[ofa-general] OFED, the backported <linux/scatterlist.h> header and sg_init_table()

[Jon Mason]

On Wed, May 06, 2009 at 08:30:55AM +0300, Jack Morgenstein wrote:
> On Tuesday 05 May 2009 18:06, Jon Mason wrote:
> > No, we currently duplicate all the scatterlist functionality.  Including
> > ncrypto.h would greatly simplify the backport headers, but it is a
> > RHEL5.2/5.3 only solution.  If this change is needed for all other
> > backports, then a better solution will be needed.
> > 
> Each backport has its OWN directory.  The backports are not identical
> for all kernels.  There is absolutely no problem with handling backports
> per kernel/per distribution. Therefore, the RHEL 5.2/5.3 solution can be
> used for those backports alone, without affecting any of the others.
> Other backports will have a different change.

Yes, the point I was trying to make is that the fix I have will only
apply to RHEL5.  If a more sweeping change is needed, then something
else will need to be done.  I believe the root issue with the reported
was on RHEL5.3, so this will probably solve their problem unless they
need it for all OFED supported releases.
 
> For RHEL5.2/5.3, my concern is that if someone will actually write an ncrypto kernel
> application, and include ncrypto.h along with the infiniband headers, there will be
> compilation problems because the scatterlist functionality fixes will appear twice.

Excellent point.  The patch I just sent out should prevent this from
happening as well.

> Specifically, OFED 1.4.1 has the following INDIVIDUAL/independent backports, and
> each one is handled differently:
> 2.6.16
> 2.6.16_sles10
> 2.6.16_sles10_sp1
> 2.6.16_sles10_sp2
> 2.6.17
> 2.6.18
> 2.6.18-EL5.1
> 2.6.18-EL5.2
> 2.6.18-EL5.3
> 2.6.18_FC6 (also for EL5.0)
> 2.6.18_suse10_2
> 2.6.19
> 2.6.20
> 2.6.21
> 2.6.22
> 2.6.22_suse10_3
> 2.6.23
> 2.6.24
> 2.6.25
> 2.6.26
> 2.6.9_U4
> 2.6.9_U5
> 2.6.9_U6
> 2.6.9_U7
> 
> - Jack