[ewg] [GIT PULL compat-rdma] qib for OFED 4.8

Thu Jun 29 11:20:29 PDT 2017

All,

I do object. I do not see the sense in releasing OFED 4.8 where it completely breaks the functionality of the previous OFED release. I'd like to see the issues fixed before the release of OFED 4.8.

I understand that there is a bandwidth issue from the maintainer. How could I be of assistance in this regard? Given, that I not up to speed on the SRP codebase or setup to debug it properly. But, that could be changed with some assistance. Vlad, what area of the code should I be looking at?

Thanks,
Marty

-----Original Message-----
From: Davis, Arlin R [mailto:arlin.r.davis at intel.com] 
Sent: Thursday, June 29, 2017 2:10 PM
To: Davis, Arlin R <arlin.r.davis at intel.com>; Vladimir Sokolovsky <vlad at dev.mellanox.co.il>; Bart Van Assche <Bart.VanAssche at wdc.com>; rsdance at soft-forge.com; Hanania, Amir <amir.hanania at intel.com>; Woodruff, Robert J <robert.j.woodruff at intel.com>; Marty Schlining <mschlining at ddn.com>
Cc: Mike Davis <mdavis at ddn.com>; ewg at lists.openfabrics.org; Cedric Fernandes <cfernandes at ddn.com>
Subject: RE: [ewg] [GIT PULL compat-rdma] qib for OFED 4.8

All,

Here is a quick summary of OFED 4.8 and SRP blocking bugs. Please correct/comment if I have it wrong.

- SL7.2 and SL7.3 SRP support is a must for DDN, need latest upstream SRP 4.11 fixes
- No resources available to backport or integrate the SRP 4.11 base or fixes into OFED 4.8
- ib_srp_backport (4.11 base) installs/builds/runs on OFED 4.8 for both 7.2 and 7.3
- SL7.3+OFED+srp_backport fixes 2632
- SL7.2+OFED+srp_backport still hitting bug 2632
- SL7.2+inbox_infiniband+srp_backport (no OFED) still hitting 2632 and 2634 (slightly different but still seeing CM DREQ)
- Bug 2632, 2634 still open, no resolution and no ETA on fix from maintainer

Given that we still have the same issues without OFED installed and no ETA on fixes, I would suggest moving forward and releasing OFED 4.8 GA. We can document issues and process for applying upstream SRP fixes in the release notes. 

If there are no objections, I would like Vlad to modify OFED release notes, add the 2 remaining bugs to known issues, and add backport procedure for SRP upstream as follow:

http://bugs.openfabrics.org/show_bug.cgi?id=2632
http://bugs.openfabrics.org/show_bug.cgi?id=2634

The ib_srp-backport procedure is as follows:
 1. Install the OS (RHEL, SLES, ...).
 2. Install OFED.
 3. Install the ib_srp-backport driver:
 	git clone https://github.com/bvanassche/ib_srp-backport
 	cd ib_srp-backport
 	make rpm
 	sudo rpm -U $PWD/rpmbuilddir/RPMS/*/*.rpm

We can then release GA package tomorrow with these changes and latest daily build.
Any other release notes changes?

Thanks,

Arlin

> -----Original Message-----
> From: ewg [mailto:ewg-bounces at lists.openfabrics.org] On Behalf Of 
> Davis, Arlin R
> Sent: Friday, June 23, 2017 12:59 PM
> To: Vladimir Sokolovsky <vlad at dev.mellanox.co.il>; Bart Van Assche 
> <Bart.VanAssche at wdc.com>; rsdance at soft-forge.com; Hanania, Amir 
> <amir.hanania at intel.com>; Woodruff, Robert J 
> <robert.j.woodruff at intel.com>; mschlining at ddn.com
> Cc: mdavis at ddn.com; ewg at lists.openfabrics.org; cfernandes at ddn.com
> Subject: Re: [ewg] [GIT PULL compat-rdma] qib for OFED 4.8
> 
> It seems like we now have a way to apply the latest SRP upstream fixes 
> to OFED 4.8, tested with RH 7.2 and RH7.3. However, we still have 2 
> critical bugs
> (1 new one) on RH 7.2 even after the upstream fixes are applied. RH 
> 7.3+OFED4.8+srp_backport seems to be fine.
> 
> http://bugs.openfabrics.org/show_bug.cgi?id=2632
> http://bugs.openfabrics.org/show_bug.cgi?id=2634
> 
> How does DDN want to proceed?
> Is it acceptable to apply upstream version/fixes to OFED 4.8 as 
> follow, document in release notes?
> Can you tell us if the remaining bugs are related to OFED 4.8 or if 
> RH7.2+infiniband+srp_backport has the same issues?
> 
> The ib_srp-backport procedure is as follows:
> 1. Install the OS (RHEL, SLES, ...).
> 2. Install OFED.
> 3. Install the ib_srp-backport driver:
> 	git clone https://github.com/bvanassche/ib_srp-backport
> 	cd ib_srp-backport
> 	make rpm
> 	sudo rpm -U $PWD/rpmbuilddir/RPMS/*/*.rpm
> 
> Thanks,
> 
> Arlin
> 
> > SRP is included already in OFED-4.8 as a part of linux-4.8 kernel 
> > and I don't see any reason to include SRP code twice.
> > If some bug fixes have to be added there then there is a specific 
> > place in compat-rdma git tree for the relevant patches.
> > If somebody is interested to add an additional SRP standalone 
> > package then please send me the relevant patches for 
> > compat-rdma/ofed_scripts and build git trees.
> >
> > Regards,
> > Vladimir
> >
> >
> > On 06/20/2017 09:01 PM, Davis, Arlin R wrote:
> > > Ok. Vlad, what would it take to get this backported version of SRP 
> > > into OFED
> > 4.8?
> > >
> > >> -----Original Message-----
> > >> From: Bart Van Assche [mailto:Bart.VanAssche at wdc.com]
> > >> Sent: Tuesday, June 20, 2017 10:49 AM
> > >> To: vlad at dev.mellanox.co.il; rsdance at soft-forge.com; Davis, Arlin 
> > >> R <arlin.r.davis at intel.com>; Hanania, Amir 
> > >> <amir.hanania at intel.com>; Woodruff, Robert J 
> > >> <robert.j.woodruff at intel.com>; mschlining at ddn.com
> > >> Cc: ewg at lists.openfabrics.org; cfernandes at ddn.com; mdavis at ddn.com
> > >> Subject: Re: [ewg] [GIT PULL compat-rdma] qib for OFED 4.8
> > >>
> > >> Hello Arlin,
> > >>
> > >> Adding a URL to the OFED README is not the same as including a 
> > >> version of the ib_srp-backport driver in OFED. How can OFED users 
> > >> know what version of the ib_srp-backport driver has been tested 
> > >> in combination with which OFED version?
> > >>
> > >> Thanks,
> > >>
> > >> Bart.
> > >>
> > >> On Tue, 2017-06-20 at 17:18 +0000, Davis, Arlin R wrote:
> > >>> Marty,
> > >>>
> > >>> Would this work for DDN? We could add the URL and the SRP 
> > >>> backport readme to the OFED release notes. I believe you had an 
> > >>> issue when applying the backport to SL 7.2 and OFED 4.8 RC4 but 
> > >>> I think Bart fixed that
> > >> already.
> > >>> Thanks, Arlin
> > >>>
> > >>>> Hello Arlin and Robert,
> > >>>>
> > >>>> A backported version of the 4.11 ib_srp driver that already has 
> > >>>> been tested is available at 
> > >>>> https://github.com/bvanassche/ib_srp-
> > backport.
> > >>>> That driver builds fine against kernel.org kernels, RHEL/CentOS 
> > >>>> kernels, openSuSE and SLES kernels and also against MOFED and OFED.
> > >>>> Has it already been considered to use that code base instead of 
> > >>>> duplicating the backporting effort of the ib_srp driver?
> > >>>>
> > >>>> Thanks,
> > >>>>
> > >>>> Bart.
> 
> _______________________________________________
> ewg mailing list
> ewg at lists.openfabrics.org
> http://lists.openfabrics.org/mailman/listinfo/ewg