On 2 May 2017 at 13:10, Graham Inggs <gin...@debian.org> wrote:
> I've been unable to reproduce this failure on asachi.debian.org.
> It built without failure ten times yesterday, and again once today.

I requested a give back by mailing debian-wb-team, but it hasn't happened yet.
If it is successful, it completes the llvm-symver transition, and if
not, it provides us with more data for determining whether the problem
lies with different hardware or with different kernel versions.
Either way, it shouldn't change the status of this bug.

> There is definitely something wrong though, according to the julia arm64
> build logs [1], the build has failed in the same way since January every
> time it was attempted on arm-arm-04.
>
> The most recent success a week ago was on arm-arm-03 and against
> llvm-3.8/1:3.8.1-22 which should have included the versioned symbols.

Looking at recent (>= December 2016) build success and failures [1]
combined with machine information [2]:

Always succeeds on:
arm-linaro-01 APM X-Gene Mustang board
arm-arm-01 Juno Versatile Express board (ARMv8-A)
arm-linaro-03 APM X-Gene Mustang board
arm-arm-03 APM X-Gene Mustang board
asachi.debian.org APM X-Gene Mustang board

Always fails on:
arm-arm-04 AMD Seattle rev B1
arm-conova-01 ?

Looking at kernel versions from the build logs:

Always succeeds on:
Linux 3.16.0-4-arm64 armhf (aarch64)

Always fails on:
Linux 4.8.0-0.bpo.2-arm64 arm64 (aarch64)
Linux 4.9.0-0.bpo.2-arm64 arm64 (aarch64)
Linux 4.9.0-2-arm64 arm64 (aarch64)


[1] https://buildd.debian.org/status/logs.php?pkg=julia&arch=arm64
[2] https://db.debian.org/machines.cgi

Reply via email to