August 2013 - kernelnewbies@kernelnewbies.org

Who is 'Freeing init memory ...'
by Woody Wu Aug. 15, 2013

Aug. 15, 2013

Hi, Near the end of kernel startup, there is a message 'Freeing init memory ..'. This process on embedded system takes several seconds. After searched kernel source, however, I cannot find the same literal text in all the .c files. So, I want to know, who was printing this message and how can I disable the behavior to save some boot time since I don't care losting of several hundreds kilo bytes of memory. Thanks in advance. woody -- I can't go back to yesterday - because I was a different person then

3 4

handling of many sysfs attributes using a single show function
by George White Aug. 14, 2013

Aug. 14, 2013

Hi all, I am creating a module that has a lot of configuration parameters (of _Bool type). These parameters are set to 0 or 1 using sysfs attributes registered in an attribute group of one kobject. Each attribute show just printfs the var and attrbute store just scanfs the corresponding variable of the same name. I have just one show and one store function that is shared among all attributes, in an effort to minimise repetition. My problem is that I don't know how to make this single 'show' function simple enough, so that it does not have to parse the name of each file read in order to find out which variable to work with. To make this more clear, here is my code right now for the show function of the attribute group: _Bool system_status = 1; _Bool confoptionA= 1; _Bool confoptionB = 0; ... _Bool confoptionZ = 0; static ssize_t option_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { size_t count = 0; if (strcmp(attr->attr.name, "system_status") == 0){ count = sprintf(buf, "%d\n", system_status); } else if (strcmp(attr->attr.name, "confoptionA") == 0){ count = sprintf(buf, "%lu\n", confoptionA); } else if (strcmp(attr->attr.name, "confoptionB") == 0){ count = sprintf(buf, "%d\n", confoptionB); } else if (strcmp(attr->attr.name, "confoptionC") == 0){ count = sprintf(buf, "%d\n", confoptionC); } ..... return count; } This repeats for all my 40something options. Surely this is not ideal and there has to be a better way to do it! Any ideas? Thanks, GeorgeW

1 0

Configure DMA atributes?
by Peter Senna Tschudin Aug. 14, 2013

Aug. 14, 2013

There are some PCI Ethernet drivers with code similar to: From: drivers/net/ethernet/realtek/8139cp.c /* Configure DMA attributes. */ if ((sizeof(dma_addr_t) > 4) && !pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)) && !pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) { pci_using_dac = 1; } else { pci_using_dac = 0; rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) { dev_err(&pdev->dev, "No usable DMA configuration, aborting\n"); goto err_out_res; } rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) { dev_err(&pdev->dev, "No usable consistent DMA configuration, aborting\n"); goto err_out_res; } } What is this code supposed to do? I did tests inside two KVM virtual machines using rtl8139 virtual network interface. One VM configured for x86_64 and the other configured for i686. On both VMs this code makes no visible difference at all. I can remove the code completely, or leave only parts of it, the result is always the same. The 8139cp module is loaded successfully and the network works after loading the module. Nothing is shown on dmesg. If I comment the assignment to pci_using_dac, the compiler issues warnings, but it makes no visible difference to assign 1 or 0. If I had real hardware what could I brake by playing with this code? Thanks! -- Peter

3 2

IA64 Help on a SGI Altix 4700 System
by Barclay Jameson Aug. 14, 2013

Aug. 14, 2013

I have tried to get a 3.2 and higher Kernel with support for more than 64 CPUs working on an SGI Altix 4700 with no luck. Every Kernel I have built keeps reporting the same error: Non-existent Memory Address Error. I am not sure if this is a Kernel issue or not but I will supply the output from the POD that was entered due to MCA. Any pointers or help would be greatly appreciated. Short Error output: 000 051.21^1#0a: index time stamp type component subcomponent 000 051.21^1#0a: ----- ------------------ --------- ------------ ------------ 000 051.21^1#0a: 0 0x000000c92deef702 MD_HW 051.21^1#0 Non-existent Memory Address Error 000 051.21^1#0a: 1 0x000000ce43920f08 PI_HW 051.21^1#0 RRB Time-out Error 000 051.21^1#0a: 2 0x000000ce43b16400 PROC_MCA 051.21^1#0a Bus Check More Detailed Error Output: 000 051.21^1#0a: SH2_EVENT_OCCURRED : 0x0000008180000003 000 051.21^1#0a: MD Hardware Interrupt Pending 000 051.21^1#0a: SH2_FIRST_ERROR : 0x0000000000000002 000 051.21^1#0a: MD Hardware Interrupt Pending 000 051.21^1#0a: SH2_MEM_ERROR_SUMMARY : 0x0000007800000002 000 051.21^1#0a: Non-existent Memory Address Error 000 051.21^1#0a: SH2_MEM_FIRST_ERROR : 0x0000000000000002 000 051.21^1#0a: MD_HW_INT: Non-existent Memory Address Error 000 051.21^1#0a: SH2_MISC_ERR_HDR_UPPER : 0x0000000001f00004 000 051.21^1#0a: Non-Existant Memory Address Error Header Captured 000 051.21^1#0a: Echo: 0x1f 000 051.21^1#0a: SH2_MISC_ERR_HDR_LOWER : 0x8800010000000000 000 051.21^1#0a: Source : pi chiplet, nasid 0x0 000 051.21^1#0a: Command : NCRD, Non-coherent read 000 051.21^1#0a: Read Operation 000 051.21^1#0a: SH2_MISC_ADRS_ERR_HDR_LOWER_A : 0x80000001014cf070 000 051.21^1#0a: Address <37:0>: 0x1014cf070 000 051.21^1#0a: Read Operation 000 051.21^1#0a: SH2_MD_HW_TIME_STAMP : 0x800000fa22fade06 000 051.21^1#0a: 000 051.21^1#0a: PI_HW :051.21^1#0 :RRB Time-out Error 000 051.21^1#0a: 000 051.21^1#0a: SH2_EVENT_OCCURRED : 0x0000008180000003 000 051.21^1#0a: PI Hardware Interrupt Pending 000 051.21^1#0a: SH2_FIRST_ERROR : 0x0000000000000002 000 051.21^1#0a: SH2_PI_ERROR_SUMMARY : 0x0000000000000010 000 051.21^1#0a: RRB Time-out Error 000 051.21^1#0a: SH2_PI_FIRST_ERROR : 0x0000000000000010 000 051.21^1#0a: RRB Time-out Error 000 051.21^1#0a: SH2_PI_ERROR_DETAIL_1 : 0xfe200001014cf071 000 051.21^1#0a: SH2_PI_ERROR_DETAIL_2 : 0x000000001f0801f1 000 051.21^1#0a: Address : 0x1014cf070 000 051.21^1#0a: Table Select : 0x4 000 051.21^1#0a: Command : RESERVED_FE 000 051.21^1#0a: IsReal : 0x1 000 051.21^1#0a: RRB Idx : 0x1f 000 051.21^1#0a: WRB Idx : 0x0 000 051.21^1#0a: IRB Idx : 0x0 000 051.21^1#0a: Error Code : 0x4 000 051.21^1#0a: Echo : 0x1f 000 051.21^1#0a: Source : not available 000 051.21^1#0a: Supplemental : 0x0 000 051.21^1#0a: AXB Queue : 0x0 000 051.21^1#0a: SH2_PI_HW_TIME_STAMP : 0x800000feba7afc05 000 051.21^1#0a: 000 051.21^1#0a: PROC_MCA :051.21^1#0a :Bus Check 000 051.21^1#0a: 000 051.21^1#0a: processor lid : 0x0000000000000000 000 051.21^1#0a: cpu: A nasid: 0x0 000 051.21^1#0a: processor state parameter : 0x20010000fff21120 000 051.21^1#0a: rendevous was not attempted 000 051.21^1#0a: min state is valid 000 051.21^1#0a: not continuable 000 051.21^1#0a: machine check is isolated 000 051.21^1#0a: more info available 000 051.21^1#0a: ip logged is not precise 000 051.21^1#0a: min state is not precise 000 051.21^1#0a: shared MCA 000 051.21^1#0a: bus check 000 051.21^1#0a: PAL recovery status: 000 051.21^1#0a: error was isolated and contained, continuable if sw can recover 000 051.21^1#0a: processor error map : 0x0000000001000000 000 051.21^1#0a: processor code id: 0 000 051.21^1#0a: logical thread id: 0 000 051.21^1#0a: processor bus level 1 error 000 051.21^1#0a: processor structure: bus 000 051.21^1#0a: bus check : 0x1880000000800141 000 051.21^1#0a: bus transaction size: 1 000 051.21^1#0a: external bus error 000 051.21^1#0a: transaction type: partial read 000 051.21^1#0a: bus error severity: 0 000 051.21^1#0a: bus hierarchy: 0 000 051.21^1#0a: UCE detected on incoming 000 051.21^1#0a: ia64 instruction set 000 051.21^1#0a: machine check corrected 000 051.21^1#0a: target address valid 000 051.21^1#0a: target identifier : 0x00000001014cf071

1 0

Reduce boot time console messages?
by Woody Wu Aug. 13, 2013

Aug. 13, 2013

Hi, My kernel is printing too many messages at boot time from a special device driver (j2ffs filesystem). I belive these kernel messages are level DEBUG. If don't change the kernel code, is there anyway to print only some higher level of messages? Thanks in advance. -- I can't go back to yesterday - because I was a different person then

2 4

Driver laboratory classes on university
by Grzegorz Dwornicki Aug. 13, 2013

Aug. 13, 2013

Hi I'm working on own phd thesis. It includes some kernel hacking for practical application of my research. I wish to make some progress with programming classes on my university. We have a lot of high level programing like object oriented languages but none on low level and kernel programing. I wish to know how other universities run classes on operating systems. I have named this topic about drivers because I think this will be most fun for the students. But fell free to give me any advices. I have more than one year to write conspect and get the required hardware. But first I need to learn myself all I can. I don't know where to start. Do I need to know electronic and physics good? Is there some easy to program hardware but at the same time easy to show how this is done from scratch? At first this would be an optional classes but I know there are a lot of interested students in this subjects soo this may become a part of operating systems laboratory classes. They are as you may expect - mandatory. I wish to give them a least a good start. This isn't as simple as decide what I need to know for my thesis... Links for articles, hardware on some stores or just it's prod. Id will be helpful. Thanks for help Gregory

6 7

BKL- BIG kernel lock + smp
by nidhi mittal hada Aug. 12, 2013

Aug. 12, 2013

Hi all, I have an smp linux redhat kernel running on many cpu cores. I know that one process running on one of cpu has tried for Big kernel lock . I am finding that another process on another CPU is trying for BKL. My question is :- ------------------------- Being on different processors , can they both try and acquire BKL ? Can two processes running on different CPUs, acquire BKL , at same time? Another doubt (for same infrastructure) *BKL and preemption and spinlock* 1) If i find from core for a process task struct that *lock_depth = 0 *, *does that ensure this process has acquired BKL ?* as without taking lock it remain -1 ?? 2) If i get preempt_count as non negative but *not 1*, instead a very big number 67174400 , does that ensure that *kernel preemption has been disabled by this process and its running in atomic context ?* 3) If a process has taken BKL , kernel premption is disabled from that point of time, now same process if tries to acquire another spinlock , thats held by some other process... will it cause a deadlock ? i assume, other process if gets to run on other cpu, will run and release the lock and hence will be acquired by this process.. is it right ? Thanks Nidhi

2 1

Where is the system call table in linux kernel v3.9?
by Iker Pedrosa Aug. 12, 2013

Aug. 12, 2013

Hi Guys, I am a newbie to linux kernel and I am trying to do some of the exercises/examples of the Linux Kernel Development book by Robert Love. For the moment I'm trying to create a system call (Chapter 5) but I am unable to do the first step which states the following: "Add an entry to the end of the system call table.This needs to be done for each architecture that supports the system call (which, for most calls, is all the architectures).The position of the syscall in the table, starting at zero, is its system call number. For example, the tenth entry in the list is assigned syscall number nine." My problem is that I am unable to find the file that holds the table. In the book the file that needs to be changed is the entry.S but it no longers exists in v3.9. In an example that I have found on the internet, which is done using v3.0, the file to change is syscall_table_32.S. But I've got the same problem, it doesn't exist. So anybody can help me to find the table? I know that I should be using v2.6 of the kernel but I don't know if that version will work with the distribution that I'm using. The question is also in stackoverflow so if someone wants to answer there I won't have any problem. The link to the page is the following: http://stackoverflow.com/questions/17652555/where-is-the-system-call-table-… -- Iker Pedrosa <ikerpedrosam(a)gmail.com>

6 12

assert macro to check null
by Saket Sinha Aug. 10, 2013

Aug. 10, 2013

I have developed a driver which has become unstable. Now when I debug,I find that it is crashing at http://lxr.free-electrons.com/source/kernel/cred.c?v=3.8#L431 get_cred(new); where null is struct cred *new. I think this new might be NULL. Now I plan to add an assert in our macro that deals with creds, like for asserting passed object isn't NULL. Any suggestions as to how I can do that? Regards, Saket Sinha

3 4

Understanding device trees(dtsi files??) and mastering them.
by manty kuma Aug. 9, 2013

Aug. 9, 2013

I am currently working on a platform which uses device trees. I donot have any understanding on them. What are device trees? I see they are in the form of .dtsi and they are having their own format. Also all these files are present in platform specific code(). So i can make out that these are related to hardware initialization or something similar. But to avoid any misconceptions, it would be nice if someone with prior experience on this would share the basic details about it. Also please site to good references if you know have any. Thanks, Sandeep kumar

2 1