[LU-84] Allow building with quota support against kernel devel headers Created: 11/Feb/11 Updated: 04/Apr/11 Resolved: 04/Apr/11 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.1.0 |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Minor |
| Reporter: | Ned Bass | Assignee: | Zhenyu Xu |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Chaos 4.4 |
||
| Rank (Obsolete): | 5089 |
| Description |
|
The configure script now treats 64-bit quota limits as mandatory. In 1.8, this check continued with limited quota support if the check failed. For CHAOS systems we build Lustre against the kernel development headers, not full kernel source. Therefore the 64-bit quota limits check is guaranteed to fail |
| Comments |
| Comment by Ned Bass [ 11/Feb/11 ] |
|
See http://review.whamcloud.com/231 (obsolete) |
| Comment by Andreas Dilger [ 11/Feb/11 ] |
|
How is it possible to build Lustre servers against the development headers instead of the full kernel source? The ldiskfs code depends on access to fs/ext4 in order to be built. If this is for the client, then the quota headers should not be necessary at all. |
| Comment by Ned Bass [ 11/Feb/11 ] |
|
We build ldiskfs in the CHAOS kernel and use the --with-ldiskfs-inkernel configure option. |
| Comment by Christopher Morrone [ 11/Feb/11 ] |
|
Andreas, LLNL has been building ldiskfs as part of the kernel for years. (Not that I am entirely fond of that solution...) With --with-ldiskfsinkernel (or whatever it is), the fs/ext4 kernel source is not necessary. |
| Comment by Peter Jones [ 14/Feb/11 ] |
|
Bobijam I suggest that you discuss this issue with Johann when he joins us this week Regards Peter |
| Comment by Zhenyu Xu [ 18/Feb/11 ] |
|
Copy review comment from Johann: Thanks for the patch Ned. Actually, the quota patch we pushed upstream through Jan Kara introduces QFMT_VFS_V1 in include/linux/quota.h. This was needed because OCFS2 quota required a special identifier (i.e. QFMT_OCFS2). To detect 64-bit quota support, i think we should just check for QFMT_VFS_V1 in include/linux/quota.h. The good news is that RHEL6 uses the same upstream patch, so QFMT_VFS_V1 is present in include/linux/quota.h. The only problem is our old version of the RHEL5 patch which needs to be realigned with what was pushed upstream. We still have a compile problem since the v2_disk_dqinfo & v2_disk_dqheader structs are unfortunately used by fsfilt (that's because lustre creates the quota files on disk by itself, which is really wrong IMO, but that's another topic). Copying those 2 structs to the lustre source as done in this patch is ok as a short-term workaround IMO, until we fix the real problem. |
| Comment by Ned Bass [ 23/Feb/11 ] |
|
On Fri, Feb 18, 2011 at 02:44:34AM -0800, Johann Lombardi (Code Review) wrote: Oops, I missed that somehow. > The good news is that RHEL6 uses the same upstream patch, so QFMT_VFS_V1 is I'll work on an updated patch that performs the old test if we can find Thanks, |
| Comment by Ned Bass [ 03/Mar/11 ] |
|
New version submitted to address Johann's comments: |
| Comment by Build Master (Inactive) [ 03/Mar/11 ] |
|
Integrated in Ned Bass : f68ef518bc5022589abef729b0f7984c963c5607
|
| Comment by Build Master (Inactive) [ 11/Mar/11 ] |
|
Integrated in Brian J. Murrell : 86cc82b3abf5269be858e33e91eb06ee49324d41
|
| Comment by Build Master (Inactive) [ 16/Mar/11 ] |
|
Integrated in Brian J. Murrell : 9c511c3b609fbe376bae3051e32de9170497081e
|
| Comment by Build Master (Inactive) [ 16/Mar/11 ] |
|
Integrated in Brian J. Murrell : 9c511c3b609fbe376bae3051e32de9170497081e
|
| Comment by Build Master (Inactive) [ 16/Mar/11 ] |
|
Integrated in Brian J. Murrell : 9c511c3b609fbe376bae3051e32de9170497081e
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Bobi Jam : 2e42217fb7f3501847ce759a3b33ad62d6bb3918
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Bobi Jam : 2e42217fb7f3501847ce759a3b33ad62d6bb3918
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Bobi Jam : 178ea9b28519a6f6fbe4b350a7f1d8fa2750a055
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Bobi Jam : 178ea9b28519a6f6fbe4b350a7f1d8fa2750a055
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Brian J. Murrell : ac3c214e4e6f3100210b8e457d057590496938b0
|
| Comment by Build Master (Inactive) [ 18/Mar/11 ] |
|
Integrated in Brian J. Murrell : ac3c214e4e6f3100210b8e457d057590496938b0
|
| Comment by Build Master (Inactive) [ 25/Mar/11 ] |
|
Integrated in Ned Bass : 22d92bda573fb3f071d65c5ffc0a857612cfbbf1
|
| Comment by Build Master (Inactive) [ 28/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Build Master (Inactive) [ 31/Mar/11 ] |
|
Integrated in Oleg Drokin : 560ade217a6de8960792b776ea0417e38c41192f
|
| Comment by Peter Jones [ 04/Apr/11 ] |
|
This appears to have been landed to master. Please reopen if any further work is required. |