July 2012 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

What is PageSlab?
by Sarbojit Ganguly July 17, 2012

July 17, 2012

Hello Community, I was browsing the tree and in slub.c/ function kfree(), I found "PageSlab" I could not find its definition anywhere. Also, googling about it did not yield much help except for pointers to various patches _using_ this. Can anyone explain what it is and where it is defined (if at all it has a definition). -- Regards, Sarbojit

3 3

lookup operation query.
by rajesh singarapu July 17, 2012

July 17, 2012

Hi All, I have couple of questions in filesystems area. 1) I want to understand how lookup operation (of inode operation) find if current dirent is actually a directory or mount point? 2) if a write happening in mid of a file (and mid of a page frame once pulled in to memory) and write at that location lead to split of that page, but doesn't completely fill the second page. in this case, clearly there is some space unfilled in second page. (I read this is going to be filled with '0's) but how read/write rotines of filesystem exactly calculates the offsets in this case correctly. These can be simple questions. but I want to know how these are handled, in say ext3 or any filesystem. thanks Rajesh

1 0

How to measure time spent in context switch
by Manavendra Nath Manav July 17, 2012

July 17, 2012

How to measure time spent in context switch in Linux i.e. the time the application stays suspended while the kernel is doing something else. I don't want to use any profiling tools. What approach should I take to write my own C code, which can also run as kernel module if required, to measure time spent accurately. Will it also require to modify the scheduler code? Does /proc gives any statistics in this regard? Ref: http://stackoverflow.com/questions/2368384/write-a-c-program-to-measure-tim… -- Manavendra Nath Manav

3 4

How to calculate the TCP checksum
by Adel Kodmani July 17, 2012

July 17, 2012

I am writing a kernel module that actually modifies the TCP header(both at sending and receiving) and I need to re-calculate the header because of that. I am capturing the packets using netfilter, so I have access to the whole SKB, I am using Linux 2.6.35-22 Looking into the source code of the TCP/IP, I found this function void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb); The comment above it says this routing computes the IPv4 TCP checksum Trying to use this function, I called it passing to it skb->sk and skb With no success so far. So my question is, how do I correctly calculate the TCP checksum? I know there's a psudoheader that must be created, and added to the calculation but there must be a function in the kernel I can call that already does that. Thanks in advance, Adel

3 3

Linux kernel report 2012 from ELC
by Peter Teoh July 16, 2012

July 16, 2012

http://elinux.org/images/a/a0/Lf_elc12_corbet.pdf -- Regards, Peter Teoh

2 1

[RFC]confusion about syscall
by 王哲 July 15, 2012

July 15, 2012

Hi all: i write two simple program to invoke syscall getpid. the first program (getpid1.c) is as followed: #include <stdio.h> #include <unistd.h> int main(void) { unsigned long value = 0; __asm__("int $0x80":"=a"(value):"0"(20)); return 0; } and use objdump -d a.out disassembling it : ... 80483c2: b8 14 00 00 00 mov $0x14,%eax 80483c7: 89 c3 mov %eax,%ebx 80483c9: 89 d8 mov %ebx,%eax 80483cb: cd 80 int $0x80 ... and the second program: #include <stdio.h> #include <unistd.h> int main(void) { unsigned long value = 0; value = getpid(); return 0; } and disassembling it:( objdump -d a.out) ... 08048300 <getpid@plt>: 8048300: ff 25 00 a0 04 08 jmp *0x804a000 8048306: 68 00 00 00 00 push $0x0 804830b: e9 e0 ff ff ff jmp 80482f0 <_init+0x3c> ... 080483e4 <main>: 80483e4: 55 push %ebp 80483e5: 89 e5 mov %esp,%ebp 80483e7: 83 e4 f0 and $0xfffffff0,%esp 80483ea: 83 ec 10 sub $0x10,%esp 80483ed: c7 44 24 0c 00 00 00 movl $0x0,0xc(%esp) 80483f4: 00 80483f5: e8 06 ff ff ff call 8048300 <getpid@plt> 80483fa: 89 44 24 0c mov %eax,0xc(%esp) 80483fe: b8 00 00 00 00 mov $0x0,%eax ... question: why i can't find the "movl $0x14,%eax" "int 0x80" in the second program after disassembling? any advice will be help! thanks in advance! wanny

3 8

Trying to find all the kernel modules needed for my machine using shell script
by Aft nix July 13, 2012

July 13, 2012

Hi, I'm developing kernel modules right now, and the build times are starting to get under my skin. As a side effect i'm taking way too many "coffee" breaks during builds. So i was looking for a way to build only the stuffs i need for my platform. Chapter 7 and 8 of "linux kernel in a nutshell" gave a good detail of how to do that by hand. Its a good read : http://www.kroah.com/lkn/ But Although i understand the stuffs, this is still a lot of tweaks to make that work. 2.6.32 and later kernels added a new target "make localmodconfig". which scans through "lsmod" and change the .config appropriately. So i thought i found my "automation". But this perl script has some problem too. This thread describes the problems : https://bbs.archlinux.org/viewtopic.php?pid=845113 There was also a proposed solution which apparently worked for others , is to run the script directly instead of using make's target. Although for me, make localmodconfig does not work at all. its because of the following : $make clean $make mrproper $cp /boo/config-'uname -r' .config $make localmodconfig and it haults with vboxguest config not found!! nf_defrag_ipv6 config not found!! vboxsf config not found!! vboxvideo config not found!! The thing is my kernel development environment is inside virtualbox. These vbox modules were installed when i chose to install "virtualbox guest addtion". And the netfilter module might be a Distribution specific module(Lot of netfilter modules are not part of the mainline kernel, so its not a shock to me), which is not included in mainline kernel. Now the workaround this obviously unloading these module and trying again. But I'm thinking if there is patch for the "streamline_config.pl" which will enable the user to exclude certain modules if s/he wants. Problem is i have "zero" knowledge about perl and i like it that way. So my problems in "nutshell" 1) Patching streamline_config.pl so i can give a list of module name as argument which it will exclude from processing the config file. 2) I've tried to run the script by myself. But running it directly fails with a error. [root@kernel-host-rh6 linux-2.6]# sh scripts/kconfig/streamline_config.pl scripts/kconfig/streamline_config.pl: line 45: my: command not found scripts/kconfig/streamline_config.pl: line 47: my: command not found scripts/kconfig/streamline_config.pl: line 48: chomp: command not found scripts/kconfig/streamline_config.pl: line 50: syntax error near unexpected token `(' scripts/kconfig/streamline_config.pl: line 50: `my @searchconfigs = (' Again my "zero" knowledge about perl does not helping my case. Additional info: I have faced the same thing in 2.6.32, 2.6.33, 2.6.35, 3.4, 3.4.rc1 .....(I constantly switch version for various reasons, everywhere the problem is same) You can access the script from kernel.org gitweb: http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=blob;f=… Its from 3.4.0. `git log --stat scripts/kconfig/streamline_config.pl` shows some activity recently on this file. But my problem persists. 3) Is there any way i can get the "Centos specific" patches in "plain patch" format? That means i dont wanna do things in src-rpm way. If the patches are available in normal patch format i can keep everything tidy in my kernel git tree. -- -aft

2 4

Re: [CentOS] Trying to find all the kernel modules needed for my machine using shell script
by Aft nix July 13, 2012

July 13, 2012

On Fri, Jul 13, 2012 at 7:34 PM, Akemi Yagi <amyagi(a)gmail.com> wrote: > On Fri, Jul 13, 2012 at 5:29 AM, Aft nix <aftnix(a)gmail.com> wrote: >> Hi, >> >> I'm developing kernel modules right now, and the build times are >> starting to get under my skin. As a side effect i'm taking way too >> many "coffee" breaks during builds. > (snip) >> 3) Is there any way i can get the "Centos specific" patches in "plain >> patch" format? That means i dont wanna do things in src-rpm way. >> If the patches are available in normal patch format i can keep >> everything tidy in my kernel git tree. > > The CentOS way of "Build Your Own Kernel Modules" is described in this > CentOS wiki article: > > http://wiki.centos.org/HowTos/BuildingKernelModules > > I suggest you try method 1 first to see if you can build your module. > Once confirmed, you may want to proceed to method 3 to build a > kABI-compatible module. > > You can ask further questions about kmods either here, or better yet, > on the ELRepo's mailing list. ELRepo ( http://elrepo.org ) builds a > number of kernel modules for RHEL/CentOS/SL. > > Akemi > So bottom line is, i can't get the patches other than src-rpms? I better learn this thing, i guess. cheers. _______________________________________________ > CentOS mailing list > CentOS(a)centos.org > http://lists.centos.org/mailman/listinfo/centos -- -aft

1 0

[TCP]: question about counter in tcp_mark_head_lost
by 李易 July 12, 2012

July 12, 2012

hi all, In tcp_mark_head_lost() fucntion, when sack enabled and fack disabled, why does the packet counter variable "cnt" increase only when current skb is marked as TCPCB_SACKED_ACKED? code: if (tcp_is_fack(tp) || tcp_is_reno(tp) || (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) cnt += tcp_skb_pcount(skb); Is this right ? I have a question when retransmition queue looks as below: sacked=0 --> sacked=SACK_ACKED --> sacked=0 --> sacked=SACK_ACKED then, when we call tcp_mark_head_lost(sk, 1, 0) this will mark the two skbs, which are previously marked 0 , as TCPCB_LOST. So, this function marks two skbs as lost, although we pass parameter "packets" as "1". I do not know why this function has this semantic , which is not consistent with its declaration. Can anyone explanation this, thanks in advance.

1 0

Changing the page attributes
by Arun KS July 11, 2012

July 11, 2012

Hello Experts, I have a requirement to change __log_buf memory(kernel printk buffer) as non cacheable memory. How can I change the page attributes for changing cache policy to non cacheable? Any pointers will be helpful. Thanks, Arun

3 5