scjody@clusterfs.com
2007-Apr-04 19:01 UTC
[Lustre-devel] [Bug 12052] build problems in b_new_cmd
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by using the following link: https://bugzilla.lustre.org/show_bug.cgi?id=12052 (In reply to comment #0)> It looks like this version has never been built with SuSE. > It has no ldiskfs patch series for SuSE.That is correct. You''re really breaking new ground here.> There appear to be 2 different instances of checking for show_task in configure, > one that is correct & one that is wrong. Don''t know the effect of having both.There is a check as part of the lnet build and again as part of the lustre build. Is that what you''re referring to? You don''t provide enough of the ./configure output for me to tell.> configure in this version now requires the linux kernel to be config''ed with > certain CRYPTO options enabled. We hadn''t been enabling these in our standard > kernel, but I worked around this by changing my kernel config for this test. > Are these kernel configs really needed? Why?We plan on including Kerberos support in 1.8, and some of this code already exists on b_new_cmd. This is likely why these options need to be on.> Once configure gets done, still won''t build. Seems to be failing during ldiskfs > build due to missing sources in linux name ima*.c Are these RHEL specific? Is > it OK to just delete them from new_sources list is lustre/ldiskfs/Makefile.in ?Absolutely not. IAM is the Index Access Module, part of ext3 used to store data for CMD. It is added by ext3-iam-separate.patch and there are also two other iam patches. You''ll have to add these to the SLES 10 ldiskfs series. There are also many other patches missing from the ldiskfs series found on HEAD, all of which are likely required. I''m not sure if the patches will apply to SLES 10''s ext3 code - unfortunately, this is unlikely but trying is the only way to know.