October 2012 - kernelnewbies@kernelnewbies.org

why the kernel need define ZONE_HIGHEM
by Fan Yang Oct. 17, 2012

Oct. 17, 2012

HI ALL: I don't knowe why the linux designer need to define the ZONE_HIGHEM, and why they defined it 896M. thanks yangfan

2 1

Network Byte order not reached reading from a sock RAW
by Pietro Paolini Oct. 17, 2012

Oct. 17, 2012

Hello, I am struggling with the byte order question on a x86_32 arch, I am doing some modifications on a program which actually works fine on a MIPS arch. I do a reading from a RAW socket in this way: /* Configure socket */ if ((fd = socket(AF_INET, SOCK_RAW, IPPROTO_IGMP)) < 0) { perror("Error on socket creation, exit"); exit(1); } .... if (setsockopt(fd, IPPROTO_IP, IP_PKTINFO, (void *)&va, sizeof(va))) { perror("Error on setsockopt, exit"); exit(1); } va = 0; .... .... struct msghdr mhdr; struct __in_pktinfo *pktinfo = NULL; ... ... nrd = recvmsg(env->mrouter_fd, &mhdr, 0); ... ip = (struct iphdr *)iov.iov_base; When I print the saddr (or daddr) of the received ip packet it is printed as host byte order instead of what I am expecting, the network byte order. I can just use the htonl() family functions for solve the problem but I would like understand if it is the normal behavior or if there is an issues on my code, or if the device driver of my NIC can influence the question. Many thanks, Pietro.

3 6

any way to see the running kernel's "vermagic" string?
by Robert P. J. Day Oct. 16, 2012

Oct. 16, 2012

perhaps i've just overlooked it but is there an easy way to see the current kernel's calculated "vermagic" string, as defined in the header file include/linux/vermagic.h thusly? ... snip ... #define VERMAGIC_STRING \ UTS_RELEASE " " \ MODULE_VERMAGIC_SMP MODULE_VERMAGIC_PREEMPT \ MODULE_VERMAGIC_MODULE_UNLOAD MODULE_VERMAGIC_MODVERSIONS \ MODULE_ARCH_VERMAGIC i'm just interested in examining the vermagic string you can see for any loadable module, as in: $ modinfo vfat filename: /lib/modules/3.6.0-oct8+/kernel/fs/fat/vfat.ko author: Gordon Chaffee description: VFAT filesystem support license: GPL srcversion: 68B6C644F1532E86ADCB2CB depends: fat intree: Y vermagic: 3.6.0-oct8+ SMP mod_unload modversions and following the load logic in kernel/module.c to see what happens at each step of a module load, and that (obviously) involves some comparisons. thanks. rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

4 6

Conntrack statistics
by Silviu Popescu Oct. 16, 2012

Oct. 16, 2012

Hello, I am currently playing with the netfilter subsystem in a 2.6.7 kernel and a 2.6.28 kernel. In the 2.6.28 kernel I can get the number of packets and bytes for a conntrack using something like: struct nf_conn_counter *cnt = nf_conn_acct_find(conntrack); However, this relies on conntrack extensions, which do not exist in 2.6.7. Does anyone have any idea how I could accomplish the same effect in 2.6.7? I mean, the counters have to be somewhere, I just don't know where or how to reach them by having the conntrack structure. All the best, Silviu Popescu

1 0

any history to class.txt's "devclass_attribute"?
by Robert P. J. Day Oct. 16, 2012

Oct. 16, 2012

currently writing an extensive tutorial for all of kobjects and ktypes and sysfs and attributes and so on, so i'm perusing the stuff under Documentation/driver-model, and i read this in "class.txt": ... snip ... Exporting Attributes ~~~~~~~~~~~~~~~~~~~~ struct devclass_attribute { struct attribute attr; ssize_t (*show)(struct device_class *, char * buf, size_t count, loff_t off); ssize_t (*store)(struct device_class *, const char * buf, size_t count, loff_t off); }; Class drivers can export attributes using the DEVCLASS_ATTR macro that works similarly to the DEVICE_ATTR macro for devices. For example, a definition like this: static DEVCLASS_ATTR(debug,0644,show_debug,store_debug); is equivalent to declaring: static devclass_attribute devclass_attr_debug; ... etc etc ... until now, i'd never heard of a "devclass_attribute", and "git log" certainly doesn't appear to explain how it came to be. anyone know what the story is behind "devclass_attribute"? methinks it can be tossed from the current Documentation. thoughts? rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

1 0

[help] kernel boot parameter "mem=? disparity
by Jun Hu Oct. 15, 2012

Oct. 15, 2012

Hi Guys: My machine has 8G memory, when I use kernel boot parameter mem=5G , it only display ��4084 M�� using ��free �Cm ��. where the ��5120-4084 = 1036M �� memory run? # /bin/uname -a Linux dom0-93 3.0.13-0.27-default #1 SMP Wed Feb 15 13:33:49 UTC 2012 (d73692b) x86_64 x86_64 x86_64 GNU/Linux dom0-93:~ # free -m total used free shared buffers cached Mem: 4084 259 3824 0 6 84 -/+ buffers/cache: 169 3915 Swap: 1916 0 1916 dom0-93:~ # cat /proc/cmdline root=/dev/disk/by-id/cciss-3600508b100104839535656314933001b-part1 resume=/dev/disk/by-id/cciss-3600508b100104839535656314933001b-part2 splash=silent crashkernel=256M-:128M vga=0x317 mem=5120M why ? details info ,please see the attachment.

2 1

(no subject)
by Niroj Pokhrel Oct. 15, 2012

Oct. 15, 2012

Hi, I'm new to linux and kernel . I'm ongoing with the linux device drivers. I've followed the the book LDD but i'm lost about how to call my driver's specific method from the user space. Eg: if have developed a character device and inserted the module then how can i make sure that when I read or write that it implement the functions via the methods i have implemented in my module. -- Niroj Pokhrel NIT Jamshedpur, B.Tech,Electronics and Communication

4 4

compelling reason to use a kset in kernel code?
by Robert P. J. Day Oct. 14, 2012

Oct. 14, 2012

it's not at all clear to me what the value is in using a kset when writing some kernel code. i can see the example program, samples/kobject/kset-example.c, but that program doesn't seem to demonstrate what the purpose is of the kset created. as another example, consider the different filesystem implementations under the fs/ directory, and grep to see who uses what to create the appropriate entry under /sys/fs/: $ grep -r "create_and_add" * 9p/v9fs.c: v9fs_kobj = kobject_create_and_add("9p", fs_kobj); btrfs/sysfs.c: btrfs_kset = kset_create_and_add("btrfs", NULL, fs_kobj); configfs/mount.c: config_kobj = kobject_create_and_add("config", kernel_kobj); debugfs/inode.c: debug_kobj = kobject_create_and_add("debug", kernel_kobj); dlm/lockspace.c: dlm_kset = kset_create_and_add("dlm", &dlm_uevent_ops, kernel_kobj); ecryptfs/main.c: ecryptfs_kobj = kobject_create_and_add("ecryptfs", fs_kobj); exofs/sys.c: exofs_kset = kset_create_and_add("exofs", NULL, fs_kobj); exofs/sys.c: EXOFS_ERR("ERROR: kset_create_and_add exofs failed\n"); ext4/super.c: ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj); fscache/cache.c: cache->kobj = kobject_create_and_add(tagname, fscache_root); fscache/main.c: fscache_root = kobject_create_and_add("fscache", kernel_kobj); fuse/inode.c: fuse_kobj = kobject_create_and_add("fuse", fs_kobj); fuse/inode.c: connections_kobj = kobject_create_and_add("connections", fuse_kobj); gfs2/sys.c: gfs2_kset = kset_create_and_add("gfs2", &gfs2_uevent_ops, fs_kobj); namespace.c: fs_kobj = kobject_create_and_add("fs", NULL); ocfs2/cluster/sys.c: o2cb_kset = kset_create_and_add("o2cb", NULL, fs_kobj); ocfs2/stackglue.c: ocfs2_kset = kset_create_and_add("ocfs2", NULL, fs_kobj); $ as you can see, some use kobject_create_and_add() while others use kset_create_and_add(), and it's not clear to me what the difference is in *why* one was chosen over the other. thoughts? i'm sure there's a simple distinction i'm just missing here. rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

1 1

new tutorial: ktypes
by Robert P. J. Day Oct. 13, 2012

Oct. 13, 2012

i just finished another online tutorial to accompany the earlier kobject one -- this one's to try to explain ktypes: http://www.crashcourse.ca/wiki/index.php/Kernel_ktypes i'm not sure i'm terrifically happy with it -- i wrote the whole thing in less than an hour and, while it's long, a lot of it is just cut and paste from kernel header and source files, so that was pretty quick, and i'm still going over it to see if i can make it better. i'm currently writing a lot of new kernel courseware and all of it is going to be publicly available, so these tutorials are sort of a "test run" to work out the bugs, and i'm extremely interested in feedback. for the most part, i want to design the new courseware around things that one can actually *see* being used in the kernel. at one end, there's the philosophy that you can learn how things work by examining, building and running simple (toy) examples. and there are some really good examples, like the ones under samples/kobject/ in the kernel source tree. at the other end, there's using the kernel source itself to demonstrate the principles, and i'm going to try to do both to see if that's the best way to explain things. so before i publicize that new ktypes tutorial on my twitter and linkedin accounts, any feedback? more later ... rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

1 0

short tutorial on kobjects, attributes and sysfs
by Robert P. J. Day Oct. 13, 2012

Oct. 13, 2012

http://www.crashcourse.ca/wiki/index.php/Kernel_kobjects enjoy. rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

2 1