January 2014 - kernelnewbies@kernelnewbies.org

Understanding the LDM: Removing a device's parent
by Rick Ramstetter Jan. 10, 2014

Jan. 10, 2014

I need a strategy for handling unregistered parent devices. struct device *parent = get_ptr_to_parent(); struct device *child = kzalloc ( a device ) child->parent = get_device(parent); ... device_register(child) Later on: struct device *parent = get_ptr_to_parent(); device_unregister(parent); /* device_unregister(child) not invoked */ The parent device is not released because my child device is still holding a reference. In my case, the child device is not useful without the parent being registered. How can the child receive notification that the parent has been unregistered? The parent device and child device are being created by different drivers. I've been referencing: http://lxr.free-electrons.com/source/drivers/base/core.c#L1259 Thanks, --Rick

1 0

Need help in enable configuration when compile Kernel module
by m silverstri Jan. 10, 2014

Jan. 10, 2014

I am using 3.10.x kernel tree. My kernel module needs config VIDEOBUF2. That is defined in drivers/media/v4l2-core/Kconfig: # Used by drivers that need Videobuf2 modules config VIDEOBUF2_CORE select DMA_SHARED_BUFFER tristate So I put 'CONFIG_VIDEOBUF2_CORE=y' in my Kernel config file and compile. From the Kconfig it has CONFIG_VIDEOBUF2_CORE has no dependency and I think adding CONFIG_VIDEOBUF2_CORE=y to my kernel config should work. I am modify the right kernel config file since I set other flags like CONFIG_VIDEO_DEV=y and that works. The generated .config does not contain 'CONFIG_VIDEOBUF2_CORE=y' and the compilation fails with a bunch of undefined reference to `vb2_buffer_done' undefined reference to `vb2_buffer_done' undefined reference to `vb2_buffer_done' undefined reference to `vb2_buffer_done' I really appreciate if someone can help me with this. Thank you.

4 5

Get all available ip address in linux kernel
by Guibin(Bill) Tian Jan. 9, 2014

Jan. 9, 2014

Does anybody know how to get all the ip addresses from linux kernel? I can google some code that runs in user mode. But not in kernel. s = socket(AF_INET, SOCK_STREAM, 0); if(s < 0) { printf("socket screwup\n"); exit(1); } memset(&ifr, 0, sizeof(ifr)); ifc.ifc_len = sizeof(buff); ifc.ifc_buf = buff; if(ioctl(s, SIOCGIFCONF, &ifc) < 0) { printf("SIOCGIFCONF screwed up\n"); exit(1); } pifr = (struct ifreq *)(ifc.ifc_req); for(i = ifc.ifc_len / sizeof(struct ifreq); --i >= 0; pifr++) { strcpy(ifr.ifr_name, pifr->ifr_name); //eth0 eth1 ... if(ioctl(s, SIOCGIFADDR, &ifr) < 0) { printf("ip screwed up\n"); exit(1); } memcpy(ip, ifr.ifr_addr.sa_data+2, 4); printf("%s\n",inet_ntoa(*(struct in_addr *)ip)); }

1 0

Find inetrrupts using VMSTAT
by Anil Joshi Jan. 9, 2014

Jan. 9, 2014

Hi All, Can anyone please guide on how to find interrupts on the particular cpu in the system using vmstat? Regards Anill

2 1

Difference between prio, normal_prio, static_prio and rt_prio
by manty kuma Jan. 9, 2014

Jan. 9, 2014

Hi all, I see the following different priorities listed in task_struct. - prio, - normal_prio, - static_prio - rt_prio What do they signify? Am seeing a issue(watchdog) in Android where my worker function that pets the watchdog is waiting in the queue but couldnot be executed. Instead some process is eating the CPU. There is no Busy while loop or some thing that i can find. But the priorities are as follows : *Process that is eating the CPU :* prio = 96, static_prio = 101, normal_prio = 96, rt_priority = 3, *Kworker in which my pet_watchdg is scheduled :* prio = 100, static_prio = 100, normal_prio = 100, rt_priority = 0, Can someone explain the scenario? Am little amused as why rt_priority is set when i am working on Android which is not a realtime system. Request quick responses(I am expected to conclude within today). :) Cheers, Manty

1 0

Spinlocks & schedule_timeout
by SaNtosh kuLkarni Jan. 9, 2014

Jan. 9, 2014

Hi, I wanted to verify/clarify if using schedule_timeout after acquiring a spinlock is safe?. So basically you have a situation where you are working on a critical code and you acquire a spin lock and you end up using a schedule_timeout after say setting the current task to Interruptible state 'cause you are waiting for some user space code to fill in some buffer from which you want to read. So my doubt is simple, is it safe to acquire a spinlock and use schedule_timeout. I assume the task is put in the run queue and will "ALWAYS" be scheduled back after the specified delay in schedule_timeout. -- *Regards,Santosh*

3 2

Issue compiling Kernel Module
by Arun M Kumar Jan. 9, 2014

Jan. 9, 2014

I have recently started with Device Drivers in Linux and have made a simple Hello world module. My code is this //--------------------------------------------------------- /* My first kernel Module */ #include <linux/init.h> #include <linux/module.h> static init my_init(void){ printk(KERN_ALERT "Hello World\n"); return 0; } static void my_exit(void){ printk(KERN_ALERT "asta la vista baby\n"); } MODULE_LICENSE("GPL"); module_init(my_init); module_exit(my_exit); //------------------------------------------------------------- and the makefile i am using is like this... //------------------------------------------------------------- all: make -C /home/arun/Linux_Source M=`pwd` clean: make -C /home/arun/Linux_Source M=`pwd` clean obj-m := Hello_World.o //--------------------------------------------------------------------- I receive the following error on running "$make" //--------------------------------------------------------- make -C /home/arun/Linux_Source M=`pwd` make[1]: Entering directory `/home/arun/Linux_Source' ERROR: Kernel configuration is invalid. include/generated/autoconf.h or include/config/auto.conf are missing. Run 'make oldconfig && make prepare' on kernel src to fix it. make[1]: *** No rule to make target `Device'. Stop. make[1]: Leaving directory `/home/arun/Linux_Source' make: *** [all] Error 2 //---------------------------------------------------------- PS: The kernel Source i am using is a Copy of my Git Repository and is a copy of the branch 3.10.25. my system details are: Linux linux-nwya.site 3.10.25-2.10-desktop #2 SMP PREEMPT Tue Jan 7 00:25:03 IST 2014 i686 i686 i386 GNU/Linux

3 2

How can I add a new program to 'samples' directory
by m silverstri Jan. 8, 2014

Jan. 8, 2014

I am trying to add a sample 'v4l2-mfc-encoder' program to '{raspberrypi dir}linux/samples/' The 'v4l2-mfc-encoder' program's Makefile is built using 'arm gcc': # Toolchain path TCPATH = arm-linux-gnueabi- KERNELHEADERS = /home/kamil/praca/w1-party/kernel/headers/include CC = ${TCPATH}gcc AR = "${TCPATH}ar rc" AR2 = ${TCPATH}ranlib make -j4 But when I compile raspberry bi source, I get error saying: /usr/bin/arm-linux-gnueabi-ld: cannot find samples/v4l2-mfc-encoder/built-in.o: No such file or directory make[1]: *** [samples/built-in.o] Error 1 make[1]: Target `__build' not remade because of errors. make: *** [vmlinux] Error 2 make: Target `_all' not remade because of errors. What does it mean by 'built-in.o' No Such File or directory? There is no built-in.o in my 'v4l2-mfc-encoder' Make file.

1 0

Missing parallel port interrupts
by Eric Fowler Jan. 8, 2014

Jan. 8, 2014

I am writing a simple driver that hooks the parallel port irq and when written to, enables the port for generating interrupts (the pins 9&10 are wired together) and then toggles the high bit once for every byte in the written input. This means that if I do: >echo "0123456789" >> /dev/foobar I expect to toggle that bit 10 times, and generate 10 interrupts. My ISR is a simple counter that bumps a number every time it is called. I am seeing that I am missing about 85% or more of my interrupts. Sometimes I get 2/10 but more often only 1/10. I have tried udelay()-ing between the port toggles but no go, and of course that is not really a solution. I am not trying to protect the counter but nothing else is accessing it. I would expect to see *some* losses here, but not that many. What is going on? Code follows. Thanks loads Eric ///////////////////////////////-----------cut here------------------------ #include <linux/kernel.h> #include <linux/module.h> #include <linux/fs.h> #include <linux/ioport.h> #include <linux/cdev.h> #include <linux/slab.h> #include <linux/init.h> #include <linux/delay.h> #include <asm/uaccess.h> #include <linux/seq_file.h> #include <linux/interrupt.h> MODULE_LICENSE("GPL"); #define NUM_DEVICES (1) #define DEVNAME "foobar" #define FOOBAR_MAJOR (248) #define FOOBAR_MINOR (0) int foobar_init(void); void foobar_exit(void); int foobar_release(struct inode *, struct file*); ssize_t foobar_read(struct file *, char __user *, size_t, loff_t *); ssize_t foobar_write(struct file *, const char __user *, size_t, loff_t *); int foobar_open(struct inode * inode, struct file * file); void set_port_and_irq(void); static irqreturn_t irqhandler(int irq, void * data); int foobar_set_irq(void); int port_address = -1; module_param(port_address, int, 0); unsigned int irq = 0; module_param(irq, uint, 0); int irqavail = -1; size_t chars_read = 0; static struct file_operations foobar_i_fops = { .owner = THIS_MODULE, .read = foobar_read, .write = foobar_write, .open = foobar_open, .release = foobar_release, }; struct resource * region = NULL; static unsigned long mask; static unsigned long probe; static dev_t g_dev; unsigned dev_id = (unsigned)&foobar_i_fops; int foobar_init(void) { int result = 0; result = register_chrdev(0, DEVNAME, &foobar_i_fops); g_dev = MKDEV(result, FOOBAR_MINOR); if(result < 0) { printk(KERN_WARNING "foobar: can't get device number, returning %d", result); return result; } printk(KERN_WARNING "foobar: Module %s got device number %d, minor is %d, result is %d\n", THIS_MODULE->name, MAJOR(g_dev), MINOR(g_dev), result); g_dev = MKDEV(MAJOR(g_dev), FOOBAR_MINOR + NUM_DEVICES); set_port_and_irq(); region = request_region(port_address, 1, DEVNAME); printk(KERN_WARNING "Port address is 0x%x, IRQ is %d, region is 0x%p\n", port_address, irq, region); if(region == 0) { printk(KERN_WARNING "Can't get region"); return -ENODEV; } mask = probe_irq_on(); return 0; } void foobar_exit(void) { int major = MAJOR(g_dev); release_region(port_address, 1); printk(KERN_WARNING "unregister_chrdev(%d) called for %s, dev # is %d\n", major, DEVNAME, g_dev); unregister_chrdev(major, DEVNAME); } int foobar_open(struct inode * inode, struct file * file) { irqavail = request_irq(irq, irqhandler, IRQF_SHARED, DEVNAME, (void*)dev_id); printk(KERN_WARNING "IRQ:%d\tPort Address:0x%x\tAvailable:%s\n", irq, port_address, irqavail == 0 ? "true" : "false"); return irqavail; } int foobar_release(struct inode * inode, struct file * filp) { printk(KERN_WARNING "foobar_release() is called, fil is 0x%p, inode is 0x%p\n", filp, inode); if(irqavail == 0) { printk(KERN_WARNING "Freeing irq %d, dev_id is 0x%x\n", irq, dev_id); free_irq(irq, (void*)dev_id); } else printk(KERN_WARNING "NOT freeing irq %d, dev_id is 0x%x\n", irq, dev_id); probe = probe_irq_off(mask); return 0; } ssize_t foobar_read(struct file * filp, char __user * buf, size_t count, loff_t * f_pos) { if(chars_read == 0) { printk(KERN_WARNING "Read is called, nothing to return."); return 0; } sprintf(buf, "Chars>%d\n", chars_read); chars_read = 0; printk(KERN_WARNING "Read is called, returning %s, length is %d\n", buf, strlen(buf)); return count; } ssize_t foobar_write(struct file * filp, const char __user * buf, size_t count, loff_t *f_pos) { size_t idx; unsigned long address = (unsigned long)port_address; printk(KERN_WARNING "Write is called, count is %d, address is 0x%lx, buffer is %s\n", count, address, buf); outb_p(0x10, address + 2); for(idx = 0; idx < strlen(buf); ++idx) { outb_p(0x00, address); udelay(100); outb_p(0xff, address); udelay(100); } outb_p(0x00, address + 2); udelay(5); printk(KERN_WARNING "probe_irq_off(0x%lx) returned 0x%lx", mask, probe); if(f_pos) * f_pos = 0; return count; } static irqreturn_t irqhandler(int irq, void * data) { chars_read++; printk(KERN_WARNING "Interrupt is fired, IRQ is %d, data is 0x%p, chars %d\n", irq, data, chars_read); return IRQ_HANDLED; } void set_port_and_irq(void) { if(irq < 0) switch(port_address) { case 0x378: irq = 7; break; case 0x278: irq = 2; break; } switch(irq) { case 2: port_address = 0x278; break; case 7: port_address = 0x378; break; } } module_init(foobar_init); module_exit(foobar_exit);

1 0

Error in Make for Kernel Module
by Arun M Kumar Jan. 7, 2014

Jan. 7, 2014

I am starting with Device Drivers for Linux System Details: Linux linux-nwya.site 3.10.25-2.10-desktop #2 SMP PREEMPT Tue Jan 7 00:25:03 IST 2014 i686 i686 i386 GNU/Linux I created a Simple Hello_world Kernel Module. "Hello_World.c" //--------------------------------------------------------- /* My first kernel Module */ #include <linux/init.h> #include <linux/module.h> static init my_init(void){ printk(KERN_ALERT "Hello World\n"); return 0; } static void my_exit(void){ printk(KERN_ALERT "Exit\n"); } MODULE_LICENSE("GPL"); module_init(my_init); module_exit(my_exit); //--------------------------------------------------------------------- The Makefile is as follows: The file name is "Makefile" //--------------------------------------------------------------------- obj-m := Hello_World.o //--------------------------------------------------------------------- on running the make command i get the following error. //---------------------------------------------------------------------------- make: Entering directory `/home/arun/Linux_Source' ERROR: Kernel configuration is invalid. include/generated/autoconf.h or include/config/auto.conf are missing. Run 'make oldconfig && make prepare' on kernel src to fix it. make: *** No rule to make target `Device'. Stop. make: Leaving directory `/home/arun/Linux_Source' //------------------------------------------------------------------------------- I have tried "make prepare" but the issue remains.... any suggestions would be appreciated. Thanks in advance.

2 1