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

Identifying whether a user-process or kernel-thread execution takes place by looking at CPU registers
by limp May 14, 2011

May 14, 2011

Hi all, About a month ago I've made a question in the ML regarding the contents of the CR3 register and how the value of this register change when different processes/threads are executed. One of the conclusions of the conversation we had was that when kernel threads are executed, the CR3 register keeps the value of the process under which the kernel thread is executed (which is the previously executed process). That is, the kernel thread "borrows" the page-table of the previously executed process for executing. The problem is that by looking *only* at the CR3 register, we can't really tell if the user process in which the CR3 value corresponds to (each process is assigned a unique PGD value which results to a unique CR3 value) *or* a kernel-thread that uses its tables is executing. I was wondering if any other CPU register (apart from CR3) can indicate if a user-process or a kernel thread under it (and which one) is executed. Is it possible to know such a thing *only* by looking at CPU registers? I'll be happy to clarify any of the above. Thank you all in advance. John K.

3 4

confused by char dev registration in a gpio driver
by Robert P. J. Day May 13, 2011

May 13, 2011

i'm sure i'm going to embarrass myself here but i was perusing the char drivers for nice examples, and i ran across this excerpt in pc8736x_gpio.c: ===== begin ===== if (major) { devid = MKDEV(major, 0); rc = register_chrdev_region(devid, PC8736X_GPIO_CT, DEVNAME); } else { rc = alloc_chrdev_region(&devid, 0, PC8736X_GPIO_CT, DEVNAME); major = MAJOR(devid); } if (rc < 0) { dev_err(&pdev->dev, "register-chrdev failed: %d\n", rc); goto undo_request_region; } if (!major) { major = rc; dev_dbg(&pdev->dev, "got dynamic major %d\n", major); } ===== end ===== i'm good with most of that -- if the (parameter) major is explicit, then a dev_t of "devid" is created and register_chrdev_region() is used. on the other hand, if major is zero, then alloc_chrdev_region() is used for *dynamic* allocation of the major number. in both cases, the return code "rc" is saved and, if it's < 0, we have an error. and that's where the confusion comes in. i always thought both of those routines returned a simple zero to indicate success. but look at those last few lines -- that return code is assigned to "major", at which point it's *that* value that's printed. wouldn't that just be zero all the time? and wouldn't it also print that this was a "dynamic" major even if the user specified an explicit major number at load time? this second point is more nitpicky, but what about that first point? wouldn't a successful registration always print an allocated major number of zero? rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

3 6

any real or previous usage of the IRQ-related seqlock calls?
by Robert P. J. Day May 13, 2011

May 13, 2011

i was curious about the level of usage of the various locking features in the kernel, so i was poking around using http://lxr.linux.no, and checked out the seqlock-related calls, and noticed that all of the IRQ-related seqlock calls don't seem to be used in the current kernel tree at all: http://lxr.linux.no/linux+v2.6.38/include/linux/seqlock.h#L239 if you select any of those calls starting at "write_seqlock_irqsave", it doesn't look like any of them are currently used. do they simply have no current value? just curious. rday -- ======================================================================== Robert P. J. Day Ottawa, Ontario, CANADA http://crashcourse.ca Twitter: http://twitter.com/rpjday LinkedIn: http://ca.linkedin.com/in/rpjday ========================================================================

1 0

Module to sanitize freed memory ?
by Chris Laberti May 13, 2011

May 13, 2011

Hi, Is there a kernel module that sanitizes (overwrites) freed memory ? Or maybe does sysctl offer something like this ? Regards, Chris

3 3

Pericom PI7C9X110 bridge driver
by madhukar m May 13, 2011

May 13, 2011

Hi, Does anybody worked on "pericom PI7C9X110 bridge"(PCI-Ex--to--PCI) in Non-transparent Mode. if so, do you have any reference code driver development in Linux. Or any Basic register initialization for the bridge to work for data-communication(in BackPlane) between two bridges in NT mode. Thanks in Advance.. Madhukar.

1 0

what is the __io_address macro for?
by Christopher Harvey May 13, 2011

May 13, 2011

http://lxr.linux.no/#linux+v2.6.38/arch/arm/mach-realview/include/mach/hard… Seems to be used in hardware that can be emulated with QEMU. The offset and mask seem totally arbitrary to me, but multiple mach types use the same ones. I'm trying to use it to access the arm global timer. I can use it to access the CPU local timer, but simply adding the offset to the global timer registers causes QEMU to land in the GIC CPU interface. This isn't for the realview board btw, just a similar example. I feel like there is something fundamental I'm missing that is preventing me from understanding why: __io_address(BASE + 0x600) //local timer, works, no problems __io_address(BASE + 0x200) //global timer, but reads from gic cpu interface? that happens. (I am using QEMU btw) arm docs for offsets here: (bottom table) http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0407f/CACCJF… Thanks for reading, Chris.

1 0

Re: module loading order
by Vikram Narayanan May 12, 2011

May 12, 2011

>> This is a question that arose from a older kernel ( 2.6.21 ) but I think >> it's generic to all kernels. Actually this might not a question on the >> kernel itself at all. So here it goes, >> >> I'm using a USB 3G modem with this kernel. The system loads three >> modules for it: pl2303, option and usbserial. Problem is, it looks like >> the order or loading is different ( I'm judging this by the order they >> appear in lsmod output): >> >> Sometimes, lsmod says: >> pl2303 18628 0 >> option 9664 0 >> usbserial 28296 2 pl2303,option >> >> And sometimes, it says: >> option 9664 0 >> pl2303 18628 0 >> usbserial 28296 2 option,pl2303 >> >> Problem is, if option is before pl2303, the device won't work properly. >> So my question is: who decides which is loaded first, or, why does lsmod >> give different order? > > It's random and there is no way to specify the order unless you make one > module a dependancy of the other. Isn't modules.dep file take care of this? > And for these modules, they should not be dependant on each other, so > you can't do that. It does I guess. In my system, the modules.dep has, kernel/drivers/usb/serial/pl2303.ko: kernel/drivers/usb/serial/usbserial.ko Doesn't it mean that pl2303 is dependent on usbserial? - Thanks, Vikram

3 6

KGDB Setup problem
by Prabhu nath May 12, 2011

May 12, 2011

Dear All / Spiro, I tried setting up KGDB on my machine but somehow I am not finding my luck for past few months. Sprio. in his earlier mail had said about his success of setting up KGDB, Can you please share your notes of how to set up KGDB. If possibe please share the screen shots. Let me feel happy about the gdb prompt. Following is my set up. I have two x86 single core machines. C1 and C2. C1 = target C2 = Host On Target I have built a custom kernel 2.6.34 with KGDB enabled. Here is the snapshot of my .config CONFIG_MAGIC_SYSRQ=y CONFIG_DEBUG_INFO=y CONFIG_HAVE_ARCH_KGDB=y CONFIG_KGDB=y CONFIG_KGDB_SERIAL_CONSOLE=y My vmlinux is around 65MB. *Scenario 1:* I connected a null modem cable from C1 to C2. Tested the cable by redirecting the kernel boot logs from C1 to C2 thru the cable and cable is fine. I connected C1 to a LCD monitor (console=tty0) through a VGA cable so that I used tty0 to view normal kernel booting logs and null modem cable to work with debugger My grub.conf is title Fedora (2.6.34) root (hd0,1) kernel /vmlinuz-2.6.34 ro root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 KEYTABLE=us console=tty0 kgdbwait kgdb8250=ttyS0,115200 initrd /initramfs-2.6.34.img *The result was, on booting up the kernel kgdb was not waiting. * *Scenario 2. *I connected a null modem cable from C1 to C2. Tested the cable by redirecting the kernel boot logs from C1 to C2 thru the cable and cable is fine. I connected C1 to a LCD monitor (console=tty0) through a VGA cable so that I tty0 to view normal kernel booting logs and null modem cable to work with debugger My grub.conf is title Fedora (2.6.34) root (hd0,1) kernel /vmlinuz-2.6.34 ro root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 KEYTABLE=us console=tty0 kgdbwait kgdboc=ttyS0,115200 initrd /initramfs-2.6.34.img *The result is, on booting up the kernel kgdb is waiting for the input ( This log was seen on my LCD monitor) but Alt-SysRq-g did not work here to switch to the debugger * *Scenario 3:* I connected a null modem cable from C1 to C2. Tested the cable by redirecting the kernel boot logs from C1 to C2 thru the cable and cable is fine. My grub.conf is title Fedora (2.6.34) root (hd0,1) kernel /vmlinuz-2.6.34 ro root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 KEYTABLE=us console=ttyS0,115200n8 kgdbwait kgdboc=ttyS0,115200 kgdbcon initrd /initramfs-2.6.34.img * The result is, on booting up the kernel kgdb is waiting for the input ( This log was seen on my minicom console) but Alt-SysRq-g did not work and it threw some garbage. * On the host side I executed the following commands. # gdb ./vmlinux (gdb) set remotebaud 115200 (gdb) target remote /dev/ttyS0 Some of the documents say that after executing the last command I should see the gdb prompt from where I can set breakpoints, single stepping, but didn't show up for me :( *Note:* If it really matters. I was working on my host machine through a putty session. Request you or anyone to bail me out from the clutches of this precarious problem. Thanks, Prabhu

1 1

tools to identify bugs in driver
by Amit Nagal May 12, 2011

May 12, 2011

Hi , What are the mandatory tools that must be run to identify potential bugs ( say memory , locking etc ) in a linux device driver ? Regards Amit Nagal

3 2

Re: KGDB set up problem
by Spiro Trikaliotis May 11, 2011

May 11, 2011

Hello Prabhu, your mail formatting is somehow distorted. However, I will try to answer anyway: * On Wed, May 11, 2011 at 11:03:29PM +0530 Prabhu nath wrote: > My grub.conf is > title Fedora (2.6.34) > root (hd0,1) > kernel /vmlinuz-2.6.34 ro > root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM > rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 > KEYTABLE=us > console=tty0 kgdbwait > kgdb8250=ttyS0,115200 [...] > The result was, on booting up the kernel kgdb was not waiting. Here, the command line parameter are wrong. First: kgdb8250 is called kgdboc now. Second: Order matters. You must specify kgdbwait AFTER kgdboc in order for it to take effect. > Scenario 2. > I connected a null modem cable from C1 to C2. Tested the cable by > redirecting the kernel boot logs from C1 to C2 thru the cable and cable > is fine. > I connected C1 to a LCD monitor (console=tty0) through a VGA cable so > that I tty0 to view normal kernel booting logs and null modem cable to > work with debugger > My grub.conf is > title Fedora (2.6.34) > root (hd0,1) > kernel /vmlinuz-2.6.34 ro > root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM > rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 > KEYTABLE=us > console=tty0 kgdbwait > kgdboc=ttyS0,115200 > initrd /initramfs-2.6.34.img > The result is, on booting up the kernel kgdb is waiting for the input ( > This log was seen on my LCD monitor) but Alt-SysRq-g did not work here > to switch to the debugger Is it really waiting? I could not get it to work unless I switched the order of kgdbwait and kgdboc. I think I also found it written somewhere on the net that order matters here. Now, regarding sysrq: - What is the output of: # cat /proc/sys/kernel/sysrq Either specify a 1, or a bit mask which selectively switches everything on. -OR- add "kernel.sysrq = 1" to /etc/sysctl.conf. Having said this, Alt-Sysrq-g does not work for me, either! But: # echo g > /proc/sysrq-trigger Works for me. I have not yet figured out why this is. Having said all this: Do you want to use the magic sysrq to initially break into the debugger? This does not work, and it is not necessary (if you switch the order of kgdboc and kgdbwait). > Scenario 3: > I connected a null modem cable from C1 to C2. Tested the cable by > redirecting the kernel boot logs from C1 to C2 thru the cable and > cable is fine. > My grub.conf is > title Fedora (2.6.34) > root (hd0,1) > kernel /vmlinuz-2.6.34 ro > root=UUID=4a51e22e-ec12-469f-80ff-17afe3e6f15f rd_NO_LUKS rd_NO_LVM > rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 > KEYTABLE=us > console=ttyS0,115200n8 > kgdbwait kgdboc=ttyS0,115200 kgdbcon I never tried to have the console and kgdboc on the same serial port. Is this even supposed to work? BTW, your idea about writing an article on how I set it up is a good idea. I will try to find the time to do so. HTH Spiro. -- Spiro R. Trikaliotis http://opencbm.sf.net/ http://www.trikaliotis.net/ http://www.viceteam.org/

1 0