- kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

How to check if my kernel driver is leaking memory
by m silverstri March 13, 2014

March 13, 2014

Hi, I am developing a kernel driver. What should I test to make sure my kernel driver is not leaking memory? 1. under normal operation (when applications open and close my driver properly) 2. in error situation (when application open my driver and then it crashes without close my driver property) How can I find out the memory usage used by my driver? so that I can monitor it over time. Thank you.

5 5

where is the printf source for busybox?
by Kim Chan March 13, 2014

March 13, 2014

Hi, I've had this question for some time past and I would like to know the answer now.. I wanted to follow a printf in busybox but couldn't find the source of the printf function. Then I throught probably the printf is provided by the system library where printf is connected to proper linux system call. (Is it printk?) So I checkd LD_LIBRARY_PATH of my shell and found libc.so under /lib. I could see by 'nm libc.so' that printf is in the library. But where can I see the printf source? Is it under gcc source? in GNU sdk, under glibc-2.16.0, I tried # grep 'int printf(' * -r but nothing comes up. In linux kernel source doing the same gaves me arch/x86/boot/boot.h:int printf(const char *fmt, ...); arch/x86/boot/printf.c:int printf(const char *fmt, ...) arch/um/include/shared/kern.h:extern int printf(const char *fmt, ...); arch/um/os-Linux/user_syms.c:extern int printf(const char *, ...); arch/powerpc/boot/stdio.h:extern int printf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); so printf is defined only for x86 architecture? I understand printf is for user program not kernel, so I want to see how the printf is connected to kernel print function. (the system call, I know in sparc, linux system call is trap 0x90, function 59 when using assembly) Where can I find the printf (user program) source? Thanks in advance. Chan

3 9

kernel build problems with KDB patch
by RS March 13, 2014

March 13, 2014

Hello I want to use KDB to debug the module. I have done this: 1. make mrproper 2. download the patch from the site (ftp://oss.sgi.com/www/projects/kdb/download/) Include kdb-v4.4-2.6.32-common-6.bz2 and kdb-v4.4-2.6.32-x86-6.bz2 3. patch -p1 < kdb-v4.4-2.6.32-common-6 patch -p1 < kdb-v4.4-2.6.32-x86-6 4. make menuconfig Kernel hacking ---> Compile the kernel with frame pointers(set Y) Built-in Kernel Debugger support (set Y) KDB off by default(set Y) 5. make ERROR LOG: arch/x86/kernel/traps.c:418 error:expected identifier or '(' before 'if' arch/x86/kernel/traps.c:437 error:expected identifier or '(' before 'if' arch/x86/kernel/traps.c:441 error:expected identifier or '(' before 'if' ... conflicting types for 'reassert_nmi' ...note:previous definition of 'reassert_nmi' was here ... What should I do ? Thanks, HeChuan

2 1

Re:Re: A new way to dive into the kernel! [Eudyptula Challenge]
by Rahul Garg March 12, 2014

March 12, 2014

Hi all, I am also thinking of taking this challenge. But I found very limited information on submission. So can you help me what we need to submit for the Assigment 1, will dmesg (or /proc/kmsg) output and module(assignement1.c and Makeflie) be enough ?? Also is there any group of forum where we can discuss assignments ? Regards Rahul

6 9

Does register_chrdev function make device file under /dev ?
by 김찬 March 12, 2014

March 12, 2014

Hi, This should be a basic question. Does the regisetr_chrdev function make the device file under /dev directory? I am running linux on our embedded system and inside vcs_init, register_chrdev is called as below. int __init vcs_init(void) { unsigned int i; if (register_chrdev(VCS_MAJOR, "vcs", &vcs_fops)) panic("unable to get major %d for vcs device", VCS_MAJOR); I can see the function is returning ok but I don't see any vcs* file under /dev. Does it only register the device on /sys or /proc directory and not under /dev? Thanks in advance. Chan

5 6

Not mounting rootfs
by Flavio Ceolin March 12, 2014

March 12, 2014

Hi folks, I trying to boot a kernel with a rootfs generated with buildroot under qemu but with any success, the output is: [ 0.814014] Magic number: 6:562:333 [ 0.814522] tty tty54: hash matches [ 0.815121] rtc_cmos 00:00: setting system clock to 2014-03-11 14:19:08 UTC (1394547548) [ 0.820886] List of all partitions: [ 0.821097] No filesystem could mount root, tried: [ 0.821399] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) [ 0.821877] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.14.0-rc5-ARCH-00265-gb01d4e6 #1 [ 0.822184] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 [ 0.822550] ffffffffffffffff ffff880007107df8 ffffffff814f7bc4 ffffffff816e5af0 [ 0.822903] ffff880007107e70 ffffffff814f4749 ffffffff00000010 ffff880007107e80 [ 0.823216] ffff880007107e20 ffff880007107e28 ffff880007107e90 0000000000000012 [ 0.823421] Call Trace: That is the way I'm invoking qemu: qemu-system-x86_64 -boot c -m 128 -nographic -kernel ./path/to/bzImage -hda ./buildroot-2013.05/output/images/rootfs.ext2 -append "root=/dev/ram0 console=ttyS0 init=/sbin/init" Any tip is extremely welcome :) --- Flavio Ceolin

2 1

How to use sparse to check my driver
by m silverstri March 12, 2014

March 12, 2014

I am trying to use sparse to check my driver that I developed. >From http://kernelnewbies.org/Sparse, It said 'make C=2 drivers/staging/wlan-ng/', what does it check it I do 'make C=2' and how does it actually call 'spare'? Do I need to modify my Makefile to add a new target or something? Thank you.

2 1

Please help.. trying to use USB keyboard to open a shell on LCD display
by Kim Chan March 12, 2014

March 12, 2014

Hello, folks, It is a comfort to me that I have people to ask questions here and I hope I could be answering to some hard quetions someday. I've beeing trying to attach a USB keyboard to our embedded system to open a shell on the LCD display using framebuffer console. The USB host controller is from the opencore(USB1.1) and it works for USB memory. Some of the boot arguments are (I've tried some variants) "console=tty1 " "fbcon=map:0000,font:SUN8x16,vc:1-3 " "root=/dev/ram " The problem is, after the final sys_execve() which executes /sbin/init of the ramdisk, I cannot see any printf results frome the busybox init_main function not to mention the shell. So I cannot use shell commands like getty. Using uart as the console I can see printf results in the busybox init_main routine. (I use "console=ttyS,mmio,0xff003000,115200n8 " in that case) Using some prints on separate uart print routine, I can see from the kbd_event function that the key value is going to current vc which is using tty1. The key value is then handed to put_queue function where the character is somehow copied to tty buffer. Flush_to_ldisc is scheduled for it and it will probably get passed on to the virtual console. By the way, I guess the keyboard data is delivered to current fg_console (vc 0 which is using tty1). Added to this, there is a notifier_call_chain and the notifier block is registered by vcs_poll_data_get function which I don't know weather is necessary or not. So one of my questions is if vcs is necessary for my case. (I guess not. There is no other notifier blocks registered) I gathered that without inittab (I don't use inittab which should be ok for most cases) the assumed default inittab tty2::askfirst:-/bin/sh tty3::askfirst:-/bin/sh tty4::askfirst:-/bin/sh is being applied by the busybox so there maybe 3 shells on tty2,tty3, and tty4 running somewhere. But even if there are shells running I don't know how to switch to those ttys. I tried using Ctrl-Alt F2 (or F3, F4) using the USB keyboard to switch to tty2,3,4 to see if any shell comes up on the LCD but couldn't see it. I tried using inittab with no avail before. And I don't know how the printf in busybox is connected to the kernel. I guess busybox talks with stdio(stdout,stdin..) and the file is connected to the kernel anyhow. I've digged up how vc driver is connected to the console switch functions and to the bit blitting functions and how the printk is handled in the kernel. But these console, tty, framebuffer console.. things are very complex to me. and I hope somebody could point me to what should be checked. Any hint will be deeply appreciated. Best Regards, Chan _______________________________________________ Kernelnewbies mailing list Kernelnewbies(a)kernelnewbies.org<mailto:Kernelnewbies@kernelnewbies.org> http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies

1 0

building kernel modules for linux differnet version
by Saket Sinha March 11, 2014

March 11, 2014

Hi, I have a scenario where I need to build a driver for different kernel versions. I have kernel-headers of different version in /lib/modules/ 2.6.32-xxxx 2.6.39-xxxx and currently I am booted into 2.6.32-xxxx(which I find out by uname -r) Now in my driver makefile, I generate .ko of driver make -C /lib/modules/$(shell uname -r)/build M=$(PWD) modules but I notice that the driver .ko gets generated for 2.6.39-xxxx though I am booted into 2.6.32-xxxx. Can anyone help me figure out why this is happening? Regards, Saket Sinha

3 5

Release DMA Memory
by Matt Davis March 11, 2014

March 11, 2014

I have a kernel module that communicates to a PCI device via DMA. The module allocates the shared memory using dma_alloc_coherent(). When the module is unloaded, a call to dma_free_coherent() is executed. My question is, since the DMA memory is not explicitly zeroed before deallocation, is there a chance that the PCI device could still read (non-zeroed data) and act upon the DMA memory after the module unloads and calls dma_free_coherent()? Thanks! -Matt

2 1