speculative fix to 32bit failures: set vrf_id_t to 64 bits

[louberger]

No description provided.

[LabN-CI]

💚 Basic BGPD CI results: SUCCESS, 0 tests failed

Results table

_	_
Result	SUCCESS git merge/1876 a44db78
Date	03/12/2018
Start	20:17:15
Finish	20:40:25
Run-Time	23:10
Total	1813
Pass	1813
Fail	0
Valgrind-Errors	0
Valgrind-Loss	0
Details	vncregress-2018-03-12-20:17:15.txt
Log	autoscript-2018-03-12-20:17:55.log.bz2

For details, please contact louberger

[NetDEF-CI]

Continuous Integration Result: SUCCESSFUL

Congratulations, this patch passed basic tests

Tested-by: NetDEF / OpenSourceRouting.org CI System

CI System Testrun URL: https://ci1.netdef.org/browse/FRR-FRRPULLREQ-2859/

This is a comment from an EXPERIMENTAL automated CI system.
For questions and feedback in regards to this CI system, please feel free to email
Martin Winter - mwinter (at) opensourcerouting.org.

---

Warnings Generated during build:

Checkout code: Successful with additional warnings:

Report for bgp_zebra.c
===============================================
< WARNING: line over 80 characters
< #419: FILE: /tmp/f1/bgp_zebra.c:419:
< +		zlog_debug("Rx Intf neighbor add VRF %llu IF %s addr %s", vrf_id,
< 
--
< WARNING: line over 80 characters
< #448: FILE: /tmp/f1/bgp_zebra.c:448:
< +		zlog_debug("Rx Intf neighbor del VRF %llu IF %s addr %s", vrf_id,
< 
< WARNING: line over 80 characters
< #480: FILE: /tmp/f1/bgp_zebra.c:480:
< +		zlog_debug("Rx Intf VRF change VRF %llu IF %s NewVRF %llu", vrf_id,
< 
--
< WARNING: line over 80 characters
< #1184: FILE: /tmp/f1/bgp_zebra.c:1184:
< +		zlog_debug("Tx route %s VRF %llu %s metric %u tag %" ROUTE_TAG_PRI
< 
< WARNING: line over 80 characters
< #1281: FILE: /tmp/f1/bgp_zebra.c:1281:
< +		zlog_debug("Tx route delete VRF %llu %s", peer->bgp->vrf_id, buf);
< 
Report for if.h
===============================================
< WARNING: line over 80 characters
< #302: FILE: /tmp/f1/if.h:302:
< +			"name exists already in VRF %llu!",                      \
< 
--
< WARNING: line over 80 characters
< #309: FILE: /tmp/f1/if.h:309:
< +			"name doesn't exist in VRF %llu!",                       \
< 
--
< WARNING: line over 80 characters
< #316: FILE: /tmp/f1/if.h:316:
< +			"ifindex exists already in VRF %llu!",                   \
< 
--
< WARNING: line over 80 characters
< #323: FILE: /tmp/f1/if.h:323:
< +			"ifindex doesn't exist in VRF %llu!",                    \
< 
Report for ospf_zebra.c
===============================================
< WARNING: line over 80 characters
< #78: FILE: /tmp/f1/ospf_zebra.c:78:
< +		zlog_debug("Zebra rcvd: router id update %s vrf %s id %llu", buf,
< 
Report for zclient.c
===============================================
< WARNING: line over 80 characters
< #488: FILE: /tmp/f1/zclient.c:488:
---
> #488: FILE: /tmp/f2/zclient.c:488:
Report for zebra.h
===============================================
< WARNING: do not add new typedefs
< #499: FILE: /tmp/f1/zebra.h:499:
< +typedef unsigned long long vrf_id_t;
< 
Report for zebra_pw.c
===============================================
< WARNING: line over 80 characters
< #93: FILE: /tmp/f1/zebra_pw.c:93:
< +		zlog_debug("%llu: deleting pseudowire %s protocol %s", pw->vrf_id,
< 
Report for zebra_rnh.c
===============================================
< WARNING: line over 80 characters
< #193: FILE: /tmp/f1/zebra_rnh.c:193:
< +		zlog_debug("%llu: Client %s registers for RNH %s type %d", vrf_id,
---
--
< WARNING: line over 80 characters
< #771: FILE: /tmp/f1/zebra_rnh.c:771:
< +		zlog_debug("%llu:%s: Evaluate RNH, type %d %s", vrfid, bufn, type,
< 
Report for zebra_vxlan.c
===============================================
< WARNING: line over 80 characters
< #6697: FILE: /tmp/f1/zebra_vxlan.c:6697:
< +		zlog_err("EVPN VNI Adv for non-default VRF %llu", zvrf_id(zvrf));
< 

---

CLANG Static Analyzer Summary

• Github Pull Request 1876, comparing to Git base SHA 9d16566

No Changes in Static Analysis warnings compared to base

19 Static Analyzer issues remaining.

See details at
https://ci1.netdef.org/browse/FRR-FRRPULLREQ-2859/artifact/shared/static_analysis/index.html

[louberger]

@mwinter-osr @rwestphal this seems to fix the 32 bit issues.

[louberger]

@donaldsharp this is a bit of a brute force solution to this issue. I have no idea if 64bit vrf_id_t is really needed or not... but it works

[donaldsharp]

Can we get a bit more explanation about what is going on? It's not clear at all.

Additionally this change will clearly break passing of the vrf id through zapi. And this should be fixed.

[LabN-CI]

💚 Basic BGPD CI results: SUCCESS, 0 tests failed

Results table

_	_
Result	SUCCESS git merge/1876 5255858
Date	03/12/2018
Start	21:32:14
Finish	21:55:30
Run-Time	23:16
Total	1816
Pass	1816
Fail	0
Valgrind-Errors	0
Valgrind-Loss	0
Details	vncregress-2018-03-12-21:32:14.txt
Log	autoscript-2018-03-12-21:32:53.log.bz2

For details, please contact louberger

[NetDEF-CI]

Continuous Integration Result: SUCCESSFUL

Congratulations, this patch passed basic tests

Tested-by: NetDEF / OpenSourceRouting.org CI System

CI System Testrun URL: https://ci1.netdef.org/browse/FRR-FRRPULLREQ-2860/

This is a comment from an EXPERIMENTAL automated CI system.
For questions and feedback in regards to this CI system, please feel free to email
Martin Winter - mwinter (at) opensourcerouting.org.

---

Warnings Generated during build:

Checkout code: Successful with additional warnings:

Report for zebra.h
===============================================
< WARNING: do not add new typedefs
< #499: FILE: /tmp/f1/zebra.h:499:
< +typedef unsigned long long vrf_id_t;
< 

---

CLANG Static Analyzer Summary

• Github Pull Request 1876, comparing to Git base SHA fac615f

No Changes in Static Analysis warnings compared to base

19 Static Analyzer issues remaining.

See details at
https://ci1.netdef.org/browse/FRR-FRRPULLREQ-2860/artifact/shared/static_analysis/index.html

[pguibert6WIND]

I am also against that solution.
I will look at the i386 issue

[pguibert6WIND]

from first observations, it seems netns from kernel behaves strangely on i386.
Don't know yet if this is due to my setup or not, but what is dumped here looks similar to what I observe.

to make it simple, setns() is not handled.
consequently: zebra_vrf is not enabled , but vrf library sees it as enabled.

as first patch, I will try to enforce the vrf library by enabling it only if setns() returns ok.