emulation/USB etc.

Code: Select all

--- linux/include/scsi/scsi_devinfo.h.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/include/scsi/scsi_devinfo.h	2004-01-13 04:12:19.509266640 +0100
@@ -19,4 +19,5 @@
 #define BLIST_MS_SKIP_PAGE_08	0x2000	/* do not send ms page 0x08 */
 #define BLIST_MS_SKIP_PAGE_3F	0x4000	/* do not send ms page 0x3f */
 #define BLIST_USE_10_BYTE_MS	0x8000	/* use 10 byte ms before 6 byte ms */
+#define BLIST_NORMB	            0x10000 /* Known to be not removable */
 #endif
--- linux/drivers/scsi/scsi_scan.c.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/drivers/scsi/scsi_scan.c	2004-01-13 04:13:07.728936136 +0100
@@ -536,7 +536,8 @@
 		sdev->online = FALSE;
 	}
 
-	sdev->removable = (0x80 & inq_result[1]) >> 7;
+	sdev->removable = (((0x80 & inq_result[1]) >> 7) &&
+		!(*bflags & BLIST_NORMB));
 	sdev->lockable = sdev->removable;
 	sdev->soft_reset = (inq_result[7] & 1) && ((inq_result[3] & 7) == 2);
 
--- linux/drivers/scsi/scsi_devinfo.c.orig	2004-01-13 04:03:19.000000000 +0100
+++ linux/drivers/scsi/scsi_devinfo.c	2004-01-13 04:13:40.591940200 +0100
@@ -183,6 +183,7 @@
 	{"SGI", "TP9500", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"MYLEX", "DACARMRB", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
 	{"XYRATEX", "RS", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
+	{"iRiver", "iFP Mass Driver", NULL, BLIST_NORMB},
 	{ NULL, NULL, NULL, 0 },
 };

To apply this patch, copy the above code to file 'filename.patch' and save the file in /usr/src. Then issue the following command: patch -p0 < filename.patch

Boot with the new kernel.

As I pointed in my posting, I've tried this patch and it resolved in errors and I've yet to have gotten this to work

[unfo-]

I got my iRiver functioning with my 2.6.5 kernel when I edited the files mentioned in the patch manually.
I just added the {"iRiver..... line to the end of the list and changed those other things in the appropriate lines.
make & boot & onto happiness

thanks a bunch all

[fasaxc]

Just filed a bug report for this problem. Hopefully we wont have to patch any more. http://bugs.gentoo.org/show_bug.cgi?id=56802

-Shaun

[fasaxc]

Here's how to fix if you get gibberish files on the player. (after patching the kernel)

WARNING: you will lose all the data on the player

First log out of kde etc and drop to a console (this ensures no program is accessing the player while we format it). And su to root.

Code: Select all

su

If you have the player mounted say in /mnt/player/ then unmount it now

Code: Select all

 umount /mnt/player 

Now be sure you know which device node your player is connected to, if you have no scsi or other usb disks connected it will most likely be /dev/sda. The following will tell you what sd? devices are connected.

Code: Select all

 cd /dev/ && ls sd* 

The one we're looking for will just be sda, sdb or sdc etc. There will be no accompanying sda1, sda2 etc.

Now assuming sda is the correct node we need to run fdisk to delete the partitions on the disk.

Code: Select all

fdisk /dev/sda

Press d then enter then 1 to delete partition 1 then
Press d then enter then 2 to delete partition 2 then
Press d then enter then 3 to delete partition 1 then
Press d then enter to delete partition 4

then press w then enter to exit.

Now run cfdisk to create the new partition

Code: Select all

cfdisk /dev/sda

Select new with the arrow keys and enter, then 'Primary' and I think leave the size as is.

Select 'TYPE' and checge to 06 (ie FAT16)

Select 'write' to write and exit the program.

Now run mkfs.vfat to make a new file system on the disk (emerge sys-fs/dosfstools). (the -I is neccessary to make mkfs treat the sda node as if it were a sda1 node which is the usual convention for partitions)

Code: Select all

mkfs.vfat /dev/sda -I

Now mount the file system straight away (assuming /mnt/player exists), move to it's folder and create the tuner and voice folders. Then unmount again

Code: Select all

mount /dev/sda /mnt/player
cd /mnt/player
mkdir tuner
mkdir voice
cd /
umount /mnt/player

Now remove the player and reattatch to see if it all worked.

On a side note -- if you use udev you can reconfigure your device nodes so the player always appears as say /dev/iriver-player even if you have several usb drives installed eg a dig. camera, card reader etc.

See http://webpages.charter.net/decibelshel ... l#lilogrub
for how to enable udev.
Then put this in /etc/udev/rules.d/10-local.rules

Code: Select all

# IRiver
BUS="scsi", SYSFS{model}="iFP Mass Driver", KERNEL="sd?", NAME="%k", SYMLINK="iriver-player%n"

[CybeRDukE]

Hi,

i just want to mention that 2 days ago, the ums firmware with OGG has been released.

and i created a patch for the current gentoo-dev-sources (2.6.7-r12). here we go

Code: Select all

diff -Nru linux-2.6.7-gentoo-r12.org/drivers/scsi/scsi_devinfo.c linux/drivers/scsi/scsi_devinfo.c
--- linux-2.6.7-gentoo-r12.org/drivers/scsi/scsi_devinfo.c      2004-08-07 23:20:29.000000000 +0200
+++ linux/drivers/scsi/scsi_devinfo.c   2004-08-07 23:44:39.535972056 +0200
@@ -155,6 +155,7 @@
        {"IOMEGA", "Io20S         *F", NULL, BLIST_KEY},
        {"INSITE", "Floptical   F*8I", NULL, BLIST_KEY},
        {"INSITE", "I325VM", NULL, BLIST_KEY},
+        {"iRiver", "iFP Mass Driver", NULL, BLIST_NORMB},
        {"LASOUND", "CDX7405", "3.10", BLIST_MAX5LUN | BLIST_SINGLELUN},
        {"MATSHITA", "PD-1", NULL, BLIST_FORCELUN | BLIST_SINGLELUN},
        {"Medion", "Flash XL  MMC/SD", "2.6D", BLIST_FORCELUN},
diff -Nru linux-2.6.7-gentoo-r12.org/drivers/scsi/scsi_scan.c linux/drivers/scsi/scsi_scan.c
--- linux-2.6.7-gentoo-r12.org/drivers/scsi/scsi_scan.c 2004-08-07 23:20:29.000000000 +0200
+++ linux/drivers/scsi/scsi_scan.c      2004-08-07 23:43:54.629798832 +0200
@@ -565,7 +565,7 @@
         */ 
 
        sdev->inq_periph_qual = (inq_result[0] >> 5) & 7;
-       sdev->removable = (0x80 & inq_result[1]) >> 7;
+       sdev->removable = (((0x80 & inq_result[1]) >> 7) && !(*bflags & BLIST_NORMB));
        sdev->lockable = sdev->removable;
        sdev->soft_reset = (inq_result[7] & 1) && ((inq_result[3] & 7) == 2);
 
diff -Nru linux-2.6.7-gentoo-r12.org/include/scsi/scsi_devinfo.h linux/include/scsi/scsi_devinfo.h
--- linux-2.6.7-gentoo-r12.org/include/scsi/scsi_devinfo.h      2004-08-07 23:20:29.000000000 +0200
+++ linux/include/scsi/scsi_devinfo.h   2004-08-07 23:42:45.819259624 +0200
@@ -24,4 +24,5 @@
 #define BLIST_REPORTLUN2       0x20000 /* try REPORT_LUNS even for SCSI-2 devs
                                           (if HBA supports more than 8 LUNs) */
 #define BLIST_NOREPORTLUN      0x40000 /* don't try REPORT_LUNS scan (SCSI-3 devs) */
+#define BLIST_NORMB             0x80000 /* Known to be not removable */
 #endif

i had to assign a new value for the BLIST_NORMB definition, but seems to work here.

now have fun using ogg and linux

[gazoombo]

Yay!!
The earlier patch had been getting rejected for a while, and I didn't have time to manually fix it. Thanks for the new patch, and the heads-up on the new firmware I've been waiting for.

[PinkyDaemon]

Just for you to know: With kernel linux-2.6.8-gentoo-r3 the player seems to be detected correctly as /dev/sda without patching. I had the same playback-problems as stated below (player skips to old songs which are supposed to be deleted). A solution seems to be the "sync" option in the fstab file (slow but seems to work perfectly)

prost,
PinkyDaemon

[gazoombo]

Yes, 2.6.8 added support for this device (finally!). When I mount it I always manually sync it before unmounting, just in case.

[Estariel]

hi... i have it somewhat working but need some advice... (ifp 390 T)

i use a 2.6.8 kernel so i didn't have to patch it.

if i plug in the device for the first time after a reboot, it gets detected correctly (dmesg says that). I can then mount it and transfer songs to it.

Now i have two problems:
- Some OGG files encoded at quality 7 only play for about 12 seconds on the player. the player will then play the next song.
I assume there is some highest possible bitrate and my q7 songs have a higher bitrate than that. Is there some way to convert them to Q4? i don't want to have to insert each CD again...

- I cannot unmount the player. the system always states that the dive was "busy". Thus, i have to just plug it off.
If i try to reattach the player, it doesn't get recoginzed any more. (dmesg doesn't show a thing, but the /dev/sda device node vanishes)
I have to reboot the system to be able to reattach the player. this is my biggest problem, so please help me there...

thanks!

[Sastraxi]

Try something like ogg123 | oggenc?

[gazoombo]

About the unmounting, make sure that no users are 'in' or browsing the mounted filesystem. Also you may have to wait for the drive to sync when you unmount (USB 1.1 takes a while to xfer data).

[kamagurka]

i always thought iRiver players were simply recognised as generic USB Mass Storage; why all this higgelty piggelty?
also, why would i need special firmware?

[kpoman]

hi all!
i am running linux 2.6.8.1-mm4, and it worked great (wasn't working with previous kernels) so it rocks !!!

but have a question, because i want to upgrade the firmware; last time i upgraded my firmware, i found out that the new one wouldnt allow "line-in" recording with 96kbps or higher ! that sucks !!!

i think they did that to disallow piracy; but it is a very bad idea, because i use my iriver to record my music group and i want higher quality (256kbps) so, i dont know if i should upgrade to the last one (ums + ogg + mp3) ...

can someone tell me please what is the higher bitrate i can set when using line-in ?


thanks a lot (dont wanna upgrade then downgrade, last time i did i almost killed my player)

[badchien]

i always thought iRiver players were simply recognised as generic USB Mass Storage; why all this higgelty piggelty?
also, why would i need special firmware?

I think it depends on the model. My iFP390T came with "Managed" firmware meaning you can only upload/download files to it using an iriver provided app in windows or mac os. Also for both managed and UMS there is either the MP3/WMA/ASF version or the MP3/OGG version because the player can't support all the formats at once. I switched to the UMS MP3/OGG firmware which makes the device behave like any other UMS device and therefore makes it accessible in Linux so I can upload my OGGs.

hi all!...
can someone tell me please what is the higher bitrate i can set when using line-in ?

I don't think you said which model you have? Anyway I have a iFP390T with the latest UMS-OGG firmware (v1.22) from 06 August 2004 and I can confirm that the line-in bitrate is limited to no higher than 96kbs for 44.1MHz. I haven't used line-in recording yet, but I agree, this sucks.

[kpoman]

yes, what a lame tech solution for disallowing mp3 piracy; they really badly suck for that :/
we just have to keep with the old firmware in order to record "decently" ... REVOLUTION NOW !!!