Source: texlive-bin Version: 2018.20180416.47457-1 Severity: serious Tags: sid buster patch Hi, texlive-bin FTBFS on mips*, alpha, ia64, and powerpc with the m-tx tests failing: > ============================================ > m-tx (TeX Live) 0.63: ./test-suite.log > ============================================ > > # TOTAL: 1 > # PASS: 0 > # SKIP: 0 > # XFAIL: 0 > # FAIL: 1 > # XPASS: 0 > # ERROR: 0 > > .. contents:: :depth: 2 > > FAIL: m-tx > ========== > > #! /bin/sh -vx > # Copyright 2017 Karl Berry <tex-live@tug.org> > # Copyright 2012 Peter Breitenlohner <tex-live@tug.org> > # You may freely use, modify and/or distribute this file. > > rm -f mozart* > + rm -f mozart* > > cp $srcdir/tests/mozart.mtx . > + cp ../../../utils/m-tx/tests/mozart.mtx . > > set -x > + set -x > > ./prepmx mozart \ > && diff mozart.pmx $srcdir/tests/mozart.pmx \ > || exit 1 > + ./prepmx mozart > ==> This is M-Tx 0.63 (Music from TeXt) <7 January 2018> > ==>> Input from file mozart.mtx > Writing to mozart.pmx > You have not defined Meter, assuming "C" : WARNING in preamble > No range defined for voice RH: WARNING in preamble > No range defined for voice LH: WARNING in preamble > PrePMX done. Now run PMX. > + diff mozart.pmx ../../../utils/m-tx/tests/mozart.pmx > 8c8 > --- >> 2 -1 2 4 4 0 6 0.00000 0 1 1 20 0 > + exit 1 > FAIL m-tx.test (exit status: 1) > > ============================================================================ > Testsuite summary for m-tx (TeX Live) 0.63 > ============================================================================ > # TOTAL: 1 > # PASS: 0 > # SKIP: 0 > # XFAIL: 0 > # FAIL: 1 > # XPASS: 0 > # ERROR: 0 > ============================================================================ > See ./test-suite.log > Please report to tex-k@tug.org > ============================================================================ I believe this bug is fixed in m-tx 0.63a which is available on CPAN: https://ctan.org/tex-archive/support/m-tx texlive currently only has version 0.63 which is affected by this bug. The underlying bug seems to be a buffer overflow which (naturally) invokes undefined behaviour and causes the tests to break on some architectures. I've attached a patch taken from 0.63a which fixes this for me. Thanks, James
--- a/utils/m-tx/mtx-src/preamble.c 2018-01-09 14:14:43.000000000 +0000 +++ b/utils/m-tx/mtx-src/preamble.c 2018-04-08 12:23:37.000000000 +0100 @@ -801,8 +801,8 @@ stave_size[i] = unspec; for (i = 0; i <= maxstaves; i++) nspace[i] = unspec; - nspace[i] = unspec; - stave_size[i-1] = unspec; + /* next line seems to be spurious. 0.63a RDT */ + /* begin nspace[i]:=unspec; stave_size[i]:=unspec; end; */ n_pages = 1; n_systems = 1; readStyles(); --- a/utils/m-tx/mtx-src/preamble.pas 2018-01-09 14:14:43.000000000 +0000 +++ b/utils/m-tx/mtx-src/preamble.pas 2018-04-08 12:21:23.000000000 +0100 @@ -501,7 +501,8 @@ for i:=1 to maxvoices do setVocal(i,false); for i:=1 to maxstaves do stave_size[i]:=unspec; for i:=0 to maxstaves do nspace[i]:=unspec; - begin nspace[i]:=unspec; stave_size[i]:=unspec; end; + { next line seems to be spurious. 0.63a RDT } + { begin nspace[i]:=unspec; stave_size[i]:=unspec; end; } n_pages:=1; n_systems:=1; readStyles; old_known_styles := known_styles; for i:=1 to lines_in_paragraph do omit_line[i]:=false;
Attachment:
signature.asc
Description: OpenPGP digital signature