![]() ![]() |
Mar 22 2004, 04:35 PM
Post
#31
|
|
|
Group: Members Posts: 48 Joined: 8-December 03 Member No.: 59 |
Bad news on my end. My Simpletech SD does not like to be ext2 and treated as home. I get similar findings as n8a gets with his SanDisk. Oh well... just so you know, Derek, it all looks good and runs fine until you suspend!
I'll have to break down and get a Lexar! I want to run KO-PI! |
|
|
|
Mar 22 2004, 04:42 PM
Post
#32
|
|
|
Group: Members Posts: 153 Joined: 16-January 04 Member No.: 1,464 |
Revdjenk, there's another thread around here on SD card speed, where they were talking about an alternate SD driver. It's available under the downloads section, it seems to solve a bunch of problems. Try it and let me know if it solves your problems.
|
|
|
|
Mar 23 2004, 06:58 PM
Post
#33
|
|
|
Group: Members Posts: 1 Joined: 23-July 03 Member No.: 296 |
eww.. thanks for the heads up.. was just getting ready to try this kernel myself..
QUOTE Well, as much fun as it's been to use the asyncio kernel, I just encountered a dealbreaker... at least for the present. It seems that the asyncio kernel has affected the serial_cs.o, to the effect that I get the following messages when installing bluetooth (as usual on tkcROM):
<code snip> Any ideas here, Derek? I'd like to get this working, but replacing the existing driver with another copy of serial_cs.o that I have (then doing a depmod -a and modprobe) did not seem to work, so I'm off to the original tkcROM zImage for now... |
|
|
|
Mar 24 2004, 01:05 PM
Post
#34
|
|
|
Group: Members Posts: 12 Joined: 3-August 03 Member No.: 323 |
I am using a sandisk sd card and the tkrom a3 with the prep-home script and it has been work great for about a week now no problems at all and witht he sd mounted via loopback I cannot really tell any difference from using main memory for /home. Thanks for a great script.
Colin |
|
|
|
Mar 31 2004, 12:21 PM
Post
#35
|
|
|
Group: Members Posts: 235 Joined: 5-January 04 From: Chicago, Illinois Member No.: 1,359 |
QUOTE It looks like the serial driver is compiled into the tkc kernel. When I compiled the async io kernel, I left most of the config options set to default, which ends up putting the serial driver as a module. I've attached the serail.o module, try insmoding this one before loading up your serial_cs.o modules and let me know if it works. derek, I have the file and finally have my 5500 back in the configuration to try this out, but I'm not sure of how to insmod the serial.o file. I have copied the file to the following directory: CODE # pwd
/home/root/modules/2.4.18-rmk7-pxa3-embedix/pcmcia # ls memory_cs.o prism2_cs.o serial_cs.o spectrum_cs.o orinoco_cs.o serial.o sharpzdc_cs.o # I then did # insmod serial.o, which just returned me to the prompt. What can I do to get this to work as you explained above? Thanks in advance for the baby steps |
|
|
|
Mar 31 2004, 12:35 PM
Post
#36
|
|
|
Group: Members Posts: 123 Joined: 4-March 04 From: Toronto, Canada Member No.: 2,156 |
n8a - It took me a while, but I figured it out. This is what I did:
insmod serial.o insmod serial_cs.o Installing serial.o first loads the symbols required to load serial_cs.o. You can do a lsmod to see what modules are installed. If everything works after this let me know. I figured out how to edit the modules file so that this loads automatically. |
|
|
|
Mar 31 2004, 04:36 PM
Post
#37
|
|
|
Group: Members Posts: 23 Joined: 8-December 03 From: Wellington, New Zealand Member No.: 1,084 |
Not sure if others have the same experience, but the my Netgear MA701 wasn't recognised by the async kernel. (cardinfo would return a result, but there was no kernel output to dmesg on card insert).
Swapped in the 'standard' kernel that ships with TKC ROM and the card works fine. |
|
|
|
Mar 31 2004, 06:24 PM
Post
#38
|
|
|
Group: Members Posts: 235 Joined: 5-January 04 From: Chicago, Illinois Member No.: 1,359 |
Thanks a ton, agosine. However, this is what I'm getting:
CODE # ls
memory_cs.o prism2_cs.o serial_cs.o spectrum_cs.o orinoco_cs.o serial.o sharpzdc_cs.o # insmod serial.o # insmod serial_cs.o # /etc/rc.d/init.d/bluetooth restart Shutting down Bluetooth... Can't open RFCOMM control socket: Address family not supported by protocol Done. Starting Bluetooth... modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë Can't set line discipline: Invalid argument Can't initialize device: Invalid argument Can't open RFCOMM control socket: Address family not supported by protocol Done. # /etc/rc.d/init.d/pcmcia restart # /etc/rc.d/init.d/bluetooth restart Shutting down Bluetooth... Can't open RFCOMM control socket: Address family not supported by protocol Done. Starting Bluetooth... modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë Can't get port settings: Input/output error Can't initialize device: Input/output error Can't open RFCOMM control socket: Address family not supported by protocol Done. # lsmod Module Size Used by serial_cs 4640 0 (unused) serial 25504 0 [serial_cs] sa1100_bi 23088 0 (unused) net_fd 16688 1 usbdcore 22592 0 [sa1100_bi net_fd] usbdmonitor 5488 0 devinfo 2912 0 (unused) sharp_mmcsd_m 27408 2 # modprobe serial modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë # modprobe serial_cs modprobe: line 1: Invalid dependency 1à Bà0 â Áåõÿë # Obviously, I don't really know what I'm doing with these modules... Time for me to try to research this more. As always, I welcome any tips! |
|
|
|
Mar 31 2004, 07:10 PM
Post
#39
|
|
|
Group: Members Posts: 235 Joined: 5-January 04 From: Chicago, Illinois Member No.: 1,359 |
OK, so I just remembered to do a "depmod -a", and thus I no longer have the crazy modprobe results:
CODE # depmod -a
# modprobe serial_cs # modprobe serial # /etc/rc.d/init.d/bluetooth restart Shutting down Bluetooth... Can't open RFCOMM control socket: Address family not supported by protocol Done. Starting Bluetooth... Can't get port settings: Input/output error Can't initialize device: Input/output error Done. # hciconfig # Still ain't working... |
|
|
|
Apr 1 2004, 07:50 AM
Post
#40
|
|
|
Group: Members Posts: 123 Joined: 4-March 04 From: Toronto, Canada Member No.: 2,156 |
n8a - Glad I could help a little. If I come across anything more I'll let you know.
|
|
|
|
Apr 1 2004, 09:34 AM
Post
#41
|
|
|
Group: Members Posts: 153 Joined: 16-January 04 Member No.: 1,464 |
n8a,
Here's a new copy of the async kernel. What I did this time dump /proc/ksyms from both the standard tkcrom kernel, and the async kernel, then started tweeking the async kernel until the kernel symbols match between the two. What this has is CONFIG_SERIAL and CONFIG_SERIAL_COLLIE turned on, and CONFIG_SERIAL_8250 turned off. Also, the Aironet driver is turned on. At this point, the only apparent difference is the tkcrom kernel has two ksyms related to the Raylink driver, which I couldn't find the kernel option to turn on. I don't think that one is important. Let me know if you have better luck with this kernel. |
|
|
|
Apr 2 2004, 10:37 AM
Post
#42
|
|
|
Group: Members Posts: 235 Joined: 5-January 04 From: Chicago, Illinois Member No.: 1,359 |
That did it, Derek! Very, very nice.
Well, I must say that I'm finally very happy with the state of my 5500. Thanks MUCHO. I certainly want to do some dual-booting, but man, that zImage asyncio kernel makes a HUGE speed difference compared to running home-on-sd with the -f option and the normal tkcROM zImage. The one MAJOR remaining issue is that I have rebooted the 5500 three times, with the following results: First time, after installing a ton of apps, the system booted to this: I had a shell and it complained that the file system was corrupt. It then gave me these options: Enter password to repair or press Control-D. Then: 1) reboot... after which it dies and can't be started unless I flash it again 2) I get a prompt that comes up like the following (Repair FS) # I ended up having to pop the battery out to get it to respond… I then see the “Generating SSH Keys…” Nice! Second time, error messages for about 8 minutes before the startup message began… Third time, same as the first time (again, I had installed a bunch of apps). The good thing is that since *everything* seems to be on the SD card, once it loads up, everything is intact, including data that I’ve stored. If there is a way to remedy this unfortunate process, then this will become my hands-down favorite configuration for the 5500… |
|
|
|
Apr 2 2004, 11:20 AM
Post
#43
|
|
|
Group: Members Posts: 153 Joined: 16-January 04 Member No.: 1,464 |
Well, I'm glad that the new kernel is working for you. The filesystem errors you get on reboots worries me though. The only time I've come accross that is with the modified prep-home-sd.sh that is included on the tkc-2.0alpha3 rom. That version creates the /mnt/card/rc.card script with a mkswap command at the end that turns the ramdisk into a swapfile. This of course causes the ramdisk to become corrupted after heavy application usage (including installing a bunch of packages), which will cause fsck to dump out a bunch of errors.
I know that you've grabbed the newer prep-home-sd.sh script (test5) which doesn't have this issue, but just in case, double-check your rc.card file (on the root of the sd card), and if it contains a line "mkswap /dev/iramdisk" at the end of it, coment it out (along with the following swapon command). If it isn't the mkswap issue, then I'm out of ideas, as I haven't run accross this issue otherwise. I'll be at the user-group meeting on Sunday, so I can take a look then. --derek |
|
|
|
Apr 2 2004, 01:10 PM
Post
#44
|
|
|
Group: Members Posts: 235 Joined: 5-January 04 From: Chicago, Illinois Member No.: 1,359 |
hmmm... this is what I found in /home/rc.card:
CODE cd /
umount /home fsck -y /dev/mmcda1 # Rename the devices. This is so the SD shows up # as "Internal Storage" in Qtopia mv /dev/mtdblock1 /dev/iramdisk mv /dev/mmcda1 /dev/mmcsd1 ln -s /dev/mmcsd1 /dev/mtdblock1 mount -oasync,noatime /dev/mtdblock1 /home mkswap /dev/iramdisk swapon /dev/iramdisk I just checked the version of your script that I downloaded, and it definitely ends in test5.zip. In addition, I tried to replace the script on my CF card with the script from that zip file and both files were identical down to the modified date. So, I just comment out the last two lines and it should work fine, right? I'll give it a go. Out of curiosity, why did you turn the whole ramdisk into a swapfile, then specify a 20 MB swapfile when running the script? Or is the swapfile specification at install a new feature of the test5 version? At any rate, thanks again and I'll let you know if this solves it -- Nate |
|
|
|
Apr 2 2004, 01:37 PM
Post
#45
|
|
|
Group: Members Posts: 153 Joined: 16-January 04 Member No.: 1,464 |
Let me back up a couple notches here :-)
I had originally put together the prep-home-sd script (and modified rc.sysinit script) without any swap file, then sent it into Proto to include in the tkcrom. Then we started going over ideas to improve it a bit. Proto thought it would be a good idea to then turn the ramdisk into a swap partition, since it was no longer being used for /home, so he added the mkswap & swapon to the end of the script and included that version in the tkcrom 2.0 alpha-3 release. However, since a fsck is run by rc.sysinit before the rc.card script is called, this would cause the fsck errors that you see. Meanwhile, I had fallen off the face of the plannet (read: got bogged down with work). Later, I sat down a my orignial script (without the mkswap) and overhauled it (added a swap file (not a swap partition), moved some stuff around, etc). I've been waiting to here back from Proto, but I take it that it's his turn to get over-loaded with non-Zaurus activities :-). The version of the rc.card script you posted above, is not generated by the new script but is the one generated by the (old) prep-home-sd.sh script included in the tkcrom. So, one of two possibilities: 1) you probably ran the old version of the script (the one on the tkcrom) by mistake, or 2) you ran the old version originally, and for some reason the new version didn't overwrite the rc.card that was put on the sd previously. To make sure that you are running the right version, specify the whole path of where you put it. Also, the new version has better help (the help text should pop up if you don't specify any parameters). Or, a third posibility is that I upload the wrong script. Nah, that couldn't happen :-) Good luck, --derek |
|
|
|
![]() ![]() |
|
Lo-Fi Version | Time is now: 21st May 2013 - 06:06 PM |