July 2011 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

Question about interrupts
by manish honap July 29, 2011

July 29, 2011

Hello all, When interrupts are handled through interrupt gate cpu automatically clears IF flag. but to allow nesting kernel sets the IF flag in handle_irq_event - for 2.6.11 but in 2.6.39 how the interrupts are re-enabled so as to facilitate nesting of interrupts I am unable to find local_irq_enable in do_IRQ() Thanks in advance.

2 1

will kill frees the kernel resources?
by Tirtha Ghosh July 29, 2011

July 29, 2011

Hi all, need an answer of basic question if some user-level process creates a message queue and the user kills the process with "kill -9" will kernel release the memory used by message queue?? on top, is there any linux command to see the list of message queues that are in use? regards

2 1

Regarding netlink connector documentation code.
by mindentropy July 28, 2011

July 28, 2011

Hi, In the example of the netlink connector code can somebody clarify if the coding is not right? The buf here (http://lxr.linux.no/#linux+v3.0/Documentation/connector/ucon.c#L114) is 1024 which will be cn_msg. In the nlmsghdr the buf is 128(http://lxr.linux.no/#linux+v3.0/Documentation/connector/ucon.c#L65). Should the nlmsghdr be atleast bigger than the cn_msg? I understand the payload len being 0 and this being just an example with hardcoded values. Thanks.

1 0

[net] Hacking the wholism of GNU/Linux net*
by Shawn July 28, 2011

July 28, 2011

hi guys, I have written an article[1] for newbies who are willing to learn networking stuff in linux kernel. I hope this introduction-level article can help you understand the mechanism of netfilter and NIC driver. For this article, any critical comments are welcomed! [1] http://hfg-resources.googlecode.com/files/hacking_the_wholism_of_linux_net.… -- GNU powered it... GPL protect it... God blessing it... regards Shawn

1 0

Using RPMBUILD
by Rodger Swann July 28, 2011

July 28, 2011

Hi all, I have to use rpmbuild to build my kernel. I also have to apply patches and replace source code in the kernel source tree. I have figured out how to patch the files that I need patched by adding Patch blah and then AppyPatch in the spec file. I then created a sym link to the patch file in the rpmbuild SOURCES directory. Now on to replacing the code in the source tree. I have tried untaring/unzipping the kernel source , putting in my source and retarring and zipping, I have tried symlinking it, etc. But each time the build directory contains source code that is not the code that is what I inserted and it is not clear to me where it is coming from. I am fairly frustrated so if anyway has encountered this or knows how to do it I would appreciate the help. Spoon feeding is not required a I will gladly rtfm but I will need a nudge in the right direction. Oh, here is the command that I am using: rpmbuild -bb --target=`uname -m` --with baseonly --without kabichk kernel.spec Rodger

2 1

Compiling enough of the kernel to get symbols for EXTMODs, but no more
by David Wuertele July 27, 2011

July 27, 2011

I have a problem where I seem to need to compile the kernel twice, and I am looking for a way to not have to repeat any compilation steps. I have an embedded system kernel into which I'm bundling an initramfs. That initramfs contains some modules that must be compiled external to the kernel (due to my vendor's funky build system). Before I do that I must compile the kernel, so that the external module compilation can insert the kernel symbols into mymodule.mod.c. But after the module is compiled, I must install it into the initramfs and recompile the kernel. I want the kernel to contain a bundled initramfs to simplify distribution. But I would rather not compile the kernel twice. I'm wondering if there is a target in the kernel makefile that will allow me to compile just enough of the kernel such that when I compile the external module it gets the proper symbols. I tried something like this: make -C $(LINUX) prepare scripts make -C $(LINUX) M=$(PATH_TO_MY_MODULE) modules cp $(PATH_TO_MY_MODULE)/mymodule.ko $(PATH_TO_MY_ROOTFS_STAGING)/lib make -C $(LINUX) CONFIG_INITRAMFS_SOURCE=$(PATH_TO_MY_ROOTFS_STAGING) but after booting that kernel, mymodule.ko would not load because it was missing the proper symbols. When I look at the file $(PATH_TO_MY_MODULE)/mymodule.mod.c, the block assigned to section("__versions") is empty. Apparently I need to compile more than just "prepare scripts" in my first step. I find that this works: make -C $(LINUX) CONFIG_INITRAMFS_SOURCE=$(PATH_TO_MY_ROOTFS_STAGING) make -C $(LINUX) M=$(PATH_TO_MY_MODULE) modules cp $(PATH_TO_MY_MODULE)/mymodule.ko $(PATH_TO_MY_ROOTFS_STAGING)/lib make -C $(LINUX) CONFIG_INITRAMFS_SOURCE=$(PATH_TO_MY_ROOTFS_STAGING) But that has some redundancy that I want to avoid. Is there a build target that I could use for my first step that goes far enough in building the kernel, but doesn't actually do the step of marshalling the initramfs and linking it into the kernel? Thanks, Dave

1 0

Bond, round robin mis-configuration.
by Eduard Sinelnikov July 27, 2011

July 27, 2011

Hi, In the kernel 2.6.39.3 ( /drivers/net/bond/bond_main.c). In the function ‘bond_xmit_roundrobin’ The code check if the bond is active via ‘bond_is_active_slave(slave)’ Function call. Which actually checks if the slave is backup or active What is the meaning of slave being backup in round robin mode? Correct me if I wrong but in round robin every slave should send a packet, regardless of being active or backup. Thank you, Eduard

1 0

Regarding perf top utility in linux
by ankur dwivedi July 27, 2011

July 27, 2011

Hi, The perf top utility on linux displays the samples and the pcnt values for each function. I looked into the code of perf top in builtin-top.c in the linux kernel. What i have understood is perf top does not display the actual hardware performance counters value in the samples. It just maintains a software counter for each event for each function entry in the symbol table and increments the counter when that function causes a particular event. Is my understanding correct? -- Thanks Ankur

1 0

about create_mapping function
by sandeep kumar July 27, 2011

July 27, 2011

Hi, I m trying to understand how different io memory is mapped to the CPU. While analysing that i was struck at create_mapping function. i have several questions like, 1) Why map_desc.virtual is filled with physical address of the io device. 2) What is a section and why it is needed? 3) I am not understanding "alloc_init_section(pgd, addr, next, phys, type);" what it does? -- With regards, Sandeep Kumar Anantapalli,

1 0

when early_param("string", function_name) will be called?
by sandeep kumar July 26, 2011

July 26, 2011

Hi all, In kernel code at lot of places i came across "early_param("string", function_name)" Eg: early_param("vmalloc", early_vmalloc); in arch/arm/mm/mmu.c when this function will be called? and what is the significance of the string "vmalloc" there? Thank you.. -- With regards, Sandeep Kumar Anantapalli,

2 1