[Lustre-devel] oldest lustre supported SUSE kernel

tao.peng at emc.com tao.peng at emc.com
Fri Mar 9 03:32:17 PST 2012


> -----Original Message-----
> From: Andreas Dilger [mailto:adilger at whamcloud.com]
> Sent: Friday, March 09, 2012 3:37 PM
> To: Peng, Tao
> Cc: lustre-devel at lists.lustre.org; wc-discuss at whamcloud.com
> Subject: Re: [Lustre-devel] oldest lustre supported SUSE kernel
> 
> On 2012-03-08, at 10:55 PM, <tao.peng at emc.com> <tao.peng at emc.com> wrote:
> >> Andreas Dilger <adilger at whamcloud.com> wrote:
> >> On 2012-02-06, at 2:53, <tao.peng at emc.com> wrote:
> >>> As I know, the oldest lustre supported RHEL kernel is 2.6.18-latest. What is the oldest SUSE
> kernel?
> >>
> >> 2.6.32-latest?
> >>
> >> That depends on which version of Lustre you are asking about, and whether you are asking about the
> client or server.
> >>
> >> For Lustre 2.2+ servers it will be SLES11 SP1 (2.6.32), since it matches RHEL6 (2.6.32).
> >>
> >> For clients it is RHEL5 (2.6.18) and SLES11 SP0 (2.6.27).
> >
> > We found that latest master branch doesn't build client for SLES11 SP0 (2.6.27). Does it mean master
> has moved on dropping SLES11 SP0 or it should be fixed?
> 
> This looks like the page_mkwrite() prototype changing from "struct page *" (which is in kernels before
> 2.6.29) to "struct vm_fault *" (which is used in later kernels).  I don't have a SLES11 SP0 kernel
> available to check which one it is using, but it should be the "struct page *" based on the kernel
> version.  This should already be handled by the check for HAVE_VM_OP_FAULT, but possibly this is
> broken in newer versions of the SP0 kernel?
> 
> It often happens that vendor kernels backport code/APIs from newer kernels in order to make porting
> other drivers/filesystem code easier.
> 
> As you can tell, we do not build the SLES11 SP0 client ourselves automatically, only the SP1 client,
> but if the effort is not huge it would be preferable to keep this client working.  We typically try to
> keep the range of working clients as large as possible, since administrators may not have as much
> control over the running clients as they do over the servers.
> 
> I would guess (without having looked at the SP0 kernel) that it may just be possible to change
> lustre/autoconf/lustre-core.m4 to check this prototype, and move the two ll_page_mkwrite() helpers
> into their own "#ifdef HAVE_PAGE_MKWRITE_VM_FAULT" section separate from HAVE_VM_OP_FAULT.
Thanks for confirming that it is still supported. And I verified it can be fixed as you described. I created LU-1200 and attached the fix there.

Cheers,
Tao
> 
> If this analysis is incorrect, I apologize, but I'm travelling and unable to download the whole kernel
> at this time.
> 
> Cheers, Andreas
> 
> > From: Zhang, Kelly
> > Sent: Thursday, March 08, 2012 6:44 PM
> > To: Peng, Tao
> > Cc: China COE FastData
> > Subject: RE: Conversation with Zhang, Kelly
> >
> > The latest Lustre code also fails on suse11SP0(kernel version  linux-2.6.27.19-5.1), when building
> Lustre:
> > cc1: warnings being treated as errors
> > /home/master/kernel/rpmbuild/BUILD/lustre-2.2.50/lustre/llite/llite_mmap.c:585: error:
> initialization from incompatible pointer type
> > make[8]: *** [/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50/lustre/llite/llite_mmap.o] Error 1
> > make[7]: *** [/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50/lustre/llite] Error 2
> > make[6]: *** [/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50/lustre] Error 2
> > make[5]: *** [_module_/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50] Error 2
> > make[5]: Leaving directory `/var/tmp/kernel-source-2.6.27.19-build/usr/src/linux-2.6.27.19-5.1'
> > make[4]: *** [modules] Error 2
> > make[4]: Leaving directory `/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50'
> > make[3]: *** [all-recursive] Error 1
> > make[3]: Leaving directory `/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50'
> > make[2]: *** [all] Error 2
> > make[2]: Leaving directory `/home/master/kernel/rpmbuild/BUILD/lustre-2.2.50'
> > error: Bad exit status from /var/tmp/rpm-tmp.76807 (%build)
> >
> 
> 
> Cheers, Andreas
> --
> Andreas Dilger                       Whamcloud, Inc.
> Principal Lustre Engineer            http://www.whamcloud.com/
> 
> 
> 
> 



More information about the lustre-devel mailing list