View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0000393fileGeneralpublic2022-10-09 05:412022-10-09 16:48
Reporterredshift Assigned Tochristos  
PrioritynormalSeverityminorReproducibilityalways
Status resolvedResolutionfixed 
Product Version5.43 
Summary0000393: dsd64-dsf / JW07022A.mp3 tests failing with musl libc
DescriptionIn https://bugs.astron.com/view.php?id=382 an issue was reported for compiling file 5.43 against musl libc. I confirmed that finding. When I add the patch committed for issue 382 (https://github.com/file/file/commit/1294029cdb18d4c0997f2b52df435076b8444137) I don't get success but a different failure on the same test:

```
Running test: ../tests/dsd64-dsf.testfile
../tests/dsd64-dsf.testfile: DSF audio bitstream data mono, simple-rate, 1 bit, 141184 samples
lt-test: ERROR: result was (len 65)
DSF audio bitstream data mono, simple-rate, 1 bit, 141184 samples
expected (len 94)
DSF audio bitstream data, 1 bit, mono, "DSD 64" 2822400 Hz, no compression, ID3 version 2.3.0
```

I used git bisect to check whether this is fixed by any later commit up to HEAD, and it's not, but there's one interesting thing - as soon as I hit the second-to-latest commit (https://github.com/file/file/commit/a960a2adb1c41ecb5b996390f2484035878c02f6) I get a different failure on a different test:

```
Running test: ../tests/JW07022A.mp3.testfile
../tests/JW07022A.mp3.testfile: Audio file with ID3 version 2.2.0, contains:MPEG ADTS, layer III, v1, 96 kbps, 44.1 kHz, Monaural
lt-test: ERROR: result was (len 97)
Audio file with ID3 version 2.2.0, contains:MPEG ADTS, layer III, v1, 96 kbps, 44.1 kHz, Monaural
expected (len 99)
Audio file with ID3 version 2.2.0, contains: MPEG ADTS, layer III, v1, 96 kbps, 44.1 kHz, Monaural
```

I suspected the test system stops on the first failure, so I removed the tests that run before dsd64-dsf to test whether it still fails, and it does. So, commit a960a2adb1 is adding additional failures, in this scenario.

These failures are reproducible every time for me.
Steps To ReproduceBuild against musl libc; I'm on x86-64. I'm building for Void Linux which uses this build template - https://github.com/void-linux/void-packages/blob/master/srcpkgs/file/template - which does a fairly standard configure, make, make check, make install. Here's the full configure:

./configure --prefix=/usr --sysconfdir=/etc --sbindir=/usr/bin --bindir=/usr/bin --mandir=/usr/share/man --infodir=/usr/share/info --localstatedir=/var --host=x86_64-unknown-linux-musl --build=x86_64-unknown-linux-musl --libdir=${exec_prefix}/lib64 --enable-static --disable-libseccomp
TagsNo tags attached.

Activities

christos

2022-10-09 16:48

manager   ~0003824

It's all been fixed now, thanks! It had nothing to do with muslc, it was broken magic and broken code.

Issue History

Date Modified Username Field Change
2022-10-09 05:41 redshift New Issue
2022-10-09 16:47 christos Assigned To => christos
2022-10-09 16:47 christos Status new => assigned
2022-10-09 16:48 christos Status assigned => resolved
2022-10-09 16:48 christos Resolution open => fixed
2022-10-09 16:48 christos Note Added: 0003824