Embedded Linux Project
Contents
1. ODevice unknown hardware using default OGlobal creating global configuration instance OConfig 0Config lt unknown gt ODevice reports transformation to be 0 lt unknown gt setting QWS_DISPLAY to Transformed Rot0 0 qt_init starting in daemon mode Please press Enter to activate this console BusyBox v1 4 1 2007 03 25 03 24 25 CST Built in shell ash Enter help for a list of built in commands rootQ none uname m uname m armv4tl While BusyBox is full featured it was difficult to use vi or to do tab completion or to edit shell commands in place because all control and navigation characters appeared on the screen when they are typed Perhaps these issues have been fixed in later versions I thought it might have been my terminal emulator so I tried various emulators and settings and found PuTTY to work the best but control characters still appeared root none L A B E E rcS 17L 343007 111111111111111111111111111111111111111111111111111 11111111111111111a I rcS modified 19 19 100 udhcpc i ethO b p var run udhcpc eth0 gt dev null 2 amp 1 wq kk rcS 19 19 100 rcS 18 19 94 rcS 17 19 89 rcS 18 19 94 kk rc8 18 19 94 rcS 17 19 89 rcS 16 19 84 rcS 17 19 89 x wq Kevin Wu 6 Embedded Linux Project Cross development Toolchain 5 Cross development Toolchain I learned the hard way that building a cross developmen2. fsck 1 37 21 Mar 2005 done Setting up networking done Starting hotplug subsystem usb isapnp ide scsi Setting MAC address to 00 11 22 33 44 55 Kevin Wu 14 Embedded Linux Project Mounting Root Filesystem over NFS SIOCSIFHWADDR Device or resource busy Setting up IP spoofing protection rp_filter Enabling packet forwarding done Configuring network interfaces done Starting portmap daemon portmap Cannot access the Hardware Clock via any known method Use the debug option to see the details of our search for an access method Running ntpdate to synchronize clock Recovering nvi editor sessions done INIT Entering runlevel 2 Starting internet superserver inetd Starting OpenBSD Secure Shell server sshd Starting NFS common utilities statd Starting NFS servers nfsd mountd Starting periodic command scheduler cron Starting web server apache ssl Debian GNU Linux 3 1 GESBC 93xx console GESBC 93xx login root Last login Fri Sep 1 08 00 52 2006 on ttyAMO Linux GESBC 93xx 31 Thu Mar 23 13 01 48 MST 2006 armv4l GNU Linux The programs included with the Debian GNU Linux system are free software the exact distribution terms for each program are described in the individual files in usr share doc copyright Debian GNU Linux comes with ABSOLUTELY NO WARRANTY to the extent permitted by applicable law root GESBC 93xx root ls adc_test adc_test_new devmem2_test word log adc_test
3. wuziq wuziq laptop linux scp hello ko root 192 168 0 104 root hello ko 100 1965 1 9KB s 00 00 I was able to do this with my NFS mounted root filesystem Tab completion also worked for complet ing paths on the target which I thought was pretty cool Kevin Wu 18 Embedded Linux Project Conclusions Future Work 12 Conclusions Future Work Coming soon Kevin Wu 19
4. but decided to try something that I found while googling compilation errors called crosstool It is a script that basically does all the work of trying different version combinations In fact there is a matrix created from the author s own buildlogs created by the script that shows which combinations work and don t work i looked for one that had the glibc installed on the board 2 3 2 and a gcc core dir that matched one i could just apt get from the ubuntu repos 3 3 6 i edited the corresponding combination s dat file and ex plicity included the gcc core dir variable explained here crosstool vars html and i In s f usr bin gcc 3 3 usr bin gcc just in case so here goes before i end up using glomation s toolchain ok turns out some automatically generated files weren t being generated properly crbuild glibc csu version info h missing terminating character did some googling googled for crosstool build glibc csu version info h missing terminating character and it s because of the freakin shell i thought i was using bash but it turns out ubuntu uses dash so i did the crosstool final gcc built ok Cross toolchain build complete Result in opt crosstool gcc 3 3 6 glibc 2 3 2 arm unknown linux gnu testhello C compiler can in fact build a trivial program Done I successfully compiled the obligatory hello world program and ran it on my board as well as the file utility rootC none gesbc 2440_binaries fi
5. image is in SDRAM I transfer it to the NAND flash nand write jffs2 33000000 0 filesize write jffs2 is used because the NAND is formatted with JFFS2 filesystem It reads writes filesize bytes starting at offset 0 to from memory address 33000000 The filesize variable holds the size in hex of the image read from the last tftp command Now I have to tell U Boot where the kernel image is set bootcmd nand read jffs2 33000000 0 filesize bootm 33000000 This simply tells U Boot that the bootable image is located at address 33000000 offset 0 and image size filesize Loading the kernel was successful My target booted and I saw that it was indeed the kernel I had compiled NAND read device 0 offset 0 size 1307664 1307664 bytes read OK Booting image at 33000000 Image Name Linux 2 6 18 2 Created 2008 06 18 3 11 52 UTC Image Type ARM Linux Kernel Image uncompressed Data Size 1307600 Bytes 1 2 MB Load Address 30008000 Entry Point 30008000 Verifying Checksum OK OK Starting kernel Dias doe gether as done booting the kernel Linux version 2 6 18 2 wuziq wuziq laptop linux gcc version 3 4 5 2 Tue Jun 17 20 11 48 PDT 2008 CPU ARM920T 41129200 revision O ARMv4T cr c0007177 Machine SBZ2440 Memory policy ECC disabled Data cache writeback CPU S3C2440A id 0x32440001 S3C244X core 399 651 MHz memory 133 217 MHz peripheral 66 608 MHz S3C24XX Clocks c 2004
6. is intended for anybody Depending on who you are though you might find this content to be trivial I can only hope that the circumstances that I describe in section 2 Background give you some perspective into where this is all coming from 1 2 Status Log Quotes Throughout this document I will often copy paste bits and pieces from a status log that I kept while I experimented and made progress here is an example of what some text from my status log will look like in this document i don t really bother with syntax or grammar and it appears like i am just chatting with someone online These quotes show my thought process as I worked With things happening so fast I felt it would behoove me to keep a log that I could reference later if I needed to remember something For each part that I describe in this document I will try to include a piece from my log that gives more color to the situation instead of paraphrasing everything 1 3 Code and Shell Snippets Code commands and standard output all look like this Kevin Wu 1 Embedded Linux Project Background 2 Background In the spring of 2008 I took a class called Embedded Linux It was a 10 week course at UCSD Extension and my first experience with embedded Linux We met once a week for about 2 hours Including myself there were 10 people enrolled Our textbook was Building Embedded Linux Systems by Karim Yaghmour The teacher attempted to cover one chapter pe
7. of LDD and found Linux Device Drivers 3rd Edition at lwn net I also downloaded the driver exam ples from the author s FTP site I tried executing make but it was looking for kernel headers whose version matched my kernel version 2 6 18 2 I had installed the kernel source tree on my target but my target also had slightly newer kernel headers loaded onto it as well along with s3c2410 appended to the source tree name After some sloppy renaming the make worked but upon insmod ing I got Kevin Wu 16 Embedded Linux Project Hello World Kernel Module root GESBC 93xx misc modules insmod hello ko hello version magic 2 6 18 6 s3c2410 mod unload ARMv4 gcc 4 1 should be 2 6 18 2 mod unload ARMv4 gcc 3 4 insmod error inserting hello ko 1 Invalid module format It turns out that make is using some information it finds in the kernel source tree to figure out version stuff This is actually explained in LDD3 I found out that you can point make to your kernel source tree by setting the KERNELDIR variable After setting KERNELDIR and creating arm linux gcc and arm linux ld symbolic links to my crosstool cre ated toolchain binaries the make ran successfully and modinfo showed matching magic versions wuziq wuziq laptop linux modinfo hello ko filename hello ko license Dual BSD GPL depends vermagic 2 6 18 2 mod unload ARMv4 gcc 3 4 Finally I successfully loaded and unloaded the driver
8. root GESBC 93xx misc modules lsmod Module Size Used by root GESBC 93xx misc modules insmod hello ko Hello world root GESBC 93xx misc modules lsmod Module Size Used by hello 1120 O root GESBC 93xx misc modules rmmod hello Goodbye cruel world root GESBC 93xx misc modules lsmod Module Size Used by root GESBC 93xx misc modules Kevin Wu 17 Embedded Linux Project Miscellaneous 11 Miscellaneous Here are some little tips that helped me along the way e To show the U Boot environment variables use printenv SBC2440A printenv bootdelay 1 baudrate 115200 ethaddr 08 00 3e 26 0a 5b filesize 1f44000 fileaddr 30000000 gatewayip 192 168 0 1 netmask 255 255 255 0 serverip 192 168 0 102 ipaddr ip 192 168 0 8 192 168 0 102 192 168 0 102 255 255 255 0 GESBC eth0 off bootcmd nand read jffs2 33000000 O 13f410 bootm 33000000 bootargs nfsroot 192 168 0 102 home wuziq gesbc 2440 rootfs_usb rw console ttySACO 115200 ip dhcp stdin serial stdout serial stderr serial Environment size 458 65532 bytes e To configure ethernet manually if there s no DHCP client ifconfig ethO 192 168 2 103 route add default gw 192 168 2 1 ethO ifconfig lo 127 0 0 1 echo nameserver 192 168 2 1 gt etc resolv conf To see all possible options and variables that are configurable for a configure script simply use configure help e To quickly transfer files from host to target in an ethernet connected setup use scp
9. see if either works 10 59 59 AM coworker if that doesn t work modify an existing drive to do it 10 59 59 AM me i tried the first line i get segmentation fault 11 00 20 AM coworker what about second line 11 00 26 AM me devmem2 does the second line 11 00 47 AM coworker u dont need to write a whole driver 11 00 58 AM coworker just insert some code into an existing driver 11 01 01 AM coworker to do it 11 01 07 AM coworker to see if it works 11 01 07 AM me HAMMMMMMMMMMMMMM 11 01 14 AM me very interesting 11 01 23 AM coworker thats what i would do cuz it s easy 11 01 26 AM me a driver that the kernel is already using 11 01 30 AM coworker ya 11 01 36 AM me very interesting This new information set me on a path that never crossed my mind for this class writing a Linux device driver There was a book at my workplace called Linux Device Drivers 2nd Edition LDD2 I flipped through the pages not understanding anything except for the Hello World program at the beginning I figured I would at least try writing and loading it and then go on from there This decision led me to a few hours of trying to debug compilation errors I couldn t even get the module to compile on my host I finally realized that LDD2 was aimed at kernel version 2 4 and from what I had read while debug ging compilation errors kernel version 2 6 handles modules in a very different way I checked for a newer version
10. stack chk fail basically i needed to install ncurses dev craziness so anyway now i m in the menuconfig i m unsure about almost everything currently stuck on con fig embedded I configured the kernel based on the help blurbs for each option but I soon noticed that my kernel didn t have options related to the S3C2440 or S3C2410 I was confused as to how each architecture gets added to the kernel tried to finish configuring the kernel even though i think it won t work because i don t have the config options for the s3c2440 tried doing some google searches for s3c2440 linux kernel version stuff like that at this point i m pretty tired not sure how it all comes together how does the specific information for certain chips get added to kernel code if that specific info isn t there does linux not work on that chip or does it just run a very generic version of linux that is specific to the generic architecture i mean aren t there different versions of x86 i guess different x86 chips have different features but is the kernel code mature enough such that all those different chips are included i don t feel like compiling the toolchain now maybe i will just use the glomation ones i m starting to understand why don recommends linux already on the board it would definitely be easier that would mean i just need to configure anjuta to use the glomation toolchain instead of the toolchain already on my ubuntu and i would just need to read
11. the most important of which was the ethernet interface Even though I wasn t sure what interesting thing I would do with the board I wanted the board to allow me enough options Kevin Wu 2 Embedded Linux Project SBC Project Ideas 3 SBC Project Ideas Since the project was left open ended it was up to me to decide what to do with my SBC It was difficult to come up with a concrete project idea because I did not fully know what my board could do much less how to work with an embedded Linux system Some people wanted to roll their own Linux distribution which sounded interesting but maybe too challenging given the time constraints of the class At the same time even though I bought an SBC with Linux pre installed and cross development toolchain already built I didn t want everything to be simply handed to me as such I wanted to know what building an embedded Linux system entailed to do at least a part of it and then to develop an application for it With this vague goal in mind I began to think about some possible programs I could develop that would eventually run on my SBC I wanted my SBC to do something practical to do something that I would actually use I soon discovered that learning about and working with my SBC and embedded Linux systems would severely limit the scope of any application I hoped to develop Kevin Wu 3 Embedded Linux Project Glomation GESBC 2440 4 Glomation GESBC 2440 4 1 Bootlo
12. the s3c2440 spec sheet to do some programming to manipulate the pins but then i m not really learning as much and i ve already sort of gotten myself this deep into it i just don t understand how to get the right kernel how to find out whether the s3c2440 is more or less supported in specific versions of the kernel that is the big mystery now is it more supported in debian i looked at the debian etch glomation package and it s just the filesystem layout no kernel source code or whatever I eventually figured out that I didn t need to be concerned about it i went back to the main page of arm linux co uk went to developers then i clicked on kernel compilation kerncomp php very useful info ESPECIALLY THIS note kernels later than 2 6 0 test2 do not require a rmk or vrs patch to be applied since ARM architecture support is already merged AMAZING i think that s where my confusion was i thought from the start that i needed an arm kernel patch so i looked for the latest one at kernel patches v2 6 the latest one was for a 2 6 0 kernel so i downloaded that kernel it turns out arm support was merged into kernels after that so i can just download the latest one and it should have arm as well as my s3c2440 which i have just confirmed it does sweeeeeettt Kevin Wu 8 Embedded Linux Project Linux Kernel Ultimately I used the kernel package provided by Glomation I learned that even though I had architecture support my p
13. to begin or what this endeavor would entail I began by purchasing some PV cells from Edmund Scientific and reading up on the ADC in the S3C2440 datasheet I wanted to program the ADC in the same way I programmed it for the AtTiny44 by simply writing values to a register name and checking the results stored in another register I looked up the register names in the kernel header files and found that ADCCON is the configuration register while the first 10 bits of ADCDATO would hold conversion results I wrote a program to simply read from the ADCCON register but got segmentation fault I learned that segmentation fault meant I was trying to access a memory address that I wasn t allowed to access I contacted Glomation about whether they had any sample code for talking to the ADC This is where things became complicated Dear Kevin Wu We currently do not have any sample code to access the ADC on the S3C2440 chip There are 2 ways to access the on chip ADC of the S3C2440 a Linux device driver or a simple user space program The device driver approach requires some in depth knowledge of the Linux device drivers and is a little more complicated than the user space program but it provides system level access to the on chip ADC The user space program is simple to program and provide simple access of the on chip ADC Here is a link to a user space program that can access any memory location of the Linux system devmem2 c Since the on chip
14. ADC is memory mapped you can actually access the ADC just using the devmem2 program in a sequence of writes and reads of the ADC registers Please refer to the S3C2440 data sheet on how to program the ADC Please be noted that the default installed kernel include a touch screen driver for the GESBC 2440 SBC which uses the on chip ADC as touch screen interface You will need to disable the touch screen driver from the kernel in order to use the on chip ADC Regards Glomation Customer Support I compiled and ran devmem2 on my target I read the value at the ADCCON address 0x5800000 to check whether it matched the reset value given in the datasheet but it didn t So I recompiled the kernel without touchscreen support and created a ulmage which is the kernel image format that U Boot accepts Kevin Wu 10 Embedded Linux Project Loading Kernel Onto GESBC 2440 8 Loading Kernel Onto GESBC 2440 Loading the kernel is all done in U Boot my target s bootloader In order to load the kernel image onto the NAND flash on my target I had to erase the NAND by issu ing the command nand erase clean Then I load the ulmage onto the SDRAM chip via a TFTP server running on my host I used this one with the ulmage file located in the TFTP server s shared folder set ipaddr 192 168 xxx xxx set serverip 192 168 xxx xxx tftp 33000000 uImage This sets the target and server host IP addresses and begins the TFTP transfer of uImage Once the
15. Embedded Linux Project FIRST DRAFT Kevin Wu June 24 2008 Contents AUN About The DOCUMEN lt cs hae eA ws Aw hee e A hee Rs 1 LA AWE oc a a BANG AE a Enina ew EOE ae 1 lu SWS TOUS s a oe Ponce a a ee ee E A 1 1 3 Code and Shell Snippets osos or ee ee ee ee x 1 Background 2 oak xe SR RR Re ERR ee RD RSS ce A 2 SBC PROSCUINGSS 2 as e eb a sex omo we da io oe ee 3 Glomatinn GESBC 2440 oo sada na da ER A OE a a xx 4 4 1 Bootloader uox pp XY xq 9 ERE eue Se SO Re SOA Se N 4 42 Pre installed Kernel eel 5 4 3 rui UC a EE Sh ae RO we Ree ee a A ee 5 Cross development Toolchain 0 rrr 7 5l o RO rca Rh Dew Na AA we a wea A 7 A O A NN 8 o ake a A ea ee a ee we es we we 10 Loading Kernel Onto GESBC 2440 lt a cascera sasaaa raaa ee 11 Mounting Root Filesystem over NFS leen 12 9 1 U Boot bootargs for Mounting Root Filesystem Over NFS 14 9 2 Stab ane HRAD cuc s soe 2 2h 4G ORS REO ESE aa Oe x ox Re 14 Hello World Kernel Module naaa a 16 scel NES i i d m ia eee E ae a ARE Eee Cer ace 18 Conclusions Future WORK lt ia sossa omo NS EE eS AE eee oS a 19 Embedded Linux Project About This Document 1 About This Document I am writing this document not as a part of an assignment or official project but for record keeping pur poses As such you may find the tone to be somewhat colloquial and the formatting to be unconventional 1 1 Audience This document
16. Simtec Electronics CLOCK Slow mode 2 116 MHz fast MPLL on UPLL on Unfortunately I didn t realize that my root filesystem was also located on the NAND flash so I had acciden tally erased it when I issued the nand erase clean command Kevin Wu 11 Embedded Linux Project Mounting Root Filesystem over NFS 9 Mounting Root Filesystem over NFS With time running out I needed a new root filesystem as soon as possible Although I read the chapter on how to set up the root filesystem I hadn t actually done the work I looked at the Glomation support site and saw that they provided a Debian based file system for use on a USB drive I tried downloading it but discovered it was over 200 MB so it wouldn t fit on my target by any means I attempted to unpack it onto my USB stick but there were permissions issues which I think were resulting from the fact that my USB stick was formatted as fat16 instead of ext2 Then I noticed instructions for mounting the root file system over NFS They were for a Glomation board different from mine but I used what I could and figured out the rest i m noticing on USB as root file system txt that it s just a matter of specifying where the rootfs in the bootloader i think i could just follow the user manual instructions for set up uboot boot environment and change the bootargs to root dev sdal rootfsfolder rw rootfstype vfat noinitrd init linuxrc con sole ttySAC0 115200 i think i ll t
17. ader The GESBC 2440 uses U Boot as its bootloader U Boot 1 1 2 Nov 15 2006 01 41 13 U Boot code 33C00000 gt 33C25CC8 BSS RAM Configuration Bank 0 30000000 64 MB Get flash bank O size 0x0 Total Flash bank s sizes 0x200000 protect monitor 25cc8 bytes address 0 Flash 2 MB M AN a 13 9 TI Has VGL r3 TRE MERI DHLELEEEEETETEETETHETTEETIU A Crap 0004044044 iii T E DE f HanRun HR911105A 33C5F610 Kevin Wu Embedded Linux Project Glomation GESBC 2440 NAND Flash chip found Manufacturer ID OxEC Chip ID 0x76 Samsung unknown 64Mb 1 flash chips found Total nand_chip size 64 MB 64 MB In serial Out serial Err serial Found DM9000 ID 90000a46 at address 18000000 ethaddr 08 00 3e 26 0a 5b DM9000 work in 16 bus width Not link of ethernet Hit any key to stop autoboot 1 0 SBC2440A 4 2 Pre installed Kernel A patched version of Linux 2 6 18 2 came pre installed on the GESBC 2440 NAND read device O offset 0 size 1293736 1293736 bytes read OK Booting image at 33000000 Image Name Linux 2 6 18 2 Created 2007 10 28 3 50 20 UTC Image Type ARM Linux Kernel Image uncompressed Data Size 1293672 Bytes 1 2 MB Load Address 30008000 Entry Point 30008000 Verifying Checksum OK OK Starting kernel arb dea ead ageism coseron done booting the kernel Linux version 2 6 18 2 root GESBC gcc version 4 1 2 20061115 p
18. c adc test new c halfword log word part2 log adc test o adc test new o memory log txt Kevin Wu 15 Embedded Linux Project Hello World Kernel Module 10 Hello World Kernel Module Gauging from the progress of my classmates from their weekly updates I felt that mounting a root filesys tem over NFS was presentable material for my final project Nevertheless I had one more day until the last class where we would be presenting our projects and I still wanted to do what I had originally intended with the ADC I tried devmem2 again but I was getting the same results as before Were the kernel reconfiguration and recompilation for nothing I tried writing a value to the ADCCON register address then reading the value back but the value remained the same as before the write root 192 root devmem2 0x5800000 h 0x5901 dev mem opened Memory mapped at address 0x40017000 Value at address 0x5800000 0x40017000 0x12 Written 0x5901 readback 0x12 I felt that I had done all I could do But then I talked to a couple coworkers about accessing memory locations They reminded me that the OS would prevent me from directly accessing memory and that a device driver would probably be needed 10 58 34 AM coworker so 10 58 39 AM coworker write a little c program 10 59 12 AM coworker that sets uint16 ptr 0x5800000 and writes to it 10 59 25 AM coworker and then trys 0x40017000 to write to it 10 59 28 AM coworker
19. e configures the console and ip dhcp tells U Boot to use DHCP to obtain an IP address Saving these U Boot environment variables is done with saveenv 9 2 fstab and inittab The fstab file that came with the root filesystem included a line to mount dev sda1 as the root filesystem Since the root filesystem is already mounted over NFS I simply removed that line This fixed the no entry for the root filesystem listed in etc fstab error and allowed inittab to execute during bootup inittab however was not configured properly so after executing the scripts in etc rc2 d I got INIT no more processes in this runlevel I figured out that it was because the terminal configuration at the end of the file was for a different board with different terminal settings TO 23 respawn sbin getty L ttyAMO 57600 vt100 T1 23 respawn sbin getty L ttyAM1 57600 vt100 I replaced those two lines with my terminal settings TO 23 respawn sbin getty 115200 console and it works w0000000ttt VFS Mounted root nfs filesystem Freeing init memory 112K INIT version 2 86 booting devfsd No devfs on dev not starting Cannot access the Hardware Clock via any known method Use the debug option to see the details of our search for an access method Cannot access the Hardware Clock via any known method Use the debug option to see the details of our search for an access method Cleaning up ifupdown done Checking all file systems
20. eripherals wouldn t necessarily work and that device drivers would need to be written for them I thought maybe I could do a diff between the Glomation provided kernel and the kernel of the same version from kernel org and determine what device driver patches were created i ve been looking at the files that are different and i have no idea what is going on it s all c but i don t understand it eh i wonder whether i can merge the differences into the latest kernel but i think i want to understand why there are differences and how whoever made the changes knew what changes to make how can i upgrade from glomation to the latest would it be possible to patch the glomation kernel that might not work because patching depends on certain lines being in there if stuff was changed around then the patch might not get implemented the long hard way would be to figure out WHY things were patched go through each changed file and figure out why it was changed then try to implement the same changes manually in the latest kernel Kevin Wu 9 Embedded Linux Project S3C2440 ADC Project 7 S3C2440 ADC Project After getting a better grasp of my board the toolchain and the kernel I decided that my project would be to use the ADC to report voltage from a solar cell photovoltaic cell Even though I felt that it would be a fairly simple project we had programmed the ADC on our Atmel AtTiny44 boards in my ECP3 class I didn t know where
21. le m magic helloworld file m magic helloworld helloworld ELF 32 bit LSB executable ARM version 1 for GNU Linux 2 4 18 dynamically linked uses shared libs not stripped I later had to recreate a toolchain using crosstool because compiling my kernel required gcc v3 4 3 or higher Kevin Wu 7 Embedded Linux Project Linux Kernel 6 Linux Kernel For the GESBC 2440 Glomation provides kernel version 2 6 18 2 Still I wanted my own kernel i tried getting the latest stable kernel from kernel org then getting the latest patch from arm linux org uk i patched the kernel source fine but when i did make menuconfig i got an error from mconf c did some search and it turns out it s because my gcc the ones from the fiesty repos for my c c class is too new v4 1 or whatever need to use gcc v3 4 or something instead found a way to get around it though use this patch Index package config mconf c package config mconf c revision 10888 package config mconf c working copy 101 7 101 7 static int indent 0 static struct termios ios_org static int rows cols static struct menu current_menu struct menu current menu static int child_count static int single_menu_mode so that worked but then i got a whole different slew of errors similar to the ones i got when i tried to make menuconfig with the glomation kernel source i picked a line and did a google search found this
22. ocal Loopback inet addr 127 0 0 1 Mask 255 0 0 0 UP LOOPBACK RUNNING MTU 16436 Metric 1 RX packets 0 errors 0 dropped 0 overruns 0 frame 0 TX packets 0 errors 0 dropped 0 overruns 0 carrier 0 collisions 0 txqueuelen 0 RX bytes 0 0 0 b TX bytes 0 0 0 b The first thing I noticed is that I no longer had the vi control character or tab completion problems that I had with BusyBox see subsection 4 3 I was essentially using a standard Linux shell Also this root filesystem already came loaded with many more applications including apt get root 192 etc apt get update Get 1 http http us debian org stable main Packages 5449kB Kevin Wu 13 Embedded Linux Project Mounting Root Filesystem over NFS Get 2 http http us debian org stable main Release 94B Get 3 http http us debian org stable contrib Packages 53 6kB Get 4 http http us debian org stable contrib Release 97B Get 5 http http us debian org stable non free Packages 71 0kB Get 6 http http us debian org stable non free Release 98B Fetched 5574kB in 49s 114kB s Reading Package Lists Done 9 1 U Boot bootargs for Mounting Root Filesystem Over NFS Here is how I set the boot arguments in U Boot set bootargs nfsroot 192 168 xxx xxx home wuziq gesbc 2440 rootfs_usb rw console ttySAC0 115200 ip dhcp nfsroot tells U Boot to look at address 192 168 xxx xxx and directory home wuziq gesbc 2440 rootfs usb for the root filesystem consol
23. r class using Powerpoint presentations to summarize each chapter Prior to this class I d had casual experience with Linux from trying out various distributions on my desktop computer I ve settled with using Ubuntu for both my desktop and laptop along with Windows so I used it as my host system for the class Ubuntu supposedly makes things easy I say supposedly because I can t claim that it s something I can fully appreciate since I grew up using Windows But from what I have read and from my tinkerings with Linux it seems that making Linux easy and accessible is an impossibly enormous undertaking After taking this class 1 am even more impressed Since embedded Linux was a new subject for me I spent a lot of time reading the textbook I started from the very beginning and by the end of the class I had covered through chapter 6 and had also skimmed chapter 7 In this class we had the option of either doing weekly quizzes or to do something interesting of our choosing with a single board computer SBC that could run Linux I chose the second option because I felt it would be more enlightening than simply reading the textbook and doing written quizzes For those of us who chose this path the teacher recommended that we purchase an SBC that already had a Linux distribution installed and directed us to Gumstix and Glomation I bought the Glomation GESBC 2440 because it seemed relatively new and had plenty of built in features
24. rerelease Debian 4 1 1 21 586 Sun Oct 28 03 48 37 UTC 2007 CPU ARM920T 41129200 revision 0 ARMv4T cr c0007177 Machine SBZ2440 Memory policy ECC disabled Data cache writeback CPU S3C2440A id 0x32440001 S3C244X core 399 651 MHz memory 133 217 MHz peripheral 66 608 MHz S3C24XX Clocks c 2004 Simtec Electronics CLOCK Slow mode 2 116 MHz fast MPLL on UPLL on 4 3 init and BusyBox The GESBC 2440 uses BusyBox as its login and shell After the kernel is loaded and the root filesystem is mounted the system enters runlevel 5 and runs one script etc rcS which mounts filesystems starts Opie and starts BusyBox VFS Mounted root jffs2 filesystem Freeing init memory 112K Vendor Chipsbnk Model Flash Disk Rev 5 00 Type Direct Access ANSI SCSI revision 02 SCSI device sda 2061824 512 byte hdwr sectors 1056 MB sda Write Protect is off sda assuming drive cache write through s3c2410 sdi s3c2410 sdi powered down Kevin Wu 5 Embedded Linux Project Glomation GESBC 2440 SCSI device sda 2061824 512 byte hdwr sectors 1056 MB sda Write Protect is off sda assuming drive cache write through sda sdal sd 0 0 0 0 Attached scsi removable disk sda mount etc as ramfs mount dev as tmpfs mounting all Scanning hardware You seem to already have a home root Applications directory Assuming it is the Opie Applications directory Exiting Starting Opie ODevice found Hardware SBZ2440
25. ry it the rootfs package just finished downloading and luckily my thumbstick is 1gb ok there are some permissions issues in the rootfs package i could sudo untar it to my host but not to the thumbstick hmmmm ok what about nfs looking at Ubuntu Gutsy NFS_ Server and nfs rootfs setup linux txt seeeeeems doable but then so did usb sudo apt get install nfs kernel server nfs common portmap sudo dpkg reconfigure portmap sudo etc init d portmap restart sudo etc init d nfs kernel server restart sudo export s a wuziq wuziq laptop linux gesbc 2440 rootfs_nfs rpcinfo p program vers proto port 100000 2 tcp 111 portmapper 100024 1 udp 36513 status 100024 1 tcp 38860 status 100000 2 udp 111 portmapper 100003 2 udp 2049 nfs 100003 3 udp 2049 nfs 100003 4 udp 2049 nfs 100021 1 udp 37927 nlockmgr 100021 3 udp 37927 nlockmgr 100021 4 udp 37927 nlockmgr 100003 2 tcp 2049 nfs 100003 3 tcp 2049 nfs 100003 4 tcp 2049 nfs 100021 1 tcp 60456 nlockmgr 100021 3 tcp 60456 nlockmgr 100021 4 tcp 60456 nlockmgr 100005 1 udp 32974 mountd 100005 1 tcp 46968 mountd 100005 2 udp 32974 mountd 100005 2 tcp 46968 mountd 100005 3 udp 32974 mountd 100005 3 tcp 46968 mountd seems to look ok ok now to try booting dude i think it worked s3c2410 sdi s3c2410 sdi powered down Sending DHCP requests OK IP Config Got DHCP answer from 192 168 2 1 my address is 192 168 2 109 Kevin Wu 12 Embedded Linux Project Mounting Root Filesy
26. stem over NFS IP Config Complete device eth0 addr 192 168 2 109 mask 255 255 255 0 gw 192 168 2 1 host 192 168 2 109 domain nis domain none bootserver 192 168 2 1 rootserver 192 168 2 101 rootpath Looking up port of RPC 100003 2 on 192 168 2 101 Looking up port of RPC 100005 1 on 192 168 2 101 VFS Mounted root nfs filesystem Freeing init memory 112K SBZ try run sbin init INIT version 2 86 booting devfsd No devfs on dev not starting The device node dev sdal for the root filesystem is missing incorrect or there is no entry for the root filesystem listed in etc fstab The system is also unable to create a temporary node in dev shm to use as a work around This means you have to fix this manually CONTROL D will exit from this shell and REBOOT the system Press enter for maintenance or type Control D to continue root 192 cd root 192 1s bin dev home lib mnt opt root sys usr boot etc initrd lost found nfsroot proc sbin tmp var root 192 uname Linux root 192 ifconfig etho Link encap Ethernet HWaddr 80 E2 66 60 00 01 inet addr 192 168 2 109 Bcast 192 168 2 255 Mask 255 255 255 0 UP BROADCAST RUNNING MULTICAST MTU 1500 Metric 1 RX packets 2878 errors 0 dropped 0 overruns 0 frame 0 TX packets 1538 errors 0 dropped 0 overruns 0 carrier 0 collisions 0 txqueuelen 1000 RX bytes 0 0 0 b TX bytes 0 0 0 b Interrupt 51 Base address 0xe000 lo Link encap L
27. t toolchain depends highly on the version of gcc that is used to do the build gnuarm com groups packages for binutils gcc and newlib according to the version of gcc installed on your host but before I realized this I tried building my own versions and ran into countless compilation errors Unfortunately even with the nicely grouped packages from gnuarm com I could not get past building the bootstrap gcc From my project log ok binutils build a success gcc 4 2 is ok so far now for building bootstrap gcc configure went without a hitch huh i even used without headers not sure if that is because it s been fixed or whether i m just building the correct version for the version of gcc i have installed uh oh gcec 4 1 1 gcec gthr posix h 43 21 error pthread h No such file or directory gcc 4 1 1 gcc gthr posix h 44 20 error unistd h No such file or directory so far everyone seems to be saying that i need a c library before i try to compile the bootstrap compiler so why doesn t the book say that ARGH trying gcc 4 1 to build bootstrap gcc same problem tired moving onto trying to build glomation kernel using glomation toolchain i think i still need to compile the toolchain though argh It seemed like picking the correct versions of the different components would take a lot of guess and check work which I didn t have time to do 5 1 crosstool I almost gave up and used the toolchain provided by Glomation
Download Pdf Manuals
Related Search