July 2013 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

shmem info issue
by naveen yadav July 12, 2013

July 12, 2013

Hi All I am working on tmpfs. During code analysis I found shmem driver register itself as tmpfs . cat /proc/meminfo Shmem field read NR_SHMEM enum filed and shows used memory in tmpfs [root@localhost linux-3.8.2]# cat /proc/meminfo | grep -r Shmem Shmem: 704 kB [root@localhost linux-3.8.2]# include/linux/mmzone.h enum zone_stat_item { -----cut here------- NR_SHMEM, /* shmem pages (included tmpfs/GEM pages -----cut here } I have the following query 1-> cat /proc/meminfo | grep -r Shmem show only the used tmpfs memory or it show used tmpfs memory + mata data also (because I found Shmem is bit larger then tempfs used memory ) 2-> as per the developer comments NR_SHMEM included tmpfs and GEM pages whct is Gem pages 3-> is my above code analysis is correct?

2 1

DEFINE() Macro defined in /linux/kbuild.h
by 王玮鸿 July 12, 2013

July 12, 2013

Hi, There are three macro(DEFINE, BLANK, OFFSET) contained in /linux/kbuild.h in kernerl version 2.6.32.60. All of them are inline assembly statements. I don't know what they means. Could anyone please explain them to me. Also I have written a test module to watch what they will print or do something, however I failed. I couldn't compile it . Here is my code, I just steal it from arch/x86/kernel/asm-offsets_32.c: #define PZ 4096 static int __init test_init(void) { DEFINE(PZ_asm, PZ); BLANK(); OFFSET(INODE, inode, i_hash); return 0; } And here is message when I compiled it: /home/kiongf/kernel_module_test/test/test.c: Assembler messages: /home/kiongf/kernel_module_test/test/test.c:13: Error: junk at end of line, first unrecognized character is `-' /home/kiongf/kernel_module_test/test/test.c:14: Error: junk at end of line, first unrecognized character is `-' /home/kiongf/kernel_module_test/test/test.c:15: Error: junk at end of line, first unrecognized character is `-' Did I do something wrong? Thanks.

1 0

System call porting
by Saket Sinha July 12, 2013

July 12, 2013

This is to discuss the problems I am facing with porting a driver from a 2.6.18 kernel to 3.8.3 kernel. Apart from the APIs and functions that have changed(which I have more or less replaced), the system call implementation has changed. http://lxr.linux.no/linux+v2.6.18/arch/i386/kernel/syscall_table.S#L41 http://lxr.linux.no/linux+v3.8.3/arch/x86/syscalls/syscall_64.tbl#L92 Now due to this, some unnecessary warnings and errors are coming. To explain this I am providing the compiler error which is not coming on 2.6.18 kernel but is coming on 3.8.3 kernel. the below line is generating https://github.com/HeisSpiter/hepunion/blob/master/fs/hepunion/helpers.c#L4… mode &= ~current->fs->umask; the following error- ERROR:Derefencing pointer to incomplete type This has nothing to do with any APIs or change in the structure in the newer kernel. This just the importation of mkdir code from the linux kernel. Likely to be sys_mkdir, IIRC. Talking of which, I'd think that I must sync ALL the importations from the Linux kernel with mine. Regards, Saket Sinha

3 2

tcpdump hang and system gone in freeze state
by Varun Sharma July 12, 2013

July 12, 2013

Hi, I build new kernel (2.6.32.60). when i execute tcpdump on this build kernel ,tcpdump gone in hang state and system gone in freeze state . ltrace tcpdump output is--- __libc_start_main(0x452220, 1, 0x7fff623b6158, 0x462d70, 0x462d60 <unfinished ...> strrchr("tcpdump", '/') = NULL getopt(1, 0x7fff623b6158, "aAB:c:C:dDeE:fF:G:i:IKlLm:M:nNOp"...) = -1 time(NULL) = 1373602325 gmtime(0x7fff623b5e68) = 0x313eba03e0 localtime(0x7fff623b5e68) = 0x313eba03e0 getuid() = 0 pcap_lookupdev(0x7fff623b5f30) = "eth0" pcap_create(0x314d636140, 0x7fff623b5f30, 0x11f6690, 0x313eb9ce90, 11) = 0x11f6ac0 pcap_set_snaplen(0x11f6ac0, 65535, 0x8038a0, 0x306874, 0) = 0 pcap_set_promisc(0x11f6ac0, 1, 0, 0x306874, 0) = 0 pcap_set_timeout(0x11f6ac0, 1000, 0, 0x306874, 0) = 0 pcap_activate(0x11f6ac0, 0, 0, 0x306874, 0 tcpdump hang in pcap_activate . whereas if run tcpdump in original kernel it works fine.can anybody know the reason?

1 0

Re: encrypt data
by Valdis.Kletnieks＠vt.edu July 11, 2013

July 11, 2013

On Thu, 11 Jul 2013 08:49:13 +0800, you said: Note: You're usually better off discussing things on-list. That way, everybody gets to learn something, and you have a better chance of soembody else providing information besides just me. Redirecting to list. > Thank you for your quick response but here i wanna implementation of > transparent encryption in the kernel,as user can not realized what he send > data had encrypt in kernel, and when they want get ,they will get decrypt > data. OK.. This is where it gets very tricky very fast. :) The problem is that it's *very* hard to do truly transparent but yet secure encryption. Let's work through one example in detail - cryptLUKS. cryptLUKS is designed to protect against data theft or modification of a system that's turned off (for instance, a laptop sitting in a laptop bag in a hotel or airport). To do this, the user has to enter a passphrase at disk mount time, providing the encryption key. Somebody steals the laptop, they don't have the key, and they can't access the contents of the disk. Now what happens if we configure it so tha there is no passphrase, and it automatically supplies the proper encryption key by itself? Then the bad guy who stole the laptop can just boot it, watch the system supply the encryption key, unlocking the disk, and now he can access it - you now have exactly zero actual security. (This is a trivial example of the fact that crypto systems are almost never broken by actually attacking the crypto algorithm - in almost every single case, the actual break is in how the software/hardware does key management) Note that cryptLUKS *does* have a few issues - for instance, it doesn't by itself protect against an "evil maid" attack where somebody modifies the initramfs to capture your passphrase, and there's certain classes of cold boot attacks that can also get the encryption key. So although it provides sufficient protection against "guy steals laptop to support his drug habit", it doesn't fully cover all the bases. If you have *serious* reason to think that the FBI, a rival company, a pissed-off hacker ex-lover, or somebody would want your data enough to backdoor your initramfs, then you're going to need stuff like secure boot to fully close all the holes. That's why in my first mail I asked: 1) What data you're trying to protect (how much, general type, etc) 2) Who you're trying to protect it from 3) How much non-transparency are your users willing to put up with In other words, you really need to understand the threat model, and how various cryptographic techniques address aspects of the threat. And also - it's almost always a mistake to design your own crypto system, even if your last name is Schneier or Rivest. *Everybody* can design a system they can't break themselves. Doing one that nobody else can break either is a whole different story. (Even Bruce Schneier didn't trust his own Twofish or Blowfish algorithms until they'd gotten some serious review by other top cryptography experts).

1 0

how to intercept system calls
by 于运超 July 10, 2013

July 10, 2013

Hi, I wanna encrpto some data in kernel, and these encrypto data storage on disk .i should decrypt data before system call trasmit to user mode. then what should i do ? how to get encrypt data ? I wanna intercept write function(kernel), when this function got encrypt data i will decrypt.

3 2

Use netlink to achieve user mode and kernel interaction
by 于运超 July 10, 2013

July 10, 2013

Hi, nowadays i am learning how to use netlink achieve user mode and kernel interaction. and when i finsh it ,i found kernel can not get the data from user mode. next is my code. can you tell me why? thank u user mode code: #include <stdio.h> #include <errno.h> #include <sys/types.h> #include <sys/socket.h> #include <unistd.h> #include <linux/netlink.h> #include <fcntl.h> #include <sys/stat.h> #include <stdlib.h> #include <string.h> void myerr(const char * err_string,int line) { fprintf(stderr,"line:%d ",line); perror(err_string); exit(1); } #define MAX_PAYLOAD 2046 #define NETLINK_TEST 26 /*struct req { struct nlmsghdr nlh; char buf[MAX_PAYLOAD]; }*/ int main (int argc,char * argv[]) { int fd; struct sockaddr_nl saddr,daddr; int sock_fd; struct nlmsghdr *nlhdr = NULL; struct iovec iov; struct msghdr msg; char text[MAX_PAYLOAD]; int ret; if(argc < 2){ printf("Usage:%s textfile\n",argv[0]); exit(1); } if(fd = (open(argv[1],O_RDWR|O_CREAT,S_IRWXU)) == -1){ myerr("open",__LINE__); }else{ printf("file open success!\n"); } sock_fd = socket( AF_NETLINK,SOCK_RAW,NETLINK_TEST); if( sock_fd < 0 ){ myerr("sock_fd",__LINE__); } memset(&saddr,0,sizeof(saddr)); memset(&daddr,0,sizeof(daddr)); saddr.nl_family = AF_NETLINK; saddr.nl_pid = getpid(); saddr.nl_groups = 0; if(bind(sock_fd,(struct sockaddr * )&saddr,sizeof( saddr))) { myerr("bind",__LINE__); } daddr.nl_family = AF_NETLINK; daddr.nl_pid = 0; daddr.nl_groups = 0; nlhdr = (struct nlmsghdr *)malloc(NLMSG_SPACE(MAX_PAYLOAD)); while(1){ if(read(fd,text,MAX_PAYLOAD) == -1){ myerr("read",__LINE__); } // while(read(fd,text,MAX_PAYLOAD)){ memcpy(NLMSG_DATA(nlhdr),text,strlen(text)); memset(&msg,0,sizeof(struct msghdr)); nlhdr->nlmsg_len = NLMSG_LENGTH(strlen(text)); nlhdr->nlmsg_pid = getpid(); nlhdr->nlmsg_flags = 0; iov.iov_base = (void *)nlhdr; iov.iov_len = nlhdr->nlmsg_len; msg.msg_name = (void * )&daddr; msg.msg_namelen = sizeof(daddr); msg.msg_iov = &iov; msg.msg_iovlen = 1; ret = sendmsg(sock_fd,&msg,0); if(ret == -1){ myerr("sendmsg",__LINE__); }else{ printf("usr send message to kernel\n"); } } close(sock_fd); } kernel module: #include <linux/module.h> //#include <linux/kernel.h> #include <linux/netlink.h> #include <linux/skbuff.h> #include <net/sock.h> struct sock *nl_sk = NULL; #define MAX_PAYLOAD 2046 #define NETLINK_TEST 26 void nl_data_handler(struct sk_buff *__skb) { struct sk_buff *skb; struct nlmsghdr *nlh; pid_t pid; char str[2046]; // int len = NLMSG_SPACE(MAX_PAYLOAD); int rc; printk("start reading...\n"); skb = skb_get(__skb); if(skb->len >= NLMSG_SPACE(0)) { nlh = nlmsg_hdr(skb); printk("recv:%s\n",(char *)NLMSG_DATA(nlh)); memcpy(str,NLMSG_DATA(nlh),sizeof(str)); pid = nlh->nlmsg_pid; printk("pid is %d\n",pid); kfree_skb(skb); skb = alloc_skb(NLMSG_SPACE(MAX_PAYLOAD),GFP_ATOMIC); if(!skb){ printk(KERN_ERR "allocate failed!\n"); return ; } nlh = nlmsg_put(skb,0,0,0,MAX_PAYLOAD,0); NETLINK_CB(skb).pid = 0; memcpy(NLMSG_DATA(nlh),str,sizeof(str)); printk("net_link: ready send message\n"); rc = netlink_unicast(nl_sk,skb,pid,MSG_DONTWAIT); if(rc < 0){ printk("can`t unicast skb\n"); } printk("send ok!\n"); } } static int __init net_init(void) { struct net init_net; // struct sock *nl_sk = NULL; printk("net creat start.....\n"); nl_sk = netlink_kernel_create(&init_net,NETLINK_TEST,0,nl_data_handler,NULL,THIS_MODULE); if(nl_sk == 0) { printk("create netlink faile!\n"); return -1; }else{ printk("netlink create success!\n"); } return 0; } static void __exit net_exit(void) { printk("lal\n"); sock_release(nl_sk->sk_socket); printk("netlink over!\n"); } module_init(net_init); module_exit(net_exit); MODULE_LICENSE("GPL");

1 0

Identifying kernel module from maj:min numbers
by Prabhunath G July 10, 2013

July 10, 2013

Dear All, Is there a mechanism to identify a driver module (assume that driver is loaded dynamically by insmod) by knowing a major number and minor number through sysfs. Thanks, Prabhu

2 1

makefile
by David kiarie July 10, 2013

July 10, 2013

Hi everybody, I currently started learning how to write Linux drivers.I am following Linux Device Drivers Third Edition. I have everything set right but the makefile provided seem not to work.I am familiar with makefiles but the kernel build system confuses me Here is the makefile: # If KERNELRELEASE is defined, we've been invoked from the # kernel build system and can use its language. ifneq ($(KERNELRELEASE),) obj-m := hello.o # Otherwise we were called directly from the command # line; invoke the kernel build system. else KERNELDIR ?= /lib/modules/$(shell uname -r)/build PWD := $(shell pwd) default: $(MAKE) -C $(KERNELDIR) M=$(PWD) modules endif It give the following output when I run make : make: Nothing to be done for `default'.

4 3

Rebuild a changed kernel quickly with make?
by Martin Kepplinger July 9, 2013

July 9, 2013

Hi. When I change something in the kernel and just hit make from toplevel dir, it always compiles everything again, no matter if it wasn't changed at all in the meantime. I thought make should be able to make a second build time really quick. What am I doing wrong? Is there a make target I don't know of? I don't type in any, just make. thanks, martin

5 6