Bugs item #555, was opened at 2006-05-20 16:16 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=555&group_id=106 Category: hz Group: v1.30.0 Status: Closed Resolution: Accepted Priority: 5 Submitted By: Dag Leine (dag) Assigned to: Taco Hoekwater (taco) Summary: pdfLaTeX coredumps on OpenBSD Initial Comment: I've found, that pdfLaTeX coredumps on OpenBSD if I use amsmath and microtype in some math environments. Here is a minimal latex document for showing the bug:
cat minidump.tex \documentclass{article} \usepackage{amsmath} \usepackage{microtype} \setlength{\parindent}{0pt} \begin{document} \begin{align} x \end{align} \end{document} % end of minidump.tex
pdflatex minidump This is pdfeTeX, Version 3.141592-1.30.6-2.2 (Web2C 7.5.5) entering extended mode (./minidump.tex LaTeX2e <2003/12/01> Babel
and hyphenation patterns for american, french, german, ngerman, b ahasa, basque, bulgarian, catalan, croatian, czech, danish, dutch, esperanto, e stonian, finnish, greek, icelandic, irish, italian, latin, magyar, norsk, polis h, portuges, romanian, russian, serbian, slovak, slovene, spanish, swedish, tur kish, ukrainian, nohyphenation, loaded. (/usr/local/share/texmf-dist/tex/latex/base/article.cls Document Class: article 2004/02/16 v1.4f Standard LaTeX document class (/usr/local/share/texmf-dist/tex/latex/base/size10.clo)) (/usr/local/share/texmf-dist/tex/latex/amsmath/amsmath.sty For additional information on amsmath, use the `?' option. (/usr/local/share/texmf-dist/tex/latex/amsmath/amstext.sty (/usr/local/share/texmf-dist/tex/latex/amsmath/amsgen.sty)) (/usr/local/share/texmf-dist/tex/latex/amsmath/amsbsy.sty) (/usr/local/share/texmf-dist/tex/latex/amsmath/amsopn.sty)) (/usr/local/share/texmf-dist/tex/latex/microtype/microtype.sty (/usr/local/share/texmf-dist/tex/latex/graphics/keyval.sty) (/usr/local/share/texmf-dist/tex/latex/microtype/microtype.cfg)) (./minidump.aux) (/usr/local/share/texmf-dist/tex/latex/microtype/mt-cmr.cfg)Segmentation fault (core dumped)
A quick look with gdb shows, that it hangs in line 4396 of pdfetex2.c: c = getrpcode ( f , mem [p ].hh.b1 ) ; Here p=999996 and f=-4096 Since f seems to be a table index to pdffontrbase a negative value does not make sense to my mind. I've tryed the same on a LINUX machine. There pdfLaTeX does not crash, but p and f have the same values as under OpenBSD. As the most bad work around after taking *Word I've added a line of code which leaves the routine if f becomes negative. This seems to work in the sense, that there isn't any more a core dump. regards Dag ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2006-07-26 14:52 Message: Logged In: YES user_id=421 A patch for this will be included in the next version (1.40). ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-07-26 14:08 Message: Logged In: YES user_id=1608 FWIW, I looked at it and agree with Hartmut. --- pdftex.web.orig 2006-07-25 20:22:54.000000000 +0200 +++ pdftex.web 2006-07-26 14:19:35.514413422 +0200 @@ -24600,9 +24600,9 @@ p := q; ptmp := p; end else begin - p := prev_rightmost(temp_head, q); {get |link(p) = q|} + p := prev_rightmost(link(temp_head), q); {get |link(p) = q|} ptmp := p; - p := find_protchar_right(temp_head, p); + p := find_protchar_right(link(temp_head), p); end; @{ short_display_n(p, 1); ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2006-07-26 00:04 Message: Logged In: YES user_id=929 the value of temp_head seems to be right, but maybe it must be p := prev_rightmost(link(temp_head), q); {get |link(p) = q|} p := find_protchar_right(link(temp_head), p); with link(temp_head) instead of temp_head, as "The current line to be justified appears in a horizontal list starting at |link(temp_head)| and ending at |cur_break(cur_p)|." wild guess only, seems to work. Regards, Hartmut ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2006-07-25 23:33 Message: Logged In: YES user_id=929 seems it happens at this line in pdftex.web: p := prev_rightmost(temp_head, q); {get |link(p) = q|} the temp_head before this call is 1499996 which seems to be wrong. Then after the call p = 1499996, and there is font f=-4096 attached to this node which is also wrong. So it seems that somehow temp_head got lost before call of prev_rightmost(). No clue how. Regards, Hartmut ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-07-25 23:06 Message: Logged In: YES user_id=1608 ok, will check tomorrow ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2006-07-25 21:30 Message: Logged In: YES user_id=421 This still happens with the latest beta (I've got an OpenBSD machine now myself): Program received signal SIGSEGV, Segmentation fault. 0x1c048f03 in zcharpw (p=999996, side=1 '\001') at pdftex2.c:3191 3191 c = getrpcode ( f , mem [p ].hh.b1 ) ; (gdb) bt #0 0x1c048f03 in zcharpw (p=999996, side=1 '\001') at pdftex2.c:3191 #1 0x1c053d53 in zpostlinebreak (d=1) at pdftex2.c:7715 #2 0x1c007659 in zlinebreak (d=1) at pdftexini.c:2136 #3 0x1c05bb11 in initmath () at pdftex3.c:141 #4 0x1c065091 in maincontrol () at pdftex3.c:4582 #5 0x1c0130a2 in mainbody () at pdftexini.c:5647 #6 0x1c069d9e in main (ac=2, av=0xcfbd239c) at pdftexextra.c:340 (gdb) l pdftex2.c:3191 3186 c = getlpcode ( f , mem [p ].hh.b1 ) ; 3187 lastleftmostchar = p ; 3188 } 3189 else { 3190 3191 c = getrpcode ( f , mem [p ].hh.b1 ) ; 3192 lastrightmostchar = p ; 3193 } 3194 if ( c == 0 ) 3195 return Result ; (gdb) p f $1 = -4096 (gdb) p p $2 = 999996 (gdb) p mem[p].hh $3 = {v = {LH = -268435455, RH = 675}, u = {B1 = 1, B0 = -4096}} ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=555&group_id=106