Alan Bartlett wrote: > On 08/03/2009, Farkas Levente <lfarkas at lfarkas.org> wrote: > >> now one last question/remark about the page: >> http://wiki.centos.org/HowTos/Custom_Kernel >> in sec 4. you wrote to comment out the >> "if a rhel kernel, apply the rhel config options" >> blocks, but for me it's not working. ie. in case of >> kernel-2.6.18-128.1.1.el5 if i comment out that part then i've got an >> error during kernel compile: >> ----------------------------------------- >> + for i in '*.config' >> + mv kernel-2.6.18-i686-PAE.config .config >> ++ head -1 .config >> ++ cut -b 3- >> + Arch=i386 >> + make ARCH=i386 nonint_oldconfig >> CONFIG_PATA_ISAPNP >> make[1]: *** [nonint_oldconfig] Error 1 >> make: *** [nonint_oldconfig] Error 2 >> error: Bad exit status from /var/tmp/rpm-tmp.88925 (%prep) >> ----------------------------------------- >> while if i leave it there than compile without any problem. anyway >> config-rhel-generic contains only option which are not set so why would >> be useful to comment out? >> so imho that part is not required what's more not working. >> yours. > > I am sorry but I do not see from where your problem arises. It is > possibly related to a language issue. > > What I will say is that the technique described has been necessary to > enable a user to build a customised CentOS kernel ever since the very > first kernel-2.6.18-8.el5 package was released. > > Will you please formulate your uncertainty into a question and then > post it to the CentOS-5 General Support forum. In that way my > colleague, good friend and co-maintainer of the article -- Akemi Yagi > (aka toracat) -- may also see what it is that concerns you. We may > then be able to clarify things. ok so let me clarify this. imho keep the diff from rh's spec as little as possible is a good goal. on page sec 4: http://wiki.centos.org/HowTos/Custom_Kernel you wrote to delete/comment out the "if a rhel kernel, apply the rhel config options" part which is imho unnecessary, what's more wrong. the file config-rhel-generic contains only commented out kernel config options (except the CONFIG_RAW_DRIVER=y). so these won't hurt anyone. and if someone add the same kernel config param to the resulting config then the first kernel config ie: make -s ARCH=$Arch nonint_oldconfig > /dev/null will ignore those commented out configs. so i don't really understand the reason for this advice since it won't hurt anyone but can cause errors (like in my case in the above kernel compile error). ps. anyway i also not understand why you prefer to modify rh's spec from: %define with_debug %{?_without_debug: 0} %{!?_without_debug: 1} to %define with_debug 0 instead of suggest to build with --without debug --without debuginfo? pss. i we discuss this page:-) for me it's totally the same, but why you prefer/suggest to use %define buildid .your_identifier over to add a suffix to the dist tag (while you also modify dist in centosplus kernels)? yours. -- Levente "Si vis pacem para bellum!"