Somewhat random segfault

[Mats Bengtsson]

   Hi,

   The following example results in a segfault almost every time I process
   it using the 64bit Linux version of 2.23.7, downloaded from
   lilypond.org.

   The log output is

   GNU LilyPond 2.23.7 (running Guile 2.2)
   Processing `test1.ly'
   Parsing...
   Interpreting music...[8][16][24][32]
   test1.ly:19:8: warning: unterminated decrescendo
   a4. \!
          \> ~ a4 g8 |}
   Preprocessing graphical objects...
   Finding the ideal number of pages...
   Fitting music on 1 page...
   Drawing systems...malloc(): unsorted double linked list corrupted
   Abort (core dumped)

   However, occasionally (5 attempts out of 25), lilypond processes the
   file without any errors and produces the expected PDF, so it seems like
   some ugly memory handling bug. I happened to have version 2.23.3
   installed and it also segfaults on the file. However, with 2.22.2 there
   are no problems, so it's a regression that was introduced between 2.22
   and 2.23.

   \version "2.23.7"
   soprano = \relative c' {
   \key f \major
   \time 6/8
   \repeat volta 2 {
   r4. r8 r  c \mf |
   f4 \< g8 a ( bes ) c |
   bes4 \! \> a8 g4 \! a8 |
   f4 \< a8 c ( a ) c \! |
   d4 d8 c4 f,8 |
   bes4 bes8 a ( f ) g |
   a4 b8 c4 \breathe c8 \f |
   f4 c8 d4 d8 |
   c4 a8 f4 a8 |
   c4 a8 f ( g ) a |
   g4 e8 c4 \breathe c8 |
   f4 \< g8 a ( bes ) c |
   a4. \! \> ~ a4 g8 |}
   \alternative {{f2. \!  R2.*2
   |}{f2. ~ f2. ~ f2. \bar "|." }}
   }
   \score{
     \unfoldRepeats \soprano
   }

   (If I add the missing \! in the second ending the file no longer
   triggers the SIGSEGV, but I guess that's more of a coincidence).

       /Mats