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

Problem with rootfs mounting
by stl March 22, 2011

March 22, 2011

Hello all, I am porting linux-2.6.37 to a new architecture, and I am facing the well-known problem: VFS: Cannot open root device "(null)" or unknown-block(0,0) When I was working on uClinux, I have seen that the vmlinux image was concatenated with the romfs image. The kernel was aware of the address of the magic number -rom1fs- thanks to the configuration CONFIG_MTD_UCLINUX_ADDRESS= 0x . . . . in the .config file. But, I have not found how I can do this in Linux.. I have chosen the filesystem ROMFS in the .config file. Where and how the romfs tree is created and included to the kernel image? This question concerns also all the necessary user applications like "init" and "sh". Should I use a build system tool like Buildroot or OpenEmbedded? Any help will be very useful. Thanks in advance.

2 1

Problem with rootfs mounting
by stl March 21, 2011

March 21, 2011

Hello all, I am porting linux-2.6.37 to a new architecture, and I am facing the well-known problem: VFS: Cannot open root device "(null)" or unknown-block(0,0) When I was working on uClinux, the vmlinux image was concatenated with the romfs image. The kernel was aware of the address of the magic number -rom1fs- thanks the configuration CONFIG_MTD_UCLINUX_ADDRESS= . . . in the .config file. But, I have not found how I can do thit in Linux.. Where and how the romfs tree is created and included to the kernel image? This question concerns also all the necessary user applications like "init" and "sh". Any help will be very useful. Thanks in advance.

1 0

Common signal handler system call
by mohit verma March 21, 2011

March 21, 2011

Hi folks, AFAIK there is not particular system call in Linux systems that can register a number of signals (signal numbers) at once and provide a common signal handler for all of the assembled signals. I know doing this may be hardly required in practice. But i think this type of system call should be in Linux systems which can serve handler functionality on some of signals. Please tell me the flaws proposing something like this. -- ........................ *MOHIT VERMA*

4 11

PCI Configuration Space
by Joy March 21, 2011

March 21, 2011

Hi, Please, forgive me about this silly question, but for this beginning learner, if somebody helps me, it would be appreciated. I've googled before, of course, but it's not easy to figure out. The question is simple. If I am able to read PCI device's configuration space, where is it? Is it in Main Memory (in general, DDR3) mapped? Or Is it in the device's own memory? Joy

2 1

how can I get page fault addressed with perf
by zhao bao March 21, 2011

March 21, 2011

Hi, I want to trace page faults in my program. The result I wanted like this: 1st page fault at addrA. 2nd page fault at addrB. 3rd page fault at addrC ... Perf seems providing this function, but I can't get what I want. I use Perf on Fedora 14(kernel 2.6.35). The following is what I have done in my machine. Thanks for help in advance. [root@localhost lab]# perf record -R -c 1 -d -e faults ./hello hello,world [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.010 MB perf.data (~418 samples) ] [root@localhost lab]# perf report -n # Events: 91 cycles # # Overhead Samples Command Shared Object Symbol # ........ .......... ....... ................. ...... # 10.99% 10 hello ld-2.12.90.so [.] .L198 6.59% 6 hello ld-2.12.90.so [.] _dl_load_cache_lookup 6.59% 6 hello ld-2.12.90.so [.] _dl_lookup_symbol_x 4.40% 4 hello ld-2.12.90.so [.] dl_main 3.30% 3 hello ld-2.12.90.so [.] _dl_start 3.30% 3 hello ld-2.12.90.so [.] _dl_cache_libcmp 3.30% 3 hello ld-2.12.90.so [.] do_lookup_x 2.20% 2 hello libc-2.12.90.so [.] _init 2.20% 2 hello libc-2.12.90.so [.] _IO_file_overflow_internal 2.20% 2 hello ld-2.12.90.so [.] _dl_map_object_from_fd 2.20% 2 hello ld-2.12.90.so [.] _dl_init_paths 2.20% 2 hello ld-2.12.90.so [.] _dl_setup_hash 2.20% 2 hello ld-2.12.90.so [.] .L302 .......................

2 1

Problem compiling BCM Wireless driver on 2.6.37 (lib80211 symbol issue)
by Shreyansh Jain March 21, 2011

March 21, 2011

Dear List, I recently compiled and booted up the 2.6.37 kernel on my Ubuntu (10.04), from 2.6.36.1. I have been trying to compile the Broadcom Wireless driver on 2.6.37, after making some changes in the source code. Just to add, this driver has been working fine on 2.6.36.1 which my machine ran earlier. ---8<--- $> uname -a Linux Tiger-L 2.6.37 #2 SMP Thu Mar 3 22:45:04 IST 2011 i686 GNU/Linux $> lspci | grep Broadcom 03:00.0 Network controller: Broadcom Corporation BCM4312 802.11b/g (rev 01) ---8<--- I have been trying to compile the Broadcom driver hybrid-portsrc-x86_32-v5.60.48.36.tar.gz, which I got from Broadcom website. [ It is quite possible that newer version is available on the same website, but I intend to get this version up, if possible.] This driver requires the 80211 Library kernel module (lib80211.ko) to be loaded as it accesses the security (crypto) algo functions which are available in this library. Thus: ---8<--- $> modprobe lib80211 $> lsmod | grep lib80211 lib80211 5058 0 ---8<--- Then, I made some changes in the code (as the code base was not getting directly compiled on this new kernel). I have attached the patch with this email in case that is required. Furtheron, on compiling the module, following is what I get: ---8<--- (in the driver directory) $> make clean KBUILD_NOPEDANTIC=1 make -C /lib/modules/`uname -r`/build M=`pwd` clean make[1]: Entering directory `/opt/kernels/linux-2.6.37' CLEAN /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/.tmp_versions CLEAN /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/Module.symvers make[1]: Leaving directory `/opt/kernels/linux-2.6.37' $> make KBUILD_NOPEDANTIC=1 make -C /lib/modules/`uname -r`/build M=`pwd` make[1]: Entering directory `/opt/kernels/linux-2.6.37' LD /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/built-in.o CC [M] /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/src/shared/linux_osl.o CC [M] /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/src/wl/sys/wl_linux.o CC [M] /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/src/wl/sys/wl_iw.o LD [M] /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/wl.o Building modules, stage 2. MODPOST 1 modules WARNING: "lib80211_get_crypto_ops" [/home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/wl.ko] undefined! CC /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/wl.mod.o LD [M] /home/shrey/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37/wl.ko make[1]: Leaving directory `/opt/kernels/linux-2.6.37' ---8<--- On searching for the lib80211_get_crypto_ops symbol: ---8<--- $> cat /proc/kallsyms | grep lib80211_get_crypto_ops f826273c r __ksymtab_lib80211_get_crypto_ops [lib80211] f82627b4 r __kstrtab_lib80211_get_crypto_ops [lib80211] f826278c r __kcrctab_lib80211_get_crypto_ops [lib80211] f8262020 T lib80211_get_crypto_ops [lib80211] <<<##### $> $> zcat /proc/config.gz | grep -i lib80211 CONFIG_LIB80211=m <<<##### CONFIG_LIB80211_CRYPT_WEP=m CONFIG_LIB80211_CRYPT_CCMP=m CONFIG_LIB80211_CRYPT_TKIP=m # CONFIG_LIB80211_DEBUG is not set $> ---8<--- This is what I get when I inserted the kernel: ---8<--- $> sudo insmod ./wl.ko [sudo] password for shrey: insmod: error inserting './wl.ko': -1 Invalid parameters shrey@Tiger-L:~/work/Things_for_Wifi_on_Linux/wifi_driver_broadcom_2.6.37$ dmesg | tail | grep wl [ 1025.190669] wl: no symbol version for lib80211_get_crypto_ops [ 1025.190674] wl: Unknown symbol lib80211_get_crypto_ops (err -22) $> ---8<--- As far as I understand the symbol for lib80211_get_crypto_ops function is available is exported: ---8<--- 251 return alg->ops; 252 } 253 EXPORT_SYMBOL(lib80211_get_crypto_ops); 254 255 static void *lib80211_crypt_null_init(int keyidx) 256 { 257 return (void *)1; "net/wireless/lib80211.c" 284 lines --89%-- ---8<--- and hence, it should be available to my module on request. I am not sure what is happening or where should I look next. Can anyone on the list point me out as to what I could be missing. Also, initially the compilation was giving issues about MODULE_LICENSE not being GPL, which I explicitly did by adding in the code (that is not part of the patch I have attached). I don't think that would affect this situation as the symbol lib80211_get_crypto_ops is not restricted to GPL modules only. Any help would be deeply appreciated. TIA, Shreyansh

2 4

User space context switch
by mohit verma March 21, 2011

March 21, 2011

hi all, is it possible to write a user space code to calculate the context switch time of a process ? I mean , how can the user space code know about the working of the scheduler? Is there any system call API to interact with scheduler or something other to interact with , regarding this problem? -- ........................ *MOHIT VERMA*

3 3

gsoc mentor request
by nijil yes March 19, 2011

March 19, 2011

Hi, I would like to apply for gsoc this year.Am not an expert in programming but am ready to learn and can read code.I was suggested to look into checkpoint/restart patch bystefanha from QEMU devel. I would like to know if anyone is ready to mentor this project.If yes please respond. I would like to start preparing as soon as possible as I would have to learn a few things . Or another suggestion that I could possible take for a gsoc project.It would be great if it is related to process migration or some sort of the likes. regards, yonij.

1 0

slab: cache sizes for kmalloc
by Maksym Planeta March 18, 2011

March 18, 2011

There are predefined cache sizes in <linux/kmalloc_sizes.h>. But I don't understand why exactly these sizes were chosen. I've wrote a hook were I've counted witch object sizes are the most popular. They were objects of sizes 8 and 16 bytes, but the smallest available cache has size 32 bytes. So in this cache fragmentation is about 40%. There is big fragmentation in 512 and 1024-byte caches too -- 25 and 35 percent correspondingly. Also there are empty caches, all DMA caches on my system are empty. In total there is wasting of memory. That's why, I think that caches for kmalloc can be created dynamically. For example, if I have 32-byte cache, but it's fragmentation exceeds level of 20% can be created new cache with smaller size and new objects, that fit this new size, should be allocated there. But if there are too little objects in the cache, new allocating to it can be stopped and with the lapse of time when it become empty it could be destroyed. The aim is to reduce memory waste and make fragmentation nearly equal. So I would like to know is there any sense in such cache management. If yes, I'll work on this. -- Thanks, Maksym Planeta

2 3

inconsistent lock state on call to kmem_cache_zalloc()
by Arvid Brodin March 17, 2011

March 17, 2011

On a call to kmem_cache_zalloc(), I get this: ================================= [ INFO: inconsistent lock state ] 2.6.37 #116 --------------------------------- inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage. khubd/133 [HC0[0]:SC0[0]:HE1:SE1] takes: (&(&priv->lock)->rlock){?.-...}, at: [<9017a508>] isp1760_urb_enqueue+0x24c/0x3e4 {IN-HARDIRQ-W} state was registered at: [<90043476>] lock_acquire+0x3a/0x48 [<90236210>] _raw_spin_lock_irqsave+0x28/0x50 [<9017a7ac>] isp1760_irq+0x44/0x3b4 [<9016ffec>] usb_hcd_irq+0x68/0x70 ... And I cannot understand why. (I'm also unsure about the syntax - what does "IN-HARDIRQ-W" and "HARDIRQ-ON-W" mean?) Background: I have an interrupt routine isp1760_irq() that calls spin_lock(&priv->lock) ... spin_unlock(&priv->lock) I also take this lock in isp1760_urb_enqueue() mentioned in the printout. I believe this function gets called by the USB core when an urb should be transferred. isp1760_urb_enqueue() uses spin_lock_irqsave(&priv->lock, spinflags); ... spin_unlock_irqrestore(&priv->lock, spinflags); isp1760_urb_enqueue() also calls kmem_cache_zalloc(); this is what apparently triggers the printout above (the lock have already been successfully taken). The call to kmem_cache_zalloc() never returns; the printk() text on the line after is never displayed and the system locks up. The printk() om the line before displays. I also get an additional BUG: BUG: spinlock lockup on CPU#0, khubd/133, 93dcc138 Call trace: [<9001d370>] dump_stack+0x18/0x20 [<9012e7aa>] do_raw_spin_lock+0x9a/0xc8 [<9023622a>] _raw_spin_lock_irqsave+0x42/0x50 [<9017a7ac>] isp1760_irq+0x44/0x3b4 [<9016ffec>] usb_hcd_irq+0x68/0x70 [<9004b2fc>] handle_IRQ_event+0x16/0x92 [<9004cbec>] handle_level_irq+0xb0/0xc8 [<9001ad28>] demux_eic_irq+0x38/0x50 [<9001ac36>] do_IRQ+0x36/0x5c [<90019428>] irq_level0+0x18/0x60 [<9006aa96>] kmem_cache_alloc+0x4a/0x90 [<9017a59c>] isp1760_urb_enqueue+0x2e0/0x3e4 [<90170e54>] usb_hcd_submit_urb+0x6d4/0x758 [<90171186>] usb_submit_urb+0x1ce/0x1e0 [<901718aa>] usb_start_wait_urb+0x32/0x80 [<90171bc4>] usb_control_msg+0xc0/0xd8 [<9016d6ac>] hub_port_init+0x218/0x528 [<9016e672>] hub_thread+0x5aa/0xa38 [<900358c6>] kthread+0x58/0x62 [<90027c4a>] do_exit+0x0/0x43e which seems very weird to me, since it looks like isp1760_irq() gets called even though I thought I had disabled interrupts with spin_lock_irqsave() in isp1760_urb_enqueue(). Obviously I have not understood the locking in the kernel. What have I done wrong here? Thanks, Arvid Brodin Enea Services Stockholm AB

2 4