October 2011 - kernelnewbies@kernelnewbies.org

Current and correct CPU clock and asm("cpuid")
by Peter Senna Tschudin Oct. 6, 2011

Oct. 6, 2011

Dear list members, I'm following: http://people.virginia.edu/~chg5w/page3/assets/MeasuringUnix.pdf And I'm trying to measure executing time of simple operations with RDTSC. See the code below: #include <stdio.h> #define CPU_THOUSAND_HZ 800000 typedef unsigned long long ticks; static __inline__ ticks getticks(void) { unsigned a, d; asm("cpuid"); asm volatile("rdtsc" : "=a" (a), "=d" (d)); return (((ticks)a) | (((ticks)d) << 32)); } void main() { ticks tickBegin, tickEnd; tickBegin = getticks(); // code to time tickEnd = getticks(); double time = (tickEnd-tickBegin)/CPU_THOUSAND_HZ; printf ("%Le\n", time); } How can the C code detects the correct value for CPU_THOUSAND_HZ? The problems I see are: - It is needed to collect the information for the CPU that will run the process. On Core i7 processors, different cores can run at different clock speed at same time. - If the clock changes during the execution of process, what should it do? When is the best time for collecting the clock speed? The authors of the paper are not sure about the effects of "asm("cpuid");" Does it ensure that the entire process will run on the same CPU, and will serialize it avoiding out of order execution by the CPU? Thank you very much! :-) Peter -- Peter Senna Tschudin peter.senna(a)gmail.com gpg id: 48274C36

3 7

Re: [PATCH] hwmon class driver registration with a device number
by Himanshu Chauhan Oct. 6, 2011

Oct. 6, 2011

Hi, > I can not comment on the merits of your patch. Unless I am missing > something, which may well be since I only spent a couple of minutes on > it, other device classes don't seem to provide a similar API, so I don't > know if or why it would make sense for hwmon. Maybe a driver which wants > to register a character device interface should do so independently of > hwmon. > The idea here is to sit in the same class directory as of hwmon. Devices registered with this interface will have "dev" under, for example, /sys/class/hwmon/hwmon0/dev. To do the same inside the driver will be a bit more involved than a call. In my opinion other classes should also have similar interfaces. > On the technical side, EXPORT_SYMBOL is missing. > I will take care of that. Regards Himanshu

1 0

schedule history between 2 interrupts
by Amit Nagal Oct. 5, 2011

Oct. 5, 2011

Hi , i have a gap of 2-3 ms between 2 usb interrupts callback . i want to know which kernel/user threads execute in the gap of 2-3 ms . basically i want to log schedule history between 2 usb interrupts callback . which data structures / functions can i use to furnish schedule logging history for this case ? Regards Amit Nagal

3 2

how does linux restores a thread's stack pointer, program counter and return address.
by Smital Desai Oct. 4, 2011

Oct. 4, 2011

Hi. I am using linux 2.6.39 and MIPS core. I am keen to find out how does linux restores a thread's stack pointer, program counter and return address. I have created a binary which spawns four threads. with a ps -eaL , then I kill one of the thread with kill -11 <thread_id>. In the kernel 'do_coredump' handler i can easily find out the stack pointer of the current crashed thread by reading the struct pt_regs->reg29. My concern is to find out the stack pointers of rest of the 3 threads that did not crash. I did it this way: 1. read the reg29 variable of struct thread_struct for each task (By traversing the list of task with list_for_each) 2. Dump the values at address stored in reg29. 3. Look out manually to locate the Stack pointer offset. Tried this by crashing all the threads and found that the stackpointer comes at a fixed offset and assumed that stack pointer to be correct and I went ahead. so you can say that my Stack pointer is stored at => reg29 + offset; My question now is how to find this value of 'offset' dynamically. How does the kernel keeps a track of any thread's stack pointer while restoring the context. Since my observation was that the reg29 field of struct thread_struct at a glance seems to be a stack pointer. but it is not. it is just an address where at some offset we can find our stack pointer stored. Thanks Smital Desai ________________________________ The contents of this e-mail and any attachment(s) may contain confidential or privileged information for the intended recipient(s). Unintended recipients are prohibited from taking action on the basis of information in this e-mail and using or disseminating the information, and must notify the sender and delete it from their system. L&T Infotech will not accept responsibility or liability for the accuracy or completeness of, or the presence of any virus or disabling code in this e-mail" ______________________________________________________________________

3 2

Re: Current and correct CPU clock and asm("cpuid")
by Peter Senna Tschudin Oct. 4, 2011

Oct. 4, 2011

Hi Peter, The post: http://aufather.wordpress.com/2010/09/08/high-performance-time-measuremen-i… Is very good reference. The author is clever and It answers my questions. Thanks! :-) I also found old school document from 1997 made by Intel which was proud of its brand new Pentium-II. See: http://www.ccsl.carleton.ca/~jamuir/rdtscpm1.pdf The section 3 describes how to deal with out of order execution and L1 cache. Thanks! Peter On Mon, Oct 3, 2011 at 9:57 PM, Peter Teoh <htmldeveloper(a)gmail.com> wrote: > On Tue, Oct 4, 2011 at 2:57 AM, Peter Senna Tschudin > <peter.senna(a)gmail.com> wrote: >> Hi Peter, >> >> Thanks for the repply. I've realized that I have no need to transform >> the arbitrary number in something like seconds because I'm interested >> in comparing them. >> >> Is it safe to say that if I do not make the division by >> CPU_THOUSAND_HZ I have the number of clock cycles that were "spent" >> between the calls to getticks()(including some for getticks() itself)? >> >> Please see below. >> >> Thank you! >> >> Peter >> >> On Mon, Oct 3, 2011 at 1:17 PM, Peter Teoh <htmldeveloper(a)gmail.com> wrote: >>> why not u put a sleep(1) here like this: >>> >>>> ticks tickBegin, tickEnd; >>>> tickBegin = getticks(); >>>> >>> >>> sleep(1); >>> >>>> >>>> tickEnd = getticks(); >>>> double time = (tickEnd-tickBegin)/CPU_THOUSAND_HZ; >>>> >>> Then u know that it is reading the TSC values for 1 sec. And by >>> running the same program on different system u will get different >>> "time" values, and then u divide by that values for THAT system - so >>> that eventually running the same program on different system will get >>> u the same difference of ticks, which in our present case is "1". >>> After this "normalization", you can run your system with any timing >>> difference, and maximum achievable resolution is of course 1 sec. Is >>> that what u wanted? >> >> That sounds as great idea but: >> - may dynamic clock rate and multiple CPU cores mess with your proposal? >> - How precise is sleep about sleeping for 1 second? >> - I hope that the out of order execution mechanism of the CPU gets >> frustrated with your proposal and runs the instructions in the order >> we're expecting (tickBegin-> sleep-> tickEnd). How can we be sure that >> the instructions were run in correct order? >> >>> >>> BTW, modern OS does not use TSC any more, but yes, your assembly can >>> still access and read TSC. But the OS usually read from HPET (which >>> is how sleep(1) calculate the time differences) and to read the HPET >>> here is a link: >>> >>> http://www.fftw.org/cycle.h >> >> Looking cycle.h I found this familiar code(starts on line 216): >> >> /*----------------------------------------------------------------*/ >> /* >> * X86-64 cycle counter >> */ >> >> static __inline__ ticks getticks(void) >> { >> unsigned a, d; >> asm volatile("rdtsc" : "=a" (a), "=d" (d)); >> return ((ticks)a) | (((ticks)d) << 32); >> } >> > > Oh no, you are right, I just re-quote the link from Wiki which says > 'code to read the high-resolution timer on many CPUs and compilers' > ......ok, RDTSC is nevertheless a high resolution timer as well..... > >> The code found on cycle.h is so similar to the one I was using that I >> guess that both codes were written by the same author. I got the code >> I'm using from the paper at: >> http://people.virginia.edu/~chg5w/page3/assets/MeasuringUnix.pdf >> >>> >>> And query the OS via: >>> >>> cat /sys/devices/system/clocksource/clocksource0/* >>> hpet acpi_pm >>> hpet >>> > > The above is from my x86 Ubuntu 10.04 laptop. > >>> and u can see from above that "tsc" is missing from my system. >>> (linux kernel is 2.6.35-22) >>> >>> For TSC, I am not sure what is the highest resolution u can go, but in >>> a modern SoC chip, with 600Mhz core speed (speaking of PowerPC >>> http://en.wikipedia.org/wiki/PowerPC_e500) the fastest execution is >>> 600 millions instruction per sec, assuming the instruction is one insn >>> per clock. With this kind of speed, TSC is a very bad for measuring >>> time differences. >> >> This is my mistake. I did not told you about my tests will run only on x86 arch. >> > > Sorry to you too....I forgotten to mention that my hpet output is from > x86 arch. Anyway, TSC is nevertheless a valid timer, after some > research, I found its resolution is as good as HPET: > > http://aufather.wordpress.com/2010/09/08/high-performance-time-measuremen-i… > > But it did highlight lots of risks with TSC. > > And reading further: > > http://stackoverflow.com/questions/3388134/rdtsc-accuracy-across-cpu-cores > > http://stackoverflow.com/questions/3835111/whats-the-most-accurate-way-of-m… > > http://the-b.org/Linux_timers > > beware of something called "constant TSC" or 'invariant tsc', and > overflow time (all different timers are given except for TSC, in link > above) - if your duration is longer than that, the timer would have > turnaround before that and gave you inaccurate figures. > >>> >>> On Mon, Oct 3, 2011 at 9:27 AM, Peter Senna Tschudin >>> <peter.senna(a)gmail.com> wrote: >>>> Dear list members, >>>> >>>> I'm following: >>>> >>>> http://people.virginia.edu/~chg5w/page3/assets/MeasuringUnix.pdf >>>> >>>> And I'm trying to measure executing time of simple operations with RDTSC. >>>> >>>> See the code below: >>>> >>>> #include <stdio.h> >>>> #define CPU_THOUSAND_HZ 800000 >>>> typedef unsigned long long ticks; >>>> static __inline__ ticks getticks(void) { >>>> unsigned a, d; >>>> asm("cpuid"); >>>> asm volatile("rdtsc" : "=a" (a), "=d" (d)); >>>> return (((ticks)a) | (((ticks)d) << 32)); >>>> } >>>> >>>> void main() { >>>> ticks tickBegin, tickEnd; >>>> tickBegin = getticks(); >>>> >>>> // code to time >>>> >>>> tickEnd = getticks(); >>>> double time = (tickEnd-tickBegin)/CPU_THOUSAND_HZ; >>>> >>>> printf ("%Le\n", time); >>>> } >>>> >>>> How can the C code detects the correct value for CPU_THOUSAND_HZ? The >>>> problems I see are: >>>> - It is needed to collect the information for the CPU that will run >>>> the process. On Core i7 processors, different cores can run at >>>> different clock speed at same time. >>>> - If the clock changes during the execution of process, what should >>>> it do? When is the best time for collecting the clock speed? >>>> >>>> The authors of the paper are not sure about the effects of >>>> "asm("cpuid");" Does it ensure that the entire process will run on the >>>> same CPU, and will serialize it avoiding out of order execution by the >>>> CPU? >>>> >>>> Thank you very much! :-) >>>> >>>> Peter >>>> >>>> >>>> -- >>>> Peter Senna Tschudin >>>> peter.senna(a)gmail.com >>>> gpg id: 48274C36 >>>> >>>> _______________________________________________ >>>> Kernelnewbies mailing list >>>> Kernelnewbies(a)kernelnewbies.org >>>> http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies >>>> >>> >>> >>> >>> -- >>> Regards, >>> Peter Teoh >>> >> >> >> >> -- >> Peter Senna Tschudin >> peter.senna(a)gmail.com >> gpg id: 48274C36 >> > > > > -- > Regards, > Peter Teoh > -- Peter Senna Tschudin peter.senna(a)gmail.com gpg id: 48274C36

1 0

Re: Kernelnewbies Digest, Vol 10, Issue 39
by chao peng Oct. 3, 2011

Oct. 3, 2011

help 2011/9/23 <kernelnewbies-request(a)kernelnewbies.org> > Send Kernelnewbies mailing list submissions to > kernelnewbies(a)kernelnewbies.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies > or, via email, send a message with subject or body 'help' to > kernelnewbies-request(a)kernelnewbies.org > > You can reach the person managing the list at > kernelnewbies-owner(a)kernelnewbies.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kernelnewbies digest..." > > > Today's Topics: > > 1. Optimizing the performance of the Scheduler in the Kernel 3.0 > (Edward Alvarez) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 22 Sep 2011 15:35:21 +0000 > From: Edward Alvarez <edwardalvarezm(a)gmail.com> > Subject: Optimizing the performance of the Scheduler in the Kernel 3.0 > To: kernelnewbies(a)kernelnewbies.org > Message-ID: > <CAPzfdHbs7QBGpAwtPYFrA3y33Orpb%3DmBcwS6XH6Le66h5mMv3w(a)mail.gmail.com > > > Content-Type: text/plain; charset="iso-8859-1" > > Greetings, > <kernelnewbies(a)kernelnewbies.org> > > I want to optimize the performance of the scheduler in the Kernel 3.0 in > order to increase the number of calls a Sip proxy server can handle. > > I read about the scheduler in the Kernel 2.6 and was able to find the > variables MIN_TIMESLICE, MAX_TIMESLICE, PRIO_BONUS_RATIO, MAX_SLEEP_AVG and > STARVATION_LIMIT. If I tune this variables I can tweak the performance of > the scheduler, but in the new kernel I cant find any documentation that > allows to successfully modify the scheduler. > > I would gladly appreciate any information related to the scheduler in the > new kernel, or documentation about the new variables that are similar to > the > ones I mentioned. > > Thanks in advance for your time. > > Edward Alvarez. >

2 1

Getting dirtied pages info
by Prateek Sharma Oct. 3, 2011

Oct. 3, 2011

Hi all, What would be a good way to get a list of all pages(lets say pfns) dirtied in a particular time interval ? I am reading about MMU_notifiers but i think a simpler, light-weight solution might work.. I came across the page-dirtied bits...are those saved somewhere by the kernel on context-switches? Thanks!

2 2

where's the linux-staging tree ?
by Víctor M. Jáquez L. Oct. 2, 2011

Oct. 2, 2011

Hi, Now that kernel.org is out of air, where's linux-stating living now? I searched in github, where everybody is storing their trees, but I couldn't find it. vmjl

3 2

Backtrace Call Stack
by Harish G. Naik Oct. 2, 2011

Oct. 2, 2011

Hi, I have been trying to get some debug information from a point in the kernel where I am seeing some errors. I am actually trying to figure out how a particular point in code was reached. Here is what I did: if (case_when_error_happens) { extern int sprint_symbol(char *buffer, unsigned long address); for (i = 0; i < depth; i++) sprint_symbol(bt[i], (unsigned long)__builtin_return_address(i)); printk("%s < %s ...", bt[0], bt[1] ...); } I get most of the information I want. But I was wondering if there was any other cleaner, more straight forward way to do this. Thanks. -Harish

2 3

What's the meaning of SUBDIRS ?
by Parmenides Oct. 1, 2011

Oct. 1, 2011

Hi, When making a kernel module outside the kernel source tree, I can use this command to make my kernel module: make -C /usr/src/linux SUBDIRS=$PWD modules According to man, the -C option will chanage the current working directory to the top level directory of kernel source tree. Furthermore, the SUBDIRS varialbe contains the path of current working directory where my module is living. So, why does 'make' need to enter /usr/src/linux first? And, What's the role of 'SUBDIRS'? Thanks.

2 1