[Savannah-hackers] Re: submission of Fast Kernel Tracing - savannah.nongnu.org

[Robert D. Russell]

Hello, is anybody there?

I submitted a project registration on 17 April, and have heard nothing
from you about this after the initial confirmation that you did
receive the registration.  I sent you an e-mail 3 weeks later, and
also got no reply.  So now it is 6 weeks since I submitted the
registration, and I still have not heard anything from you.

Could you please let me know if my registration is lost, still being
processed, has been refused, or any other status about it.

Thank you for your time,
Bob Russell

On Sun, 9 May 2004, Robert D. Russell wrote:

> Hello:
> 
> I was wondering if any action has been taken on this submission --
> it has been 3 weeks so far, and I have heard nothing, pro or con.
> 
> Thanks,
> Bob Russell
> 
> 
> On Sat, 17 Apr 2004 address@hidden wrote:
> 
> > A package was submitted to savannah.nongnu.org
> > This mail was sent to address@hidden, address@hidden
> > 
> > 
> > Bob Russell <address@hidden> described the package as follows:
> > License: gpl
> > Other License: 
> > Package: Fast Kernel Tracing
> > System name: fkt
> > Type: non-GNU
> > 
> > Description:
> > This project has developed a flexible, efficient software tool for
> > accurate performance evaluation and tracing of code in the kernel
> > and in loadable kernel modules of the Linux operating system on
> > Intel Pentium processors, both uniprocessors and multiprocessors.
> > We call the tool FKT (Fast Kernel Tracing).  It consists of a
> > number of modifications and additions to the Linux kernel that
> > collect and record data in real time, as well as some user-level
> > programs to control the probing and to perform off-line analysis.
> > The current code base is at the URL:
> > http://perso.ens-lyon.fr/samuel.thibault/fkt-20040412.tgz
> > 
> > The method consists of placing special software "probe" macros at
> > various locations in the kernel source code.  The source code must
> > then be recompiled, and compilation of the probes is controlled by
> > a configuration symbol (CONFIG_FKT).  Placement of probes is
> > determined by the kernel programmer, and can be done manually or
> > automatically via the "instrument_function" attribute provided by the GNU C 
> > compiler.  Typically, one probe is placed at the entry to, and another 
> > probe at the exit from, every kernel function of interest to the 
> > programmer.  By default, probes are always placed at the entry to and exit 
> > from every system call, trap, and interrupt, and at the process switch 
> > point in the kernel.
> > 
> > Each probe supplies a "mask" parameter with one or more bits set
> > in it.  All probes having the same bit set in their keymask form
> > a "probe-set", and a probe-set can be enabled or disabled by
> > user-level programs through the use of some new system calls.
> > When a probe is disabled, the only overhead it introduces is the cost of a 
> > bitmask test and branch.  A probe can also be supplied with other 
> > parameters of the programmer's choosing, which makes it a useful debugging 
> > tool, especially valuable when developing new modules and/or additions to 
> > the kernel.
> > 
> > Probes record data into kernel buffers that are written to a file
> > set up at the beginning of an FKT run.  Recording is done using
> > the Linux "sendpage()" mechanism, which avoids extra copying of
> > the buffers.  Each probe records an unique identification code, a
> > timestamp obtained from the hardware Time Stamp Counter, the
> > number of the processor executing the code, and the additional
> > parameters, if any.  The off-line analysis program, run after the
> > file has been completely written, maps the identification code to
> > text identifying the probe (typically the name of the function in
> > which it was placed).  At the start of a run, the kernel symbol
> > table, and the symbol tables of all loaded modules, are recorded
> > in the file so that they are kept together with the data recorded
> > by the probes.  The offline analysis can use these symbol tables
> > to map the locations generated by the GCC "instrument_function"
> > feature into symbolic names.
> > 
> > The current offline analysis offers a number of options to produce detailed 
> > traces, function nesting traces, and exact accounting for every machine 
> > cycle.  Future plans call for developing a general library that would allow 
> > users to easily develop their own analysis programs, and a graphical 
> > interface to the analysis output.
> > 
> > 
> > 
> > 
> > Other Software Required:
> > This project depends on the Linux kernel and the Intel x86 hardware.  It 
> > should be possible to extend it to run on other hardware.
> > 
> > 
> > Other Comments:
> > 
> > 
> > 
> > 
> > 
> > *In case you have to register your project again.
> > 
> > Please be aware that if your registration does not fulfill all the 
> > requirements, the Savannah administrators may ask you to register your 
> > project again.You can use the following url to do a new registration 
> > starting with the values used in this registration process.
> > Copy and paste AS ONE SINGLE URL the following content:
> > ----RERegistration-URL-BEGIN-----
> > http://savannah.nongnu.org/register/basicinfo.php?re_purpose=This%20project%20has%20developed%20a%20flexible%2C%20efficient%20software%20tool%20for%0D%0Aaccurate%20performance%20evaluation%20and%20tracing%20of%20code%20in%20the%20kernel%0D%0Aand%20in%20loadable%20kernel%20modules%20of%20the%20Linux%20operating%20system%20on%0D%0AIntel%20Pentium%20processors%2C%20both%20uniprocessors%20and%20multiprocessors.%0D%0AWe%20call%20the%20tool%20FKT%20%28Fast%20Kernel%20Tracing%29.%20%20It%20consists%20of%20a%0D%0Anumber%20of%20modifications%20and%20additions%20to%20the%20Linux%20kernel%20that%0D%0Acollect%20and%20record%20data%20in%20real%20time%2C%20as%20well%20as%20some%20user-level%0D%0Aprograms%20to%20control%20the%20probing%20and%20to%20perform%20off-line%20analysis.%0D%0AThe%20current%20code%20base%20is%20at%20the%20URL%3A%0D%0Ahttp%3A%2F%2Fperso.ens-lyon.fr%2Fsamuel.thibault%2Ffkt-20040412.tgz%0D%0A%0D%0AThe%20method%20consists%20of%20placing%20special%20software%20%
> > C%22probe%5C%22%20macros%20at%0D%0Avarious%20locations%20in%20the%20kernel%20source%20code.%20%20The%20source%20code%20must%0D%0Athen%20be%20recompiled%2C%20and%20compilation%20of%20the%20probes%20is%20controlled%20by%0D%0Aa%20configuration%20symbol%20%28CONFIG_FKT%29.%20%20Placement%20of%20probes%20is%0D%0Adetermined%20by%20the%20kernel%20programmer%2C%20and%20can%20be%20done%20manually%20or%0D%0Aautomatically%20via%20the%20%5C%22instrument_function%5C%22%20attribute%20provided%20by%20the%20GNU%20C%20compiler.%20%20Typically%2C%20one%20probe%20is%20placed%20at%20the%20entry%20to%2C%20and%20another%20probe%20at%20the%20exit%20from%2C%20every%20kernel%20function%20of%20interest%20to%20the%20programmer.%20%20By%20default%2C%20probes%20are%20always%20placed%20at%20the%20entry%20to%20and%20exit%20from%20every%20system%20call%2C%20trap%2C%20and%20interrupt%2C%20and%20at%20the%20process%20switch%20point%20in%20the%20kernel.%0D%0A%0D%0AEach%20probe%20supplies%20a%20%5C%22m
> > sk%5C%22%20parameter%20with%20one%20or%20more%20bits%20set%0D%0Ain%20it.%20%20All%20probes%20having%20the%20same%20bit%20set%20in%20their%20keymask%20form%0D%0Aa%20%5C%22probe-set%5C%22%2C%20and%20a%20probe-set%20can%20be%20enabled%20or%20disabled%20by%0D%0Auser-level%20programs%20through%20the%20use%20of%20some%20new%20system%20calls.%0D%0AWhen%20a%20probe%20is%20disabled%2C%20the%20only%20overhead%20it%20introduces%20is%20the%20cost%20of%20a%20bitmask%20test%20and%20branch.%20%20A%20probe%20can%20also%20be%20supplied%20with%20other%20parameters%20of%20the%20programmer%5C%27s%20choosing%2C%20which%20makes%20it%20a%20useful%20debugging%20tool%2C%20especially%20valuable%20when%20developing%20new%20modules%20and%2For%20additions%20to%20the%20kernel.%0D%0A%0D%0AProbes%20record%20data%20into%20kernel%20buffers%20that%20are%20written%20to%20a%20file%0D%0Aset%20up%20at%20the%20beginning%20of%20an%20FKT%20run.%20%20Recording%20is%20done%20using%0D%0Athe%20Linux%20%5C%22sen
> > page%28%29%5C%22%20mechanism%2C%20which%20avoids%20extra%20copying%20of%0D%0Athe%20buffers.%20%20Each%20probe%20records%20an%20unique%20identification%20code%2C%20a%0D%0Atimestamp%20obtained%20from%20the%20hardware%20Time%20Stamp%20Counter%2C%20the%0D%0Anumber%20of%20the%20processor%20executing%20the%20code%2C%20and%20the%20additional%0D%0Aparameters%2C%20if%20any.%20%20The%20off-line%20analysis%20program%2C%20run%20after%20the%0D%0Afile%20has%20been%20completely%20written%2C%20maps%20the%20identification%20code%20to%0D%0Atext%20identifying%20the%20probe%20%28typically%20the%20name%20of%20the%20function%20in%0D%0Awhich%20it%20was%20placed%29.%20%20At%20the%20start%20of%20a%20run%2C%20the%20kernel%20symbol%0D%0Atable%2C%20and%20the%20symbol%20tables%20of%20all%20loaded%20modules%2C%20are%20recorded%0D%0Ain%20the%20file%20so%20that%20they%20are%20kept%20together%20with%20the%20data%20recorded%0D%0Aby%20the%20probes.%20%20The%20offline%20analysis%20can%20use%20these%20
> > ymbol%20tables%0D%0Ato%20map%20the%20locations%20generated%20by%20the%20GCC%20%5C%22instrument_function%5C%22%0D%0Afeature%20into%20symbolic%20names.%0D%0A%0D%0AThe%20current%20offline%20analysis%20offers%20a%20number%20of%20options%20to%20produce%20detailed%20traces%2C%20function%20nesting%20traces%2C%20and%20exact%20accounting%20for%20every%20machine%20cycle.%20%20Future%20plans%20call%20for%20developing%20a%20general%20library%20that%20would%20allow%20users%20to%20easily%20develop%20their%20own%20analysis%20programs%2C%20and%20a%20graphical%20interface%20to%20the%20analysis%20output.%0D%0A%0D%0A%0D%0A&re_require_sw=This%20project%20depends%20on%20the%20Linux%20kernel%20and%20the%20Intel%20x86%20hardware.%20%20It%20should%20be%20possible%20to%20extend%20it%20to%20run%20on%20other%20hardware.%0D%0A&re_comments=&re_full_name=Fast%20Kernel%20Tracing&re_unix_name=fkt
> > ----RERegistration-URL-END-------
> > 
> > 
> > _______________________________________________
> >   Message sent via/by Savannah
> >   http://savannah.nongnu.org/
> > 
> > 
> > 
> > 
> 
>