February 2011 - kernelnewbies@kernelnewbies.org

setup_timer
by Mohammed Shafi Feb. 24, 2011

Feb. 24, 2011

Hi all, can some please explain what does setup_timer does. thanks in advance, shafi

2 2

Issue on 2.6.35.9 kernel with module insertion when Rootfs is NFS mounted
by naveen yadav Feb. 24, 2011

Feb. 24, 2011

Hi All, When I am trying to insert some modules on 2.6.35.9, I am getting some random crash's. There are 2 scenarios:- 1) When my rootfs is NFS mounted. In this case, when I insmod modules some get inserted and some gives crash. I have tried with following modules :- a) ext2.ko ; size 93K ; status - successfully inserted b) ext3.ko ; size 188K ; status - insertion failed c) xfs.ko ; size 823K ; status - insertion failed d) usbcore.ko ; size 243K ; status - insertion failed 2) When I created kernel Image using Initramfs, hence making all modules part of ramfs image all insertions are successfull In case of rootfs as NFS mounted here are some of the crash logs :- Crash 1 Linux#> insmod usbcore.ko Module len 238547 truncated insmod: can't insert 'usbcore.ko': invalid module format Crash 2 Linux#> insmod usbcore.ko CPU 0 Unable to handle kernel paging request at virtual address 00100100, epc == 8008fb00, ra == 8008cf9c Oops[#1]: Cpu 0 $ 0 : 00000000 00000001 84166e00 00100100 $ 4 : 84166e00 8415e870 84166e00 8415e870 $ 8 : 84086720 83c12178 00000000 000081a4 $12 : 00000000 00000000 01a0c43d 00000000 $16 : 84166e00 83c12120 8410de58 00000000 $20 : 84166100 83c0b380 00000024 00002001 $24 : 00000000 800f058c $28 : 8410c000 8410ddb8 00000000 8008cf9c Hi : 000bf0c5 Lo : 6ab94a84 epc : 8008fb00 file_move+0x20/0x68 Not tainted ra : 8008cf9c __dentry_open+0x110/0x278 Status: 11008c03 KERNEL EXL IE Cause : 0080000c BadVA : 00100100 PrId : 00019555 (MIPS 34Kc) Modules linked in: Process insmod (pid: 260, threadinfo=8410c000, task=84136b20, tls=005c9470) Stack : 8410de58 8410de68 83c0b380 00000001 8410de58 84166e00 00000000 8410de58 8410de68 00000000 00002000 8008d1ec 83c06080 8410de58 00000000 83c06080 84086720 00000000 00000000 80099f4c ebb5351b 80380000 84086720 803177a0 8410de68 800900c0 ffffffe9 00000003 00002000 8410de68 ffffff9c 00000024 00000002 8009c23c 000003fc 00000000 84150420 84136b20 00000000 8400a000 ... Call Trace: [<8008fb00>] file_move+0x20/0x68 [<8008cf9c>] __dentry_open+0x110/0x278 [<8008d1ec>] nameidata_to_filp+0x38/0x5c [<80099f4c>] do_last+0x58c/0x6fc [<8009c23c>] do_filp_open+0x1c0/0x62c [<8008ce04>] do_sys_open+0x6c/0xc8 [<80002164>] stack_done+0x20/0x3c Code: af820014 8c830004 8c820000 <ac620000> 8ce40000 ac430004 acc40000 ac860004 ace60000 Disabling lock debugging due to kernel taint note: insmod[260] exited with preempt_count 1 BUG: scheduling while atomic: insmod/260/0x10000002 Modules linked in: Call Trace: [<800051b4>] dump_stack+0x8/0x34 [<800054e0>] schedule+0x9c/0x4b4 [<80005a20>] _cond_resched+0x3c/0x60 [<80077788>] unmap_vmas+0x594/0x690 [<8007bf2c>] exit_mmap+0xc0/0x1a4 [<80024474>] mmput+0x3c/0x118 [<80028a88>] exit_mm+0x130/0x164 [<8002a89c>] do_exit+0x1b4/0x624 [<8000e3e8>] default_cu2_call+0x0/0x4c Segmentation fault I tried to debug myself and put some print in kernel/module.c function load_module, printing the length and offset of each section header. I found that on some sections the values are wrongly shown Linux#> insmod usbcore.ko load_module: umod=2aaa8008, len=238950, uargs=005c2f70 hdr->e_shnum 601 <= number of sections shown correctly header type sechdrs[i].sh_type =7 sechdrs[i].sh_offset=52 sechdrs[i].sh_size=36 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=96 sechdrs[i].sh_size=0 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=96 sechdrs[i].sh_size=72 <= correct header type sechdrs[i].sh_type =9 sechdrs[i].sh_offset=154448 sechdrs[i].sh_size=8 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=168 sechdrs[i].sh_size=64 <= correct header type sechdrs[i].sh_type =9 sechdrs[i].sh_offset=154456 sechdrs[i].sh_size=8 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=232 sechdrs[i].sh_size=64 <= correct header type sechdrs[i].sh_type =2136364925 sechdrs[i].sh_offset=4286298862 sechdrs[i].sh_size=1602090478 <= Incorrect header type sechdrs[i].sh_type =-1386227977 sechdrs[i].sh_offset=4293914174 sechdrs[i].sh_size=1429526523 <= Incorrect header type sechdrs[i].sh_type =-780708260 sechdrs[i].sh_offset=1579921686 sechdrs[i].sh_size=1562763221 <= Incorrect header type sechdrs[i].sh_type =-571181733 sechdrs[i].sh_offset=1875703162 sechdrs[i].sh_size=3600104807 <= Incorrect header type sechdrs[i].sh_type =9 sechdrs[i].sh_offset=182792 sechdrs[i].sh_size=32 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=69044 sechdrs[i].sh_size=32 <= correct header type sechdrs[i].sh_type =9 sechdrs[i].sh_offset=182824 sechdrs[i].sh_size=32 <= correct header type sechdrs[i].sh_type =1 sechdrs[i].sh_offset=69076 sechdrs[i].sh_size=48 <= correct The prints goes on this way and lastly the insertion failed with this message :- CPU 0 Unable to handle kernel paging request at virtual address ed79cc5b, epc == 80054254, ra == 800541d8 Oops[#1]: Please suggest me some ideas to debug as to why when rootfs is NFS mounted I am not able to load modules.

2 5

Problem with PCI Driver on Ubuntu 10.10 i386
by Adhyas Avasthi Feb. 23, 2011

Feb. 23, 2011

I am trying to develop a very basic PCI device driver for a dummy device I have created in qemu. The device has no functionality except that it sits on the PCI bus behind two bridges (bridge 0 on root bus, bridge 1 only device behind bridge 0, and device behind bridge 1 as the only device). It is a single function do nothing device with one MMIO BAR of 0x10000 size. The BAR is initialized by the kernel at some MMIO location but the device is not disabled, as there is no driver. I have used one of the available vendor ids for my work. My driver is not loaded automatically, so I load the driver manually after the boot of the VM, and then issues a rescan of the entire PCI bus using "echo 1 > /sys/bus/pci/rescan", but nothing happens. The probe function of my driver is never invoked, neither does the device gets visible in lspci output. Can someone help me identify what I might be doing wrong? I have attached the driver code, the lspci output, the lspci -H1 output, the lspci -H1 -xxx -s 02:02.0 output that details the registers of the device. -- Adhyas ******************************************************************** Two types have compatible type if their types are the same. — ANSI C Standard, 3.1.2.6. ********************************************************************

2 2

Re: what is __get_cpu_var() ?
by Dave Hylands Feb. 23, 2011

Feb. 23, 2011

HI Murali, On Wed, Feb 23, 2011 at 11:26 AM, Murali N <nalajala.murali(a)gmail.com> wrote: > Hi Dave, > > On Wed, Feb 23, 2011 at 11:15 AM, Dave Hylands <dhylands(a)gmail.com> wrote: >> Hi Murali, >> >> On Wed, Feb 23, 2011 at 10:34 AM, Murali N <nalajala.murali(a)gmail.com> wrote: >>> Hi Dave, >>> thanks for your reply. >> ...snip... >>>> get_cpu_var returns the contents of a per-cpu variable. >>>> >>>> __get_cpu_var contains the actual machine-dependant implementation. It >>>> looks like all of the architectures use the one in >>>> asm-generic/percpu.h >>>> >>>> In general, all of the per-cpu data is gathered together into a >>>> section. Multiple sections are allocated (one per CPU). I think that >>>> the address of the variable is really the offset within the section, >>>> and each allocated section is cache-line aligned. This offset is then >>>> added to the "offset for my cpu" to come up with the final address of >>>> the variable, which is dereferenced as a pointer dereference. There >>>> are lots of extra doo-dads to get around warnings, and to prevent the >>>> linker from producing relocation references for for the variable >>>> access (since it looks like an access of a global variable, but it's >>>> really just doing a game of using the offset of the variable within >>>> the section). >>>> >>>> So you could think of it as a very fancy offsetof macro. >>>> >>>> There are several other macros involved, perhaps you could be a bit >>>> more specific about your request? >>>> >>>> Dave Hylands >>>> >>> >>> I have one more basic question. >>> Why would we need to maintain structures like this? Is there any >>> advantage we get here? >> >> Primarily for performance reasons. For example, the kernel maintains >> lots of stats on threads and processes (I haven't looked to see if >> these are actually maintained on a per-cpu basis, but the concept >> applies). these stats are updated frequently, but only accessed >> occaisonally. If you have a global "database" of stats, then each CPU >> needs to lock the data, which creates lots of contention. By keeping >> stuff per-cpu, the cpus don't need to acquire any locks (or at the >> very least won't cause as much contention when acquiring per-cpu >> locks). This becomes especially important when there are lots of cpus. >> >> The query functions can then amalgamate the information and present it >> as if it were maintained in a global database. >> >> So if you have data which is updated frequently and only accessed >> occaisonally, or updated infrequently and accessed frequently, then >> you might have a case for using per-cpu-data. Of course you'd still >> need to profile it and see if it makes sense. >> >> Also keep in mind, that some things might not seem like it matters >> much for say a dual-core, but could make a considerable difference >> with say 32 cores. >> >> Dave Hylands >> > > So it make sense to use if i am running on more cores ( > 4 ). It really depends on the access patterns of the data. Whether it makes sense or not is something you'll probably need to profile (i.e. with and without using per-cpu variables). Dave Hylands

1 0

what is __get_cpu_var() ?
by Murali N Feb. 23, 2011

Feb. 23, 2011

Hi, can somebody explain me what "__get_cpu_var()" macro does? I try to understand this macro but i couldn't, its representation is weired!!! -- Regards, Murali N

2 4

Some doubts on MUTEX
by subin gangadharan Feb. 23, 2011

Feb. 23, 2011

Hi All, I am trying to understand how to use mutex API's properly,so while going through the documentation, there is section mentioning fast path and slow path for mutexes. For your reference I am pasting this here.(kernel/mutex.c) /* * We split the mutex lock/unlock logic into separate fastpath and * slowpath functions, to reduce the register pressure on the fastpath. * We also put the fastpath first in the kernel image, to make sure the * branch is predicted by the CPU as default-untaken. */ static __used noinline void __sched __mutex_lock_slowpath(atomic_t *lock_count); Can someone help me to understand what is the difference between fastpath lock and slowpath lock? Thanks in advance Subin Gangadharan

2 4

Re: adding a system call
by mohit verma Feb. 22, 2011

Feb. 22, 2011

mulyadi , i followed the link https://sites.google.com/a/ce-jmi.edu.in/tarique/adding-a-system-call-to-li… . modified files are: 1. /usr/src/kernels/linux-2.6.37/arch/x86/kernel/syscall_table_32.S 2. /usr/src/kernels/linux-2.6.37/include/asm-frv/unistd.h 3. /usr/src/kernels/linux-2.6.37/include/linux/syscalls.h 4. /usr/src/kernels/linux-2.6.37/Makefile and i added: 1. #define __NR_hariohm 264 __SYSCALL(__NR_hariohm, sys_hariohm) in unistd.h file. 2.asmlinkage int sys_hariohm(const char __user *filename); in file syscalls.h 3. .long sys_hariohm /* our system call */ in syscall_table_32.S 4 core-y += kernel/ mm/ fs/ ipc/ security/ crypto/ block/ *hariohm/* in makefile On Tue, Feb 22, 2011 at 11:40 PM, Mulyadi Santosa <mulyadi.santosa(a)gmail.com > wrote: > On Wed, Feb 23, 2011 at 01:01, mohit verma <mohit89mlnc(a)gmail.com> wrote: > > hi mulyadi, > > > > i used syscall number 264 . and for the coding standard , can u tell me > > where am i off the track ? > > Exactly, how do you add "264" as the syscall number? the diff...I mean... > > NB: find lindent script for style reformatting... > > -- > regards, > > Mulyadi Santosa > Freelance Linux trainer and consultant > > blog: the-hydra.blogspot.com > training: mulyaditraining.blogspot.com > -- ........................ *MOHIT VERMA*

3 3

Re: adding a system call
by Mulyadi Santosa Feb. 22, 2011

Feb. 22, 2011

On Wed, Feb 23, 2011 at 01:01, mohit verma <mohit89mlnc(a)gmail.com> wrote: > hi mulyadi, > > i used syscall number 264 . and for the coding standard , can u tell me > where am i off the track ? Exactly, how do you add "264" as the syscall number? the diff...I mean... NB: find lindent script for style reformatting... -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com

1 0

adding a system call
by mohit verma Feb. 22, 2011

Feb. 22, 2011

hi all , i followed the link and tried to add a system call in kernel-version 2.6.37 here is the code : #include<linux/linkage.h> #include<linux/kernel.h> #include<linux/fs.h> //for getname #include<linux/err.h> //for PTR_ERR asmlinkage int sys_hariohm(const char __user *filename) { char *tmp = getname(filename); int fd=PTR_ERR(tmp); if (!IS_ERR(tmp)) { printk("filename %s \n",tmp); putname(tmp); return fd; } else { return 2; } } the compilation process was silent (without any error). but when i try to boot into the kernel having this system call , i get the errors like : do_exit() ,do_group_exit() , syscall_init() error lines with some addresses. can someone help me understand what is wrong with this code? thanks in advance :) -- ........................ *MOHIT VERMA*

2 3

Re: new link list
by Javier Martinez Canillas Feb. 22, 2011

Feb. 22, 2011

> if i add a new system call and define a new link list in that and some > kernel function want to access that list > is it possible for that kernel code to make use of this? > Since the Linux kernel is monolithic, all kernel code is in the same memory address space. So it will depend of the scope of your variables. For example if your structure is global, it will be accesible from any place in the kernel. > actually i am making changes 2 side : one introducing a new kernel function > and that kernel function has to use the link list created by the system call > which i make in system call code. > Generally is not a good idea to introduce a new system call. You have to make changes in many places an recompile the kernel to make it work. It is better to use netlink or other mechanism to communicate kernel and userspace. > how can i intigrate them so that the things become easier? > > any further help is appreciable , javier. :) > I still don't know your context but a common approach is to create a kernel module that has both your data structures (linked list) and some way export that data to userspace (netlink sockets, sysfs, procfs, virtual devices, etc) -- ----------------------------------------- Javier Martínez Canillas (+34) 682 39 81 69 PhD Student in High Performance Computing Computer Architecture and Operating System Department (CAOS) Universitat Autònoma de Barcelona Barcelona, Spain

1 0