[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#877956: busybox: Not all testsuite failures are fatal

Source: busybox
Version: 1:1.27.2-1
Tags: upstream patch
Forwarded: http://lists.busybox.net/pipermail/busybox/2017-October/085868.html

Currently there is a bug in the testsuite driver where running an
old-style test resets the current failure flag to 0, so not all failures
become fatal. The above patch forwarded upstream fixes this.

Note: Some of the `du` tests seem to be dependent on the filesystem in
use (or other system-specific configuration). Builds on abel.debian.org,
exodar.debian.net and {falla,fischer}.debian.org all fail the
"du-m-works" test, which seems to be because any indirect blocks used
are counted towards the number of blocks reported in use by stat (abel
uses ext3, exodar uses ext2 and falla/fischer use UFS), pushing the
total number of bytes on disk over 1 MiB for the 1 MiB file. Presumably
the others are all fine because they are using ext4 and require no more
than 4 extents, all of which fit in the inode. Additionally, grepping
the build logs for the latest upload shows some issues with "du-k-works"
and "du-l-works" on powerpc and ppc64el which I could not reproduce on
the porterboxes:

> jrtc27@wuiet:~$ bzgrep FAIL: /srv/buildd.debian.org/db/b/busybox/1\:1.27.2-1/*.bz2
> /srv/buildd.debian.org/db/b/busybox/1:1.27.2-1/powerpc_1505669728_log.bz2:FAIL: du-k-works
> /srv/buildd.debian.org/db/b/busybox/1:1.27.2-1/powerpc_1505669728_log.bz2:FAIL: du-l-works
> /srv/buildd.debian.org/db/b/busybox/1:1.27.2-1/ppc64el_1505669547_log.bz2:FAIL: du-k-works
> /srv/buildd.debian.org/db/b/busybox/1:1.27.2-1/ppc64el_1505669547_log.bz2:FAIL: du-l-works


Reply to: