Am 11.08.2012 14:39, schrieb Andrew: > 11.08.2012 02:54, KP Kirchdoerfer пишет: >> Am 09.08.2012 21:02, schrieb Andrew: >>> 09.08.2012 21:53, KP Kirchdoerfer пишет: >>>> Hi >>>> I tried to build a recent kernel (3.2.25 and 3.2.26), but both failed >>>> with the error: >>>> >>>> >>>> CC init/version.o >>>> LD init/built-in.o >>>> LD .tmp_vmlinux1 >>>> kernel/built-in.o: In function `tick_nohz_stop_sched_tick': >>>> (.text+0x2f55c): undefined reference to `calc_load_enter_idle' >>>> kernel/built-in.o: In function `tick_nohz_restart_sched_tick': >>>> (.text+0x2f7fd): undefined reference to `update_cpu_load_nohz' >>>> kernel/built-in.o: In function `tick_nohz_restart_sched_tick': >>>> (.text+0x2f819): undefined reference to `calc_load_exit_idle' >>>> make[3]: *** [.tmp_vmlinux1] Error 1 >>>> make[2]: *** [sub-make] Error 2 >>>> make[1]: *** [all] Error 2 >>>> make[1]: Leaving directory >>>> `/opt/buildtool-master/source/i486-unknown-linux-uclibc/linux/linux-i686 >>>> >>>> Google searches make me believe that this related to the BFS scheduler >>>> patch... >>>> >>>> I've posted results using the BFS scheduler before, which gave IMHO >>>> mixed results. >>>> So if BFS is really the reason for the error above, I'd ask if using BFS >>>> is worth the unability to update the 3.2.x kernel. >>>> >>>> Of course I may be wrong in my analysis. >>>> kp >>>> >>> Ok, we can comment bfs patch now if it causes troubles. Main reason for >>> bfs including - memory saving (near 300 kb) at no cost + possible better >>> userland performance. >> Hi Andrew; >> >> I found it's as easy as just to uncomment the bfs patch, the current >> configs has changes that has been apllied after we added the bfs scheduler. >> >> For 3.2.26 I've found that unset the kernel config option CONFIG NO_HZ >> (tickless kernel) could be a workaround (even though it is >> counter-productive, cause is degrades performance AFAIR), but with the >> new 3.2.27 kernel it became worse - the bfs patch won't apply cleanly >> any longer. >> >> Using older configs (commit 7be4217c5a01a95bb5feb437ef20d99c83e5624b) >> allows to compile a 3.2.27 kernel. >> >> Unfortunately I don't know how to apply the changes made by you for i486 >> (commit ffc8ade456633b1382a90d610e88d9d63af8bb81) and David for >> versatile (commit cc7ad52de6291364b9926739ba8cfee462017922). >> >> So some help is welcome. >> >> kp >> > Use 'git revert' to revert commits that chenge configs. > > Also commit cc7ad52 is related to uClibc, not kernel.
Sorry that was the wrong one, meant 48125c0fc10c84860995f38acd6a7eb7dbd95c02 Started reading about revert, and yes git is still too powerful for me, not shure how it can be accomplished to revert commit that adds bfs to the configs, while keeping the changes (esp. for i486 and versatile) that has been made later to the config files. kp ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
