USB Storage ported to 5xxx ROM\'s ?

[Mitch]

But you still need the drivers for syncing.

[lardman]

Yes and no (if you can perform the sync using the data file from the Z on the PC a la Ko/Pi then no).

But I don\'t want to be able to sync, just to easily transfer information when I go to visit someone else\'s office. I don\'t want to have to install drivers on their computer (if they\'d even let me, or if they even had sufficient privileges to do so), just to plug it in and copy back and forth.


Si

[lardman]

I\'ve compared the 5600 and C860 kernel source and there are almost no differences (the only ones are related to being able to set a packet size as a parameter), so I guess that once ced and greg2 grab my corrected scripts they will be able to use mass storage on their 5600s.

Not ideal for my quest to get it working on the 5500 (I was hoping that it wouldn\'t work on the 5600 and that the differences between the two sets of source would be larger and would show me what to add to the 5500 kernel), but it\'ll make 5600 users happy.


Si

P.S. I\'m getting a bit ahead of myself here, no-one has actually got it working yet (my initial scripts had a typo), but I\'m 90% sure it will work.

[ced]

so how come the pxa_bi.o from the C860 source doesn\'t work on the SL5600?

[lardman]

I don\'t know, but there really are very few differences. There\'s a line in the makefile.in and config.in files (I think those are the right names) which specify that storage_fd should be built, and in the pxa.c file there are only the differences which are due to the addition of the module parameters.

Perhaps I\'ll have to dig deeper and see whether anything else is linked in (I\'m pretty sure there\'s nothing else) or try to take a look at the various #defines which may affect it.


Si

[maslovsky]

I also think that USB storage will work on 5600 - they are using the same USB support module.

[lardman]

It worked! On my 5500.

God only knows why it wouldn\'t work yesterday (I\'ve not changed anything).

Anyway, my storage_fd.o module + scripts has allowed my 256Mb CF card to be recognised and mounted on my Linux box.

Time to try WinXP.....

 
Si

P.S. The issue may have been that I hadn\'t started the storage_fd.o module before I put my Z in the cradle (or that I booted my desktop Linux box with it already in the cradle, etc.)

[lardman]

Yep, WinXP works too.

Read and write both work fine. Under Linux the device is /dev/sda (no numbers).

I suppose I should make a package for people to install. Maslovsky & ced, you\'ve both used the C860 hotplug scripts, do you think these are the best way to go (rather than my quick hack method of using 2 separate scripts to start and stop the storage funtion)?


Si

[lardman]

Looks like the above file is also the one to use for the 5600.

Here are the commands I used to compile it (after sourcing my standard GCC2.95.3 cross toolchain script):

cd whichever/dir/has/the/source/in/it

arm-linux-gcc -D__KERNEL__ -I/home/simon/dev/OE/oe-packages-exported/tmp/base/linux-2.4.18-rmk7-pxa3-embedix-r0/linux/include -Wall -Wstrict-prototypes -Wno-trigraphs -Os -mapcs -fno-strict-aliasing -fno-common -fno-common -pipe -mapcs-32 -march=armv4 -mtune=strongarm1100 -mshort-load-bytes -msoft-float -DMODULE  -DKBUILD_BASENAME=storage   -c -o storage_fd.o storage-fd.c

arm-linux-gcc -D__KERNEL__ -I/home/simon/dev/OE/oe-packages-exported/tmp/base/linux-2.4.18-rmk7-pxa3-embedix-r0/linux/include -Wall -Wstrict-prototypes -Wno-trigraphs -Os -mapcs -fno-strict-aliasing -fno-common -fno-common -pipe -mapcs-32 -march=armv4 -mtune=strongarm1100 -mshort-load-bytes -msoft-float -DMODULE  -DKBUILD_BASENAME=storage   -c -o schedule_task.o schedule_task.c

arm-linux-gcc -D__KERNEL__ -I/home/simon/dev/OE/oe-packages-exported/tmp/base/linux-2.4.18-rmk7-pxa3-embedix-r0/linux/include -Wall -Wstrict-prototypes -Wno-trigraphs -Os -mapcs -fno-strict-aliasing -fno-common -fno-common -pipe -mapcs-32 -march=armv4 -mtune=strongarm1100 -mshort-load-bytes -msoft-float -DMODULE  -DKBUILD_BASENAME=storage   -c -o storageproto.o storageproto.c

arm-linux-ld -r -a storage_fd_drv.o storage_fd.o storageproto.o schedule_task.o

mv storage_fd_drv.o storage_fd.o



Si

[rrusaw]

Code: [Select]

+++ sa1100.c    2004-02-27 19:19:34.628738128 -0500

@@ -440,6 +440,7 @@

                                         udc_interrupts, *(UDCSR), *(UDCCS0), *(UDCAR));



                        usbd_device_event (udc_device, DEVICE_RESET, 0);

+                       usbd_device_event (udc_device, DEVICE_ADDRESS_ASSIGNED,0);

                }



                if (status & UDCSR_SUSIR) {

To compile i used the 860 kernel source with the drivers/usb/device/bi directory from the 5500 kernel source. I then patched the sa1100.c driver as above (1 line)  copied the sl5500 config (arch/arm/def-configs/collie) to .config. I then manually copied the CONFIG_USBD_STORAGE section from the boxer config into the .config.

After that it was a simple matter of building the kernel and modules and copying the files over to the Z and using modified versions of lardman\'s scripts to load the appropiate modules.

Cheers and good luck

Ryan

[Anonymous]

Nice work Si!

[Anonymous]

Hah, very strange, but after reading your post I\'ve tried it again at my home PC. IT WORKED! Something could be wrong with my PC at work, but WinXP was able to recognize CF card in SL5500 just fine.

I\'m using hotplug scripts from c860 root fs. I think I can write a simple application that will alow to switch between USB storage and network modes.
I will build an IPK when I get back to work on Monday.

Regards,
maslovsky

[Anonymous]

Bad news - that only worked once. Next time I connect zaurus windows does not see  a CF card being inserted. Just like before.
I cannot figure out if this is a zaurus or windows problem.

maslovsky

[lardman]

It seems to be working every time for me now.

With my Z out of the cradle I run my script to insmod storage_fd, then just pop it in the cradle and it works. I may have switched off between insmoding and sticking it in the cradle actually, but ever since I can take it out and put it back in without any troubles. I also tend to not boot with the Z in the cradle, this might make a difference, I\'m not sure.

Something else to check is that the card can be cleanly umounted.


Si

[lardman]

Right, I\'ve managed to replicate your troubles. To get around it I have to do the following:

take Z out of cradle (might not be necessary)
run script to insmod storage_fd
suspend/switch off
turn back on and place in the cradle.

Strange.


Si