December 2010 - kernelnewbies@kernelnewbies.org

Logical, Linear and Virtual Address.
by Gustavo da Silva Dec. 23, 2010

Dec. 23, 2010

Hi Everyone! Friends, could someone explain me the difference between Kernel Logical and Kernel Virtual Address? I'm reading Linux Device Drivers 3th, but I did not understand about that. I have a macro vision about Kernel Virtual Address (virtual address is used do hide the paging mechanism and physical memory, I think), and Linear is the address used in to evaluate to the physical (using page directory+page table+offset). But and the Kernel Logical Address? Why Logical Address is Virtual Address, but not all Virtual Address is a Logical Address? Regards. Atenciosamente, Gustavo da Silva

2 2

FW: About runqueues in Linux
by Schick, Stefan Dec. 23, 2010

Dec. 23, 2010

-----Original Message----- From: Schick, Stefan Sent: Thu 12/23/2010 12:52 PM To: Sowmya Sridharan Subject: RE: About runqueues in Linux -----Original Message----- From: kernelnewbies-bounces(a)kernelnewbies.org on behalf of Sowmya Sridharan Sent: Thu 12/23/2010 11:58 AM To: kernelnewbies(a)kernelnewbies.org Subject: About runqueues in Linux Greetings List, I have certain queries regarding the maintenance of runqueues in Linux. In case of an SMP system, will a same process be added to multiple runqueues? In the sample data given below, I am seeing that the process "Appstart" is added to the runqueue on CPU1, but it is not run on that CPU. Instead it is added to CPU0's runqueue, and it is run on that CPU. I have not affined the process to any CPU, and it is not multi-threaded. CPU1 data: Cpu: 1, HRT: 1290079896.698994, Pid: 6577, Pname: Appstart added_to_runqueue, current process: irc(6103) Cpu: 1, HRT: 1290079896.706994, Pid: 6103, Pname: irc added_to_runqueue,current process: swapper(0) CPU0 data: Cpu: 0, HRT: 1290079896.702994, Pid: 6577, Pname: Appstart added_to_runqueue, Current process: eventd(5370) Cpu: 0, HRT: 1290079896.702994, Pid: 14247, Pname: Conn_agent added_to_runqueue, Current process: Appstart(6577) What is the default CPU on which a process runs, if cpu affinity is not set? Is it possible that a process added to one runqueue can be moved to another runqueue? Also I have seen swapper getting scheduled frequently even if there are runnable processes. Shouldn't swapper be run, only if there are no runnable processes? Your insights will be appreciated. Regards, Sowmya =====-----=====-----===== Notice: The information contained in this e-mail message and/or attachments to it may contain confidential or privileged information. If you are not the intended recipient, any dissemination, use, review, distribution, printing or copying of the information contained in this e-mail message and/or attachments to it are strictly prohibited. If you have received this communication in error, please notify us by reply e-mail or telephone and immediately and permanently delete the message and any attachments. Thank you Hey Sowmya, one process can only be in one runqueue at the same time. What is the default CPU on which a process runs, if cpu affinity is not set? The CPU with the shortest runqueue. Is it possible that a process added to one runqueue can be moved to another runqueue? Yes, they can be pulled over, see load_balance() in sched_fair.c Also I have seen swapper getting scheduled frequently even if there are runnable processes. Shouldn't swapper be run, only if there are no runnable processes? No, because it is possible that a Process in the expired runqueue is starving. Regards Stefan

1 0

About runqueues in Linux
by Sowmya Sridharan Dec. 23, 2010

Dec. 23, 2010

Greetings List, I have certain queries regarding the maintenance of runqueues in Linux. In case of an SMP system, will a same process be added to multiple runqueues? In the sample data given below, I am seeing that the process "Appstart" is added to the runqueue on CPU1, but it is not run on that CPU. Instead it is added to CPU0's runqueue, and it is run on that CPU. I have not affined the process to any CPU, and it is not multi-threaded. CPU1 data: Cpu: 1, HRT: 1290079896.698994, Pid: 6577, Pname: Appstart added_to_runqueue, current process: irc(6103) Cpu: 1, HRT: 1290079896.706994, Pid: 6103, Pname: irc added_to_runqueue,current process: swapper(0) CPU0 data: Cpu: 0, HRT: 1290079896.702994, Pid: 6577, Pname: Appstart added_to_runqueue, Current process: eventd(5370) Cpu: 0, HRT: 1290079896.702994, Pid: 14247, Pname: Conn_agent added_to_runqueue, Current process: Appstart(6577) What is the default CPU on which a process runs, if cpu affinity is not set? Is it possible that a process added to one runqueue can be moved to another runqueue? Also I have seen swapper getting scheduled frequently even if there are runnable processes. Shouldn't swapper be run, only if there are no runnable processes? Your insights will be appreciated. Regards, Sowmya =====-----=====-----===== Notice: The information contained in this e-mail message and/or attachments to it may contain confidential or privileged information. If you are not the intended recipient, any dissemination, use, review, distribution, printing or copying of the information contained in this e-mail message and/or attachments to it are strictly prohibited. If you have received this communication in error, please notify us by reply e-mail or telephone and immediately and permanently delete the message and any attachments. Thank you

1 0

QUestion on C language
by bhuvan gupta Dec. 23, 2010

Dec. 23, 2010

Respected sir, I am Bhuvan Gupta, a B.Tech student.(got your id from yours " USB LED driver - 1.1 ") ) With due respect, I humbly request you to please throw some light on to below written question. i will highly appreciate your efforts. [1] when we use C language,we use diferent library to do different thing.As also in your case you use module.h,init.h etc but how to write this library , in what language they are written in as standard ANSI C has round about 32 keywords ? what should i read and learn to the related topic . please suggest something (I do have knowledge of Assembly). please help,this question making me beacen it will be honour if you reply. thank you Bhuvan Gupta India

5 4

struct sock change in kernel 2.6
by lijin liu Dec. 23, 2010

Dec. 23, 2010

Hello everyone! I am trying to implement a simple tcp server in the kernel. I read the ktcpvs's source code, but it works under kernel 2.4. The struct sock changed in kernel 2.6, I have two questions about the struct: 1. Is sk_wq field in 2.6 equals to sk_sleep field in 2.4? 2. How can I access the tcp accept queue in kernel 2.6? (In kernel 2.4, we could use tp_pinfo.af_tcp.accept_queue, but it was removed in 2.6) PS: Should sk_receive_queue in the struct sock be used as tp_pinfo.af_tcp.accept_queue? Thanks a lot!

3 4

Re: Exact leak location in KMemleak Output
by Sengottuvelan S Dec. 23, 2010

Dec. 23, 2010

Sankar and All, I followed blow investigation on detecting kernel memory leak. I got different o/p which is different from yours shown. ie, 4 and 5 lines are addresses in the backtrace and not function names. http://psankar.blogspot.com/2010/11/detecting-memory-leaks-in-kernel.html comm "insmod", pid 1559, jiffies 4294961804 (age 229.610s) hex dump (first 32 bytes): fa 3e 00 00 00 00 00 00 92 5a 00 00 00 00 00 00 .>.......Z...... 0a 00 00 00 12 00 00 00 3c 38 00 00 00 00 00 00 ........<8...... backtrace: [<ffffffff8156eb1d>] kmemleak_alloc+0x2d/0x60 [<ffffffff81152469>] __vmalloc_node+0xc9/0xe0 [<ffffffff8115280c>] vmalloc+0x2c/0x30 [<ffffffffa0099047>] 0xffffffffa0099047 [<ffffffffa009906c>] 0xffffffffa009906c [<ffffffff8100204c>] do_one_initcall+0x3c/0x1a0 [<ffffffff810cea1b>] sys_init_module+0xbb/0x200 [<ffffffff81041072>] system_call_fastpath+0x16/0x1b [<ffffffffffffffff>] 0xffffffffffffffff On Wed, Oct 27, 2010 at 12:24 PM, Sankar P <sankar.curiosity(a)gmail.com>wrote: > Hi Dave, Rajat, Mulyadi and everyone, > > > Thanks for your pointers. > > Now I removed the "static __init" from the hello_init function and I > got an output of: > > unreferenced object 0xf9042000 (size 512): > comm "insmod", pid 12068, jiffies 13995923 (age 51.096s) > hex dump (first 32 bytes): > 6f 64 75 6c 65 00 00 00 00 00 00 2e 73 79 6d 74 odule.......symt > 61 62 00 2e 73 74 72 74 61 62 00 2e 73 68 73 74 ab..strtab..shst > backtrace: > [<c10b0001>] create_object+0x114/0x1db > [<c148b4d0>] kmemleak_alloc+0x21/0x3f > [<c10a43e9>] __vmalloc_node+0x83/0x90 > [<c10a44b9>] vmalloc+0x1c/0x1e > [<f9036021>] init_module+0x21/0x2f [hello_kernel] > [<c1001226>] do_one_initcall+0x71/0x113 > [<c1056c48>] sys_init_module+0x1241/0x1430 > [<c100284c>] sysenter_do_call+0x12/0x22 > [<ffffffff>] 0xffffffff > > > And, if I create a new function named myfunc and do my allocations > inside that function (instead of the init function), even though I > have a "static __init" text in the init_function I get an output: > > unreferenced object 0xf9061000 (size 512): > comm "insmod", pid 12750, jiffies 14401507 (age 110.217s) > hex dump (first 32 bytes): > 1c 0f 00 00 01 12 00 00 2a 0f 00 00 01 12 00 00 ........*....... > 38 0f 00 00 01 12 00 00 bc 0f 00 00 01 12 00 00 8............... > backtrace: > [<c10b0001>] create_object+0x114/0x1db > [<c148b4d0>] kmemleak_alloc+0x21/0x3f > [<c10a43e9>] __vmalloc_node+0x83/0x90 > [<c10a44b9>] vmalloc+0x1c/0x1e > [<f9055021>] myfunc+0x21/0x23 [hello_kernel] > [<f9058012>] 0xf9058012 > [<c1001226>] do_one_initcall+0x71/0x113 > [<c1056c48>] sys_init_module+0x1241/0x1430 > [<c100284c>] sysenter_do_call+0x12/0x22 > [<ffffffff>] 0xffffffff > > As we can see from the above output, it shows both the function name > as well as the module name where the leak is happening. Also the > address of the statement. So, my problem seems to be solved. Thank you > all for your pointers. > > Sorry for the top post and long contents. > > On Thu, Oct 28, 2010 at 12:00 AM, Rajat Sharma <fs.rajat(a)gmail.com> wrote: > > try removing static __init from function definition > > On Wed, Oct 27, 2010 at 10:40 PM, Sankar P <sankar.curiosity(a)gmail.com> > > wrote: > >> > >> Hi, > >> > >> I have an out of tree kernel module with the source: > >> > >> static __init int hello_init(void) > >> { > >> char *ptr; > >> ptr = vmalloc(512); > >> ptr = vmalloc(512); > >> ptr = vmalloc(512); > >> > >> printk(KERN_ALERT "Hello World"); > >> > >> return 0; > >> } > >> > >> Now I compile this file (hello.c) using my Makefile as follows: > >> > >> EXTRA_CFLAGS=-g > >> obj-m := hello-kernel.o > >> hello-kernel-objs := hello.o > >> > >> > >> Now if I insmod my hello-kernel.ko file and then do a `cat > >> /sys/kernel/debug/kmemleak` I get an output: > >> > >> unreferenced object 0xf8211000 (size 512): > >> comm "insmod", pid 9602, jiffies 3666022 (age 1553.469s) > >> hex dump (first 32 bytes): > >> 01 12 00 00 f2 10 00 00 01 12 00 00 fe 10 00 00 ................ > >> 01 12 00 00 0c 11 00 00 01 12 00 00 1a 11 00 00 ................ > >> backtrace: > >> [<c10b0001>] create_object+0x114/0x1db > >> [<c148b4d0>] kmemleak_alloc+0x21/0x3f > >> [<c10a43e9>] __vmalloc_node+0x83/0x90 > >> [<c10a44b9>] vmalloc+0x1c/0x1e > >> [<f8208021>] 0xf8208021 > >> [<c1001226>] do_one_initcall+0x71/0x113 > >> [<c1056c48>] sys_init_module+0x1241/0x1430 > >> [<c100284c>] sysenter_do_call+0x12/0x22 > >> [<ffffffff>] 0xffffffff > >> > >> > >> I understand that the kmemleak has correctly detected the memory leak. > >> But why is the leak showing in insmod , instead of the actual function > >> name "hello_init". Is there a way we can get the exact location inside > >> the hello.c file printed instead of the generic "insmod" location ? > >> > >> If I do a `gdb hello-kernel.ko`, the symbols are properly resolved, so > >> I am sure the .ko file has symbols. > >> > >> Any help will be appreciated. Thank you. > >> > >> -- > >> Sankar P > >> http://psankar.blogspot.com > >> > >> -- > >> To unsubscribe from this list: send an email with > >> "unsubscribe kernelnewbies" to ecartis(a)nl.linux.org > >> Please read the FAQ at http://kernelnewbies.org/FAQ > >> > > > > > > > > -- > Sankar P > http://psankar.blogspot.com > > -- > To unsubscribe from this list: send an email with > "unsubscribe kernelnewbies" to ecartis(a)nl.linux.org > Please read the FAQ at http://kernelnewbies.org/FAQ > > -- Regards, S. Sengottuvelan.

2 1

custom kernel install - query
by Sengottuvelan S Dec. 22, 2010

Dec. 22, 2010

Hi all I am trying to install new custom kernel in my ubundu. I am able to see "vmlinuz-2.6.x.x-custom-velan" and System.map-2.6.x.x-custom-velan in my /boot after #make-kpkg --initrd --append-to-version=-custom_velan kernel_image kernel_headers i don;t see initrd for this new custom kernel in /boot and also no entry is created in /boot/grub/grub.cfg. I also don't see any new file "menu.lst" in /boot/grub directory. I am using generic kernel as below. root@ubuntu:/usr/src/linux# uname -a Linux ubuntu 2.6.32-24-generic #39-Ubuntu SMP Wed Jul 28 05:14:15 UTC 2010 x86_64 GNU/Linux Can you throw some light on this. -- Regards, S. Sengottuvelan.

8 12

Need help EHCI host driver on Linux 2.6.29
by CL Lee Dec. 22, 2010

Dec. 22, 2010

Dear Experts, I am developing a EHCI host driver for Linux kernel 2.6.29 on a proprietary SoC. But for weeks now, I am troubleshooting a time out error from khubd. I believe it's related to the client device not being able to update the qtd struct which has been set at some Dma addresses, say 0x87xxxxxx. Is this possible? The main issue is the register address for chip lies in 0x5a800000 range, so is the on-chip memory. If I set the qtd structs to on chip memory(internal sram), kernel paging error will occur. So I am at a loss of what I can. Please kindly advise. Best regards, CL

4 3

Re: Recursive function in Kernel space
by Mulyadi Santosa Dec. 22, 2010

Dec. 22, 2010

On Wed, Dec 22, 2010 at 13:02, Darshan Ghumare <darshan.ghumare(a)gmail.com> wrote: > > Sir, > I have following question, > Why recursive function is not recommended in Kernel-Space? because kernel stack is limited...about 8K...or 4K if you choose to have such amount... -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com

1 0

Accessing PCI Memory Mapped Registers in Linux Kernel
by YOUNGWHAN SONG Dec. 22, 2010

Dec. 22, 2010

Hi, A few days ago, I posted about accessing PCI memory mapped registers. (FYI, the platform I use is 32bit architecture and Linux kernel compiled in 32 bit intel based architecture.) Now, I found that there is a problem to access 16b aligned registers. For example, ------------- struct foo { unsigned int TEST1; unsigned short int a1[256]; } void * regsva; struct foo *test_foo; if ((regsva = ioremap_nocache(dev->base_addr, (16 * 1024))) == NULL) { printk("ioremap() failed\n"); goto fail; } test_foo = (struct foo*)regsva; printk(KERN_DBG "value TEST1: 0x%lx\n", readl(test_foo->TEST1)); printk(KERN_DBG "value 0: 0x%x\n", readw(&test_foo->a1[0]); printk(KERN_DBG "value 1: 0x%x\n", readw(&test_foo->a1[1]); ------------- The above code in PC (intel based 32bit architecture) with the same h/w module shows that value TEST1: 0x8080 value 0: 0x9090 value 1: 0xa0a0 So, that's valid value. (I have datasheet to check if the value is correct or not.) But, when the code is run on our platform (32bit) with the same h/w module, it shows that value TEST1: 0x8080 value 0: 0xFFFF value 1: 0xFFFF Now, if I change the printk as printk(KERN_DBG "value 0: 0x%x\n", readl(&test_foo->a1[0]); printk(KERN_DBG "value 1: 0x%x\n", readl(&test_foo->a1[1]); Then, even PC shows 0xFFFF like this: value 0: 0xFFFFFFFF value 1: 0xFFFFFFFF So, if the read bit is not aligned correctly, it looks it cannot read correctly. OK, then, How can our platform (32 bit) read correctly 16b aligned PCI memory mapped register? (As long as the register is 32 bit, there is no problem.) At least, I would like to understand why the 16 bit registers and 32 bit registers are responding differently. Thanks, Daniel

4 4