I'm working on an embedded Linux system that has a specific I2C platform driver and I'm writing a custom I2C driver. Everything works fine, but I have a problem with their dependencies.
As my custom driver uses the default I2C functions, once I compile it, the make command automatically updates the modules.dep file saying that my driver depends on i2c-core to run, but that is not enough. In order to i2c-core to be configured I need to load i2c-omap first (the platform's driver) and only then my driver works properly.
Unfortunately, I can't find any dummy function to call and thus trick the make into adding another dependency when it generates my driver. Also, I would prefer an automated solution instead of modifying modules.dep with something like sed -i 's/RE1/RE2/' modules.dep.
So, is there any way to explicitly add a dependency to a module when I compile it?
Thanks!
I found an answer here: http://www.xml.com/ldd/chapter/book/ch11.html
I solved my problem calling
request_module("i2c-omap");
Anyway, this does not exactly update the dependencies file as I first intended. If anyone knows a way to do that, please add a comment here!
Related
I'm building Kernel for my embedded ARM-based system (system is a ARM7 VARISCITE DART6UL).
Since I need to build Kernel appending my local version to it, I read it's possible to edit (using make menuconfig) the right local version I need, writing in General setup->"Local version - append to kernel release".
Commands I execute to edit .config kernel are the following:
make mrproper
make imx6ul-var-dart_defconfig
make menuconfig
My question is: why my kernel release that I appended in the graphical kernel menu config file doesn't still remain saved?
Everytime I enter in menuconfig it disappear: Is there a way to fixed it, avoiding the need to rewrite each time?
Regards
Paolo
You need to copy back the .config file to arch/arm/configs/imx6ul-var-dart_defconfig
I have a custom board running Yocto (Jethro) and would like to run a single u-boot command, preboot. Obviously, breaking the boot sequence with space and running it manually works. How do I get it to run automatically? More specifically, where is the startup command sequence, by default?
Edit: Also, I am aware I can edit the environment at runtime. However, I am trying to build this change into the image so I can distribute it.
When you are in the uboot environment. Enter printenv, it will list the environment variables that uboot uses.
There is a variable name bootcmd. Currently, mine contain a bunch of if else command. Similarly, add your prefer function there for boot.
And after it is finished and tested. Use saveenv to store the edit
Here is a syntax for uboot.
Edit:
U-Boot allows to store commands or command sequences in a plain text file. Using the mkimage tool you can then convert this file into a script image which can be executed using U-Boot's autoscr command. U-boot Scripting Capabilities
Typically, your U-Boot recipe will build U-Boot for a single machine, in that case, I'd normally just patch the compiled in, default, U-Boot environment to do the right thing. This is achieved by
SRC_URI_machine += "file://mydefenv.patch"
Or (even better) use your own git tree. This would also have the additional benefit that your system might be able to boot up and to something useful, even if the environment would be totally corrupted.
Another possibility is to do it like Charles suggested in a comment to another answer, create an environment offline, and have U-Boot load it, see denx.de/wiki/view/DULG/UBootScripts
A third possibility, that I've also used sometimes, is to construct the environment offline (possibly using the same or a similar mechanism as in the link above), and the flash the environment to flash during the normal flash programming process. Though, most of the time I've done this on AT91's, using a tcl script similar to at91 Sam-Ba TCL script
No matter which method you chose, the bootcmd variable in U-Boot should hold your boot script.
The general answer is that bootcmd is run by default, and if there is persistent environment you can change the command and 'saveenv' so that it's kept.
It is easiest to modify the said bootcmd, which is executed anyway.
As an alternative to patching the kernel, it is possible to override the command in u-boot.
Create a file e.g. platform-top.h at the same place where you would place the patch file (it might already exist) and override the CONFIG_BOOTCOMMAND.
The result will look something like this:
/* ... */
/* replace the memory write with any other valid command */
#define CONFIG_BOOTCOMMAND "mw 0x1 0x1 && run default_bootcommand"
Don't forget to make the file known in your bbapend SRC_URI = "file://platform-top.h"
I am trying to interface a 16x2 LCD with Beagleboard xM using GPIO. I have done this by using a shell script and it's working very good. Now I want to achieve the same functionality by writing a kernel module. I know little bit about kernel programming as I'm in the learning phase. Need some guidance. Thanks in advance!
Writing a kernel module is different then shell scripting. You must write your own code in C++, declaring the kernel mode, and then compile it. I found one example, but don't have time to check it, so I am leaving that to you.
Here is one example of writing kernel modules, and here is one tutorial for interfacing 16x02 lcd.
If you have a script you can load it like a module in linux ,
In /etc/rcS.d folder you will find a lot of scripts like S13-some_name.sh . These scripts will be automatically run by the kernel while booting up. So you can just add your scipt here to make it as a module
So one thing have to do is find the last number used in these list of scripts and rename your driver script by prepending the next number to the last in the list
for eg:
if the last script in /etc/rcS.d is S53logger.sh
Rename your scipt as S54-name-.sh (don't forget to change attributes by chmod +x)
If the /etc/rcS.d is not present there might be a file rc.local file you just add the driver script to it
some time ago I decided to learn how to write drivers. Unfortunatelly I didn't get too far because for testing the driver it is kinda important for you to be able to unload it without the need of restarting the machine. Now I got back to it but I am just not able to get past this on my own.
Now I suppose just to make the question more straight forward you sure want to see this:
VOID Unload(PDRIVER_Object DriverObject)
{
DbgPrint("Unload\r\n");
}
and
NTSTATUS DriverEntry(PDRIVER_OBJECT pDriverObject, PUNICODE_STRING pRegistryPath)
{
DbgPrint("Entry\r\n");
pDriverObject->DriverUnload = Unload;
return STATUS_SUCCESS;
}
In DbgView it prints the "Entry" message, but when unloading the driver it doesn't print the "Unload" one. Actually trying to stop the driver service changes it's status to NOT_STOPPABLE.
Then I have to restart if I want another try.
I work with Windows 7 and with same outcomes I have tried to do it booting up with TESTSIGNING ON and with no signiture required too. A little bit odd thing is the load doesn't work either unless I create device in the entry function. Only then I can find my driver with winobj in the \Driver directory. I have also tried it out on another machine with Win7 installation but it turned out local settings are not the issue. If you recognize this kind of kernel behaviour I would really like to hear it, thanks.
Actually if know about some programm that can load .sys and then is able to unload it, drop a link or name. Source codes not required, the executable should do.. or vice versa.. thanks.
Driver development - use OSR Driver Loader
Ok the problem was I linked with /driver:wdm because I read an older tutorial. That is wrong nowadays, you can only IoCreateDevice in DriverEntry in legacy drivers that means proper option is just /driver. Sorry I dont understand the behaviour of I/O manager, but if you are having the same problem, just get rid of that wdm flag and it will unload.
I cross-compiled NET-SNMP 5.7.1 from sources to a PowerPC using ELDK-3.1.
When I try to load the snmpd daemon in my embedded board, I see the message:
# snmpd -f -Lo
pcilib: Cannot open /proc/bus/pci
pcilib: Cannot find any working access method.
Of course my PPC board has no PCI, and I wonder why is netsnmp looking for it.
In more than one place I see this same message (sourceforge, mail-archive, google-groups), but ir has no answer at all. Another variant, with a little but unhelpful responses at (archlinuxarm).
Can anybody please help me?
I'm assuming you're on a Linux target.
Net-SNMP's changelog lists "[PATCH 3057093]: allow linux to use libpci for creating useful ifDescr strings".
The configure script will search for an available libpci, and, having found one, will define
HAVE_PCI_LOOKUP_NAME and HAVE_PCI_PCI_H. To disable this code: after configuring, you can change those defines in include/net-snmp/net-snmp-config.h, then rebuild. The affected code is in agent/mibgroup/if-mib/data_access/interface_linux.c.
There's also a patch in this bug report: http://sourceforge.net/p/net-snmp/bugs/2449/
I resolved the issue using the stock snmpd that comes with the Raspbian.
In /etc/snmp/snmpd.conf file I isolated the issue to the following line
agentAddress udp:161,udp6:[::1]:161
Instead of listening on all interfaces, if I specify the the ip address of the eth0 interface i.e.:
agentAddress udp:10.0.1.5:161,udp6:[::1]:161
Then snmpd starts fine.
My speculation is that the stock snmpd tries to enumerate all possible interfaces including the pci ones.