Re: Slur_proto_engraver segfault

[David Kastrup]

David Kastrup <address@hidden> writes:

> Dan Eble <address@hidden> writes:
>
>> Forgive me for being brief, but I thought I should at least mention
>> this in case someone would like to approach it with code review:
>>
>> Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
>> 0 lilypond 0x001f35ff Slur_proto_engraver::derived_mark() const + 79
>> 1 lilypond 0x00279696 Translator::mark_smob() const + 54
>> 2 libguile.17.dylib 0x00718aaa scm_gc_mark_dependencies + 1050
>
> Without further information, it's not really possible to figure out
>what's wrong here.  The crash is while in garbage protection,
>ostensibly trying to mark already freed structures.  It's not really
>clear how those would be coming about since they should have been
>marked at the previous garbage collection and/or protected in between.

Actually, marking 0 pointers.  Ugh.

I pushed a fix to staging.

Phil, this is absolutely serious stuff, likely causing crashes for large
scores on most platforms.  Too bad we did not catch this (no idea why)
in time.

I wonder whether we have a chance to actually _retract_ 2.19.29, but
short of that, we really need to create 2.19.30 as soon as possible
rather than waiting for the full 2-week turnaround.

Sorry for messing this up.

-- 
David Kastrup