April 2014 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

about BTFIXUPSET_.. macros..
by Chan Kim April 15, 2014

April 15, 2014

Hi, in the linux initialization, inside setup_arch, I see ld_mmu_srmmu() (I'm working on SPARC architecture, SRMMU is Sparc Reference MMU) and it comprises many macro calls starting with BTFIXUP... The first two lines look like this. BTFIXUPSET_SIMM13(pgdir_shift, SRMMU_PGDIR_SHIFT); // ___ss_pgdir_shift <-- "first line" BTFIXUPSET_SETHI(pgdir_size, SRMMU_PGDIR_SIZE); // ___hs_pgdir_size ... The macro in the first line BTFIXUPSET_SIMM13 is expanded to #define BTFIXUPSET_SIMM13(__name, __val) \ do { \ ___ss_##__name[0] |= 1; \ ___ss_##__name[1] = (unsigned)__val; \ } while (0) So the first line just becomes ___ss_pgdir_shift[0] |= 1; ___ss_pgdir_shift[1] = (unsigned)SRMMU_PGDIR_SHIFT; If I look up __ss_pgdir_shift, it is found in arch/sparc/boot/btfix.S, ___ss_pgdir_shift: .word 0x73000000,0,68 .word _stext+0x000148c8,0,_stext+0x00014c94,0,_stext+0x00015348,0,_stext+0x00016310,0,_stext+0x000163c8,0,_stext+0x00016fb4,0,_stext+0x00017508,0,_stext+0x0007c31c,0,_stext+0x0007c9b0,0,_stext+0x0007da94,0,_stext+0x0007e310,0,_stext+0x0007edbc,0,_stext+0x0007f058,0,_stext+0x0007f234,0,_stext+0x0007f950,0,_stext+0x0007fbb4,0,_stext+0x00080408,0,_stext+0x00084540,0,_stext+0x00084c30,0,_stext+0x00085a60,0,_stext+0x00086598,0,_stext+0x00087b40,0,_stext+0x00087ce4,0,_stext+0x00088c54,0,_stext+0x00089f58,0,_stext+0x0008d748,0,_stext+0x003d08f4,0,__init_begin+0x0000659c,0,__init_begin+0x00007ac8,0,__init_begin+0x00008088,0,__init_begin+0x00008360,0,__init_begin+0x00008720,0,__init_begin+0x00008908,0,__init_begin+0x00008bf8,0 I think this assembly code was built before the compilation of ld_mmu_srmmu. (this BTFIXUP thing sounds like related to what is called pre-link, I saw it on the web) and This ld_mmu_srmmu function seems like to setup functions and values into specific sparc sub-architectures. (replacing genearal things with more specific things) So the first line is setting bit 0 of 0x73000000 and writing SRMMU_PGIDR_SHIFT (=2^24) to the next word. and there is a number 68 and it is followed by 68 words. I can see it looks like setting up a page directory table with certain length. But how has this assembly been built during pre-link? Is there any good reference? How are this values and functions used? I mean in what manner? Could somebody give me and idea how it's used later? Sorry for hasty questions. Best regards, Chan

1 0

Linux on PS4?
by Peter Senna Tschudin April 14, 2014

April 14, 2014

Did anyone succeed to run Linux on a Playstation 4? If so can one share instructions with me? Do one want help for making it happen? I would love to have a PS4 as a high end, compact, and cheap workstation for compiling Kernels... -- Peter

1 0

Kernel Module Compilation error
by Arun M Kumar April 14, 2014

April 14, 2014

Hi, I get the following error while compiling my very basic kernel Module (a hello world module) * //--------------------------------------------------------------------------------------------------------------* make -C /home/arun/Dev/linux-stable M=/home/arun/Dev/practice/Kmodules modules make[1]: Entering directory `/home/arun/Dev/linux-stable' WARNING: Symbol version dump /home/arun/Dev/linux-stable/Module.symvers is missing; modules will have no dependencies and modversions. CC [M] /home/arun/Dev/practice/Kmodules/Hello_world.o /home/arun/Dev/practice/Kmodules/Hello_world.c:1:0: error: code model 'kernel' not supported in the 32 bit mode /home/arun/Dev/practice/Kmodules/Hello_world.c:1:0: sorry, unimplemented: 64-bit mode not compiled in make[2]: *** [/home/arun/Dev/practice/Kmodules/Hello_world.o] Error 1 make[1]: *** [_module_/home/arun/Dev/practice/Kmodules] Error 2 make[1]: Leaving directory `/home/arun/Dev/linux-stable' make: *** [all] Error 2 *//-----------------------------------------------------------------------------------------------------------------* what could be the possible cause of this error? on my system *$uname -a* returns the following.. *Linux linux-xh55.site 3.4.6-2.10-desktop #1 SMP PREEMPT Thu Jul 26 09:36:26 UTC 2012 (641c197) i686 i686 i386 GNU/Linux*

3 2

APIs to submit bios at block layer
by KASHISH BHATIA April 14, 2014

April 14, 2014

Hi all, Could anyone explain the difference between submit_bio() and generic_make_request() ? How they are different in functionality? How lower layers treat bios submitted via submit_bio() / generic_make_request() ? When is it preferable to use api generic_make_request() than submit_bio() api? -- Regards, Kashish Bhatia

2 1

checkpatch.pl comment style warnings
by Greg Donald April 14, 2014

April 14, 2014

When I run checkpatch.pl -f, it complains about comment blocks like /* * foo * bar */ with "networking block comments don't use an empty /* line, use /* Comment..." But I found this: http://fr.it-usenet.org/thread/18772/41342/, which makes me think that particular checkpatch.pl warning should possibly be ignored for code not in net or drivers/net. 1) Are there different preferred commenting styles for networking code versus everything else? 2) Assuming the comment code above does need fixing, would an example fix be as simple as this? /* foo * bar */ 3) What about the very similar comment blocks checkpatch.pl doesn't currently complain about, do these need fixing too? /** * foo * bar */ Thanks. -- Greg Donald

3 4

Gianfar Ethernet Device Driver used in MPC8641D-HPCN
by Ashish Khetan April 14, 2014

April 14, 2014

Hi, I Have MPC8641D-HPCN based custom designed board and After some changes the Linux 3.14 is now booting on my board. we are using same phy device ie vitesse vsc8244. The Difference between my board and the original board is in only PHY interrupt ie Our custom Designed Board MPC8641D-HPCN original board eTSEC0(mac)----> phy0 ---- irq0 eTSEC0(mac)----> phy0 ---- irq10 eTSEC1(mac)----> phy1 ---- irq1 eTSEC1(mac)----> phy1 ---- irq10 eTSEC2(mac)----> phy2 ---- irq2 eTSEC2(mac)----> phy2 ---- irq10 eTSEC3(mac)----> phy3 ---- irq3 eTSEC3(mac)----> phy3 ---- irq10 After Booting Linux the first Ethernet device is only eTSEC0 is working means at booting time it detects all 4 ports and initialize all ports. The following IP addresses are assigned to each port: interface Board IP Host IP eth0 ---- 192.168.10.2 192.168.10.1 eth1 ---- 192.168.20.2 192.168.20.1 eth2 ---- 192.168.30.2 192.168.30.1 eth3 ---- 192.168.40.2 192.168.40.1 when i Tried to ping from board to host (ie ping 192.168.10.1) through eth0 its pinging. but when i tried to ping from other ports ie eth1, eth2 & eth3 its not pinging and prints "destination host not reachable". Also I tried to ping from pc(ie host) to board same result was printing. Then I started debugging, I put a printk function inside Gianfar.c file and gfar_clean_rx_ring() for printing packet sequence and packet size. I started pinging each port from pc (ie host) and its printing on each reception as: For eth0 [ 94.966113] Packet 1 size 60 [ 95.965588] Packet 2 size 120 [ 97.980729] Packet 3 size 180 [ 98.980705] Packet 4 size 240 [ 99.998872] Packet 5 size 300 and same for other interfaces too. Here is the device tree snap-shot for ethernet which I am using with this kernel (hope may be helpful): enet0: ethernet@24000 { #address-cells = <1>; #size-cells = <1>; cell-index = <0>; device_type = "network"; model = "TSEC"; compatible = "gianfar"; reg = <0x24000 0x1000>; ranges = <0x0 0x24000 0x1000>; local-mac-address = [ 00 00 00 00 00 00 ]; interrupts = <29 2 30 2 34 2>; interrupt-parent = <&mpic>; tbi-handle = <&tbi0>; phy-handle = <&phy0>; phy-connection-type = "rgmii-id"; mdio@520 { #address-cells = <1>; #size-cells = <0>; compatible = "fsl,gianfar-mdio"; reg = <0x520 0x20>; phy0: ethernet-phy@0 { interrupt-parent = <&mpic>; interrupts = <0 1>; reg = <0>; device_type = "ethernet-phy"; }; phy1: ethernet-phy@1 { interrupt-parent = <&mpic>; interrupts = <1 1>; reg = <1>; device_type = "ethernet-phy"; }; phy2: ethernet-phy@2 { interrupt-parent = <&mpic>; interrupts = <2 1>; reg = <2>; device_type = "ethernet-phy"; }; phy3: ethernet-phy@3 { interrupt-parent = <&mpic>; interrupts = <3 1>; reg = <3>; device_type = "ethernet-phy"; }; tbi0: tbi-phy@11 { reg = <0x11>; device_type = "tbi-phy"; }; }; }; enet1: ethernet@25000 { #address-cells = <1>; #size-cells = <1>; cell-index = <1>; device_type = "network"; model = "TSEC"; compatible = "gianfar"; reg = <0x25000 0x1000>; ranges = <0x0 0x25000 0x1000>; local-mac-address = [ 00 00 00 00 00 00 ]; interrupts = <35 2 36 2 40 2>; interrupt-parent = <&mpic>; tbi-handle = <&tbi1>; phy-handle = <&phy1>; phy-connection-type = "rgmii-id"; mdio@520 { #address-cells = <1>; #size-cells = <0>; compatible = "fsl,gianfar-tbi"; reg = <0x520 0x20>; tbi1: tbi-phy@11 { reg = <0x11>; device_type = "tbi-phy"; }; }; }; Is this happening because of phy interrupt or something else? Please through some light on this. how to make it working? how to disable phy interrupt? Does phy interrupt matters for gianfar interrupt or polling mode? Please give some pointer/anything to help more understanding of ethernet driver. Thanks And Regards A$hi$h

1 1

Run a program when DP monitor is plugged in/out
by arun c April 11, 2014

April 11, 2014

Hi, I want to run a program when DP monitor is plugged in/out. My linux kernel ( 3.14.0-00348-gbfcbf45) gets an interrupt when DP cable is removed or inserted. >From the graphics driver interrupt handler I can differentiate the two events. I want to run a user space program when this removal/insertion happens. What are the possible ways to achieve this? when I google for this I am seeing examples related to udev(How to run a script when a usb device is inserted/removed); but in my case it is not device insertion/removal. There is no need for a device creation when the DP monitor is plugged in/out(There is no device node available for me). Basically I want to send this removal/insertion event from my kernel interrupt handler to user space program. Regards, Arun C

3 2

how tso (tcp segmentation offload) effect TCP bidirectional test sending throughput.
by Varun Sharma April 11, 2014

April 11, 2014

Hi, if we off tso (tcp segmentation offload) on sending side , *TCP Bidirectional* test sending throughput decrease as compare to *TCP Unidirectional *test sending throughput . whereas if we on tso(tcp segmentation offload) on sending side , both testgive almost same sending throughput result. is anyone know how tso(tcp segmentation offload) effect TCP Bidirectional sending throughput ? i am using iperf-2.0.5 for this case. Thanks in advance. Regards varun

3 2

Linux hibernation support for loadable drivers
by AYAN KUMAR HALDER April 11, 2014

April 11, 2014

Hi, I am working with the linux hibernation code( swsusp hibernation framework) on an ARM based custom SOC. I am using some drivers like Nand Flash Controller driver as loadable modules. As per my understanding , during hibernation kernel calls suspend for all the drivers while creating a snapshot. During resume, after the kernel resumes the snapshot (from late_initcall - software_resme) , the kernel calls resume for all the drivers. As nand and some other drivers do not support suspend/resume, thus they fail to work on system resume. So is it fine to call driver_unregister/driver_register during hibernation/resume for the drivers which do not support suspend/resume. I do not want the user to manually remove/insert the drivers on suspend/resume. Regards, Ayan Kumar Halder

2 1

Linux hibernation support for loadable drivers
by Ayan Halder April 11, 2014

April 11, 2014

Hi, I am working with the linux hibernation code( swsusp hibernation framework) on an ARM based custom SOC. I am using some drivers like Nand Flash Controller driver as loadable modules. As per my understanding , during hibernation kernel calls suspend for all the drivers while creating a snapshot. During resume, after the kernel resumes the snapshot (from late_initcall - software_resme) , the kernel calls resume for all the drivers. As nand and some other drivers do not support suspend/resume, thus they fail to work on system resume. So is it fine to call driver_unregister/driver_register during hibernation/resume for the drivers which do not support suspend/resume. I do not want the user to manually remove/insert the drivers on suspend/resume. Regards, Ayan Kumar Halder

1 0