January 2013 - kernelnewbies@kernelnewbies.org

Re: method to get backtrace of kernel function?
by Mulyadi Santosa Jan. 29, 2013

Jan. 29, 2013

Hi :) On Tue, Jan 29, 2013 at 3:53 AM, horseriver <horserivers(a)gmail.com> wrote: > On Mon, Jan 28, 2013 at 11:09:56PM +0700, Mulyadi Santosa wrote: >> On Mon, Jan 28, 2013 at 12:17 PM, horseriver <horserivers(a)gmail.com> wrote: >> > hi: >> > >> > Is there method for getting bt of a kernel functions? >> >> IIRC, you can use dump_stack() function. > > Thanks! > > what is IIRC? IIRC --> If I Remember Correctly > Can this function be used in user application? no. But I guess I see quite similar function provided by glibc/gcc. Forgot the function name though, sorry... > I am a newbie.:) So am I :) -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com

2 1

boot linux2.6 in virtual machine
by horseriver Jan. 29, 2013

Jan. 29, 2013

hi:) I have compiled a kernel of 2.6 version . there is an bzImage file generated and I have got a initrd.img file here. Now ,I want to boot this kernel , Need I have a boot-loader ? Is GRUB right for this? Need I mount kernel image and initrd.img on /dev/XXX ? If need, when to do mount them? thanks!

2 1

TCP checksum not transmited to the wire!
by Khaled Dassouki Jan. 28, 2013

Jan. 28, 2013

Hello, I am trying to use netfilter to program my own firewall. To make a forwarding test to the local machine, I made a test program that change a certain destination IP address in the PRE ROUTING HOOK to my local machine, I changed back the source IP address in the POST ROUTING hook to the old destination address. I computed TCP and IP checksums in this hook but I still have a bad TCP checksum on the receiver. I can see in the POST ROUTING hook that my checksum was changed! Below is my code #include <linux/module.h> #include <linux/kernel.h> #include <linux/proc_fs.h> #include <linux/list.h> #include <asm/uaccess.h> #include <linux/udp.h> #include <linux/tcp.h> #include <linux/skbuff.h> #include <linux/ip.h> #include <linux/netfilter.h> #include <linux/netfilter_ipv4.h> #include <net/ip.h> static struct nf_hook_ops nfho; static struct nf_hook_ops nfho_out; unsigned int hook_func_in(unsigned int hooknum, struct sk_buff *skb, const struct net_device *in, const struct net_device *out, int (*okfn)(struct sk_buff *)) { u32 local = 1090627776; //my machine IP 192.168.1.65 struct iphdr *ip_header = (struct iphdr *)skb_network_header(skb); struct tcphdr *tcp_header = (struct tcphdr *)skb_transport_header(skb); unsigned int src_port = (unsigned int)ntohs(tcp_header->source); ip_header->daddr = local; //printk(KERN_INFO "PRE: saddr: %pI4 %u, daddr: %pI4, %u", &ip_header->saddr, ip_header->saddr, &ip_header->daddr, ip_header->daddr); //printk("IN saddr: %pI4 daddr: %pI4 len: %u check: %x\n", &ip_header->saddr, &ip_header->daddr, src_port, tcp_header->check); return NF_ACCEPT; } unsigned int hook_func_out(unsigned int hooknum, struct sk_buff *skb, const struct net_device *in, const struct net_device *out, int (*okfn)(struct sk_buff *)) { struct iphdr *ip_header = (struct iphdr *)skb_network_header(skb); u32 dest = 388809389; //the old destination ip ip_header->saddr = dest; struct tcphdr *tcp_header = (struct tcphdr *)skb_transport_header(skb); tcp_header->check = 0; int len = skb->len-20; unsigned int src_port = (unsigned int)ntohs(tcp_header->source); tcp_header->check = csum_tcpudp_magic(ip_header->saddr, ip_header->daddr, len, IPPROTO_TCP, csum_partial((char *)tcp_header, len, 0)); ip_header->check = 0; ip_send_check (ip_header); //to check changes printk( "saddr: %pI4 daddr: %pI4 check: %x\n", &ip_header->saddr, &ip_header->daddr, tcp_header->check); int i; for (i=0; i<skb->len; i++) printk("%x", skb->data[i]); printk("\n"); return NF_ACCEPT; } /* Initialization routine */ int init_module() { printk(KERN_INFO "initialize kernel module\n"); /* Fill in the hook structure for incoming packet hook*/ nfho.hook = hook_func_in; nfho.hooknum = NF_INET_PRE_ROUTING; nfho.pf = PF_INET; nfho.priority = NF_IP_PRI_FIRST; nf_register_hook(&nfho); // Register the hook /* Fill in the hook structure for outgoing packet hook*/ nfho_out.hook = hook_func_out; nfho_out.hooknum = NF_INET_POST_ROUTING; nfho_out.pf = PF_INET; nfho_out.priority = NF_IP_PRI_FIRST; nf_register_hook(&nfho_out); // Register the hook return 0; } /* Cleanup routine */ void cleanup_module() { nf_unregister_hook(&nfho); nf_unregister_hook(&nfho_out); printk(KERN_INFO "kernel module unloaded.\n"); } Would you please help me, I spent 10 days working on this issue and not solved yet. Best,

2 2

method to get backtrace of kernel function?
by horseriver Jan. 28, 2013

Jan. 28, 2013

hi: Is there method for getting bt of a kernel functions? thanks!

2 1

SWAP gone when running mainline
by Martin Kepplinger Jan. 28, 2013

Jan. 28, 2013

Hi, I run a recent re-kernel (using lubuntu 12.04) and when I boot into it, "free" says I don't _have_ (and use) any SWAP. Of course on ubuntu's kernel it works. When could I start looking around? .config? sysfs knob? I have no idea. thanks -- Martin Kepplinger

3 5

Re: Can jiffies freeze?
by Mulyadi Santosa Jan. 28, 2013

Jan. 28, 2013

On Mon, Jan 28, 2013 at 2:44 PM, sandeep kumar <coolsandyforyou(a)gmail.com> wrote: > I had put that in free_highpages() arch/arm/mm/init.c... > By that time LPJ(loops for jiffies) also not calibrated... hmm, so it's ARM....ok, so maybe I am not really sure about what I can help here. But, generally, free_highpages() should not disable timer interrupt while working -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com

1 0

locking spinlocks during copy_to_user, copy_from_user
by Pablo Pessolani Jan. 25, 2013

Jan. 25, 2013

Hi: It is well know that is not recomended to keep a spinlock locked during operations that can be preempted. Some of these operations are copy_to_user, copy_from_user. Below is the code of the write_lock() as a sample. 340 static inline void __write_lock(rwlock_t *lock) 341{ 342 preempt_disable(); <<<<<< THE ISSUE IS HERE <<<<<<<<<<<<<< 343 rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_); 344 LOCK_CONTENDED(lock, _raw_write_trylock, _raw_write_lock); 345} On write_unlock() the preemtion is enable. My question is: Is there any know consequence if I enable preemption before copy_to_user/copy_from user (keeping the spinlock locked) and then disable preemption again after the copy????? i.e.: write_lock(&lock); ....... preempt_enable(); copy_to_user(.......); preempt_disable(); ....... write_unlock(&lock); Thanks in advance. PAP

2 1

Intercepting a system call
by Paul Davies C Jan. 25, 2013

Jan. 25, 2013

Hi, [1] is the module I wrote for intercepting the system call fork(). I have taken the conventional way of hooking the system call. Firstly I found out the address of the sys_call_table from the System.map. Then I used it in the module to substitute for the original fork() with my own version of fork.My version does nothing more than printing a message and returning the original fork(). It didn't gave any errors when loaded to kernel using the insmod. However it seems that the module had no affect on the sys_call_table[] since the fork() is working perfectly after the insertion of module and it does not show me the message it was supposed to print. Can someone figure out the problem? I am using 3.2.0-4-686 version of kernel. [1] http://pastebin.com/aWN3jdQU -- *Regards,* *Paul Davies C* vivafoss.blogspot.com

2 3

locking spinlocks during copy_to_user, copy_from_user
by Pablo Pessolani Jan. 25, 2013

Jan. 25, 2013

Hi: It is well know that is not recomended to keep a spinlock locked during operations that can be preempted. Some of these operations are copy_to_user, copy_from_user. Below is the code of the write_lock() as a sample. 340 static inline void __write_lock(rwlock_t *lock) 341{ 342 preempt_disable(); <<<<<< THE ISSUE IS HERE <<<<<<<<<<<<<< 343 rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_); 344 LOCK_CONTENDED(lock, _raw_write_trylock, _raw_write_lock); 345} On write_unlock() the preemtion is enable. My question is: Is there any know consequence if I enable preemption before copy_to_user/copy_from user (keeping the spinlock locked) and then disable preemption again after the copy????? i.e.: write_lock(&lock); ....... preempt_enable(); copy_to_user(.......); preempt_disable(); ....... write_unlock(&lock); Thanks in advance. PAP

2 1

Sorry, a problem with hotmail
by Pablo Pessolani Jan. 25, 2013

Jan. 25, 2013

I sent 3 identical emails to the list because hotmail does not acknowledge them. I am sorry. PAP

1 0