I saw that the patch got merged yesterday evening. It still produces the same faillogs:
I don't know buildbot internals. If I look into:
https://buildbot.openwrt.org/master/images/builders/ipq806x%2Fgeneric/builds/722
it is referencing the latest changes including PKG_SYMVERS_DIR patch.
It might be that there has to be updated sth. other before (where buildbot is relying on; what is counting for all devices)?
It requires
first a new phase1 build round (kernel, images, SDK and imagebuilder)
http://buildbot.openwrt.org/master/images/grid
then a packages round using that SDK. And some packages archs share SDKs, or use a different SDK than some devices finally using those packages.
http://buildbot.openwrt.org/master/packages/grid
E.g. for ipq806x devices we need arm_cortex-a15_neon-vfpv4 packages that are built with armvirt32 SDK, meaning that there needs to be
I think that all targets / SDKs will be built in phase1 in the next few hours (2127accd441b... or later)
But building all packages may take 1-2 days, depending on the luck with packages build round starting with or without the fixed SDK.
Sadly it takes time to fully recover from this kind of problem in core of the build system logic.
Ps. note that this "symvers" problem may have hidden some other problem caused by the commits on last Friday, but hopefully that fixed bug was the only bug...
Looks like things got fixed:
lookign at the newest completed phase2 build (happens to be mipsel_24kc_24kf), the failurelogs look again "normal".
Just the usual failures...
https://downloads.openwrt.org/snapshots/faillogs/mipsel_24kc_24kf/
That makes me to think that other archs will recover, too. 
Ah, that looks good. I didn't know exactly how it is processed. Thx.
Based on faillogs it looks like package builds for almost all archs have now been completed, and the two last ones (powerpc_464fp and arm_cortex-a9) are being compiled right now.
Hopefully no other bugs surfaced for imagebuilder (and this symvers thing leading to missing packages was the bug causing the segment fault).
thankyou for representing the user base on the mailing list @hnyman your efforts were the missing link and are valued.
I couldn't agree with that @anon50098793 has written more. For a user without deeper knowledge about the processes and code it is very difficult to make a diagnose if sth. is failing. Because most likely the error is in front of the screen. 
Personally I'm scary to file bugs or write into mailing lists because of making possible false claims. So I'm thankfull if others with more expertise are doing this.
Looks like ImageBuilder is working normally now. I just built an image, flashed, and everything looks fine.
This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.