One final confirmation for 2.6.18.8: It seems to be booting extremely slow even with atomic cpu model. It got stuck at "[4194001.857552] SMP: Total of 8 processors activated (31985.87 BogoMIPS)." for long time. Earlier 2.6.13 was booting very fast though in same machine with same workload. Do you also see this slow boot-up?
Just for curiosity, what does [4194001.857552] stand for? --- Ali Saidi <[EMAIL PROTECTED]> escribió: > The diffs create a .config.m5 which you need to copy > to .config before > you build. > > Ali > > On Feb 6, 2008, at 11:25 AM, nathan binkert wrote: > > > Take no offense at this question, I ask it of all > programmers I work > > with (they can attest to it): > > > > Are you absolutely sure you're doing what you > think you're doing? It's > > a good rule to go back and verify all of your > assumptions about what > > you've done when things don't make sense. (I kick > myself when I > > don't follow that rule myself.) > > > > Said another way, are you sure you recompiled > everything correctly and > > that m5 is using the kernel binary that you think > it is? A hidden > > compiler error message, or an incorrectly set > config parameter, or > > even NFS can thwart these things. config.ini has > information about > > what M5 is using. Are you certain that you're > even looking at the > > correct output file? > > > > That said, there should be a .config file that > we've provided in the > > patch set. Is it not there? > > > > Nate > > > > On Feb 6, 2008 2:56 AM, abc def > <[EMAIL PROTECTED]> wrote: > >> I did configure kernel using menuconfig before > running > >> ,turned framebuffer from Graphics support and > then > >> checked in the .config that "# CONFIG_FB is not > set" + > >> no mention of FRAMEBUFFER. > >> > >> However after compiling when i try to use the > vmlinux > >> the same error is coming. > >> > >> > >> > >> > >> --- Ali Saidi <[EMAIL PROTECTED]> escribió: > >> > >>> You should figure out what the hunk failed and > fix > >>> it, but the problem > >>> is in the config file. M5 doesn't supporrt frame > >>> buffers so you need > >>> to turn off the frame buffer in your kernel > config > >>> file. > >>> > >>> Ali > >>> > >>> On Feb 5, 2008, at 12:05 PM, abc def wrote: > >>> > >>>> Hi, > >>>> > >>>> I applied the patch using patch command on > >>>> linux-2.6.18 source tree and patch seems to be > >>> working > >>>> fine, though for patch_to_2.6.18.8.diff it gave > >>> one > >>>> failure for one hunk. > >>>> > >>>> Then i compiled the kernel using cross > compiler. > >>>> > >>>> But when i try to boot with new vmlinux i got > this > >>>> panic and m5 aborted. > >>>> > >>>> panic: Device FrameBuffer not imlpmented > >>>> @ cycle 21786981600 > >>>> [read:build/ALPHA_FS/dev/baddev.cc, line 58] > >>>> Program aborted at cycle 21786981600 > >>>> > >>>> At that time in console: > >>>> Entering slaveloop for cpu 3 > >>> my_rpb=FFFFFC0000018900 > >>>> Entering slaveloop for cpu 2 > >>> my_rpb=FFFFFC0000018680 > >>>> > >>>> Now is it happening because i didnt apply patch > >>> using > >>>> MQ and something got missing?. If that is the > case > >>>> then i will try to figure out about mercurial > >>>> repository details. > >>>> > >>>> Any help will be highly appreciated. > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> --- Ali Saidi <[EMAIL PROTECTED]> escribió: > >>>> > >>>>> I just added them to the download page > >>>>> (linux_patches.tar.bz2). > >>>>> > >>>>> Ali > >>>>> > >>>>> On Jan 30, 2008, at 1:29 PM, nathan binkert > >>> wrote: > >>>>> > >>>>>> I'm certain that this isn't an M5 problem > since > >>>>> I've been using that > >>>>>> call a lot. 2.6.13 is probably a bit > outdated. > >>>>> I've been using > >>>>>> 2.6.18 lately and it works there. The port > to > >>>>> more recent versions is > >>>>>> pretty easy. Ali, aren't the patches > available > >>>>> somewhere? If not, we > >>>>>> can put them all up on the website. > >>>>>> > >>>>>> On Jan 30, 2008 9:21 AM, abc def > >>>>> <[EMAIL PROTECTED]> wrote: > >>>>>>> Hi, > >>>>>>> > >>>>>>> I am running a multithreaded application > with > >>>>> regular > >>>>>>> linux pthreads. > >>>>>>> > >>>>>>> I want to bind a single thread to a single > >>>>> simulated > >>>>>>> processor. For this purpose i am using > >>>>>>> sched_setaffinity system call in side the > >>> common > >>>>>>> function executed by all threads. > >>>>>>> > >>>>>>> But while executing the program the > >>>>> sched_setaffinity > >>>>>>> is failing and i get "Function Not > implemented" > >>>>>>> message. > >>>>>>> > >>>>>>> I am using m5 beta 4 and default version for > >>>>> kernel > >>>>>>> (2.6.13). > >>>>>>> If the kernel version is not proper for this > >>>>> feature > >>>>>>> then is there any other recent version > >>> available > >>>>> which > >>>>>>> has support for processor affinity system > call > >>>>> for > >>>>>>> alpha? > >>>>>>> > >>>>>>> Any help will be highly appreciated. > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> > ______________________________________________ > >>>>>>> ¿Con Mascota por primera vez? Sé un mejor > >>> Amigo. > >>>>> Entra en Yahoo! > >>>>>>> Respuestas > >>>>> http://es.answers.yahoo.com/info/welcome > >>>>>>> > _______________________________________________ > >>>>>>> m5-users mailing list > >>>>>>> m5-users@m5sim.org > >>>>>>> > >>>>> > >>> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > >>>>>>> > >>>>>>> > >>>>>> > _______________________________________________ > >>>>>> m5-users mailing list > >>>>>> m5-users@m5sim.org > >>>>>> > >>> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > === message truncated === ______________________________________________ ¿Con Mascota por primera vez? Sé un mejor Amigo. Entra en Yahoo! Respuestas http://es.answers.yahoo.com/info/welcome _______________________________________________ m5-users mailing list m5-users@m5sim.org http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
