bug#16128: 24.3.50; SIGSEGV at font.c:2901 : eassert (FRAME_DISPLAY_INFO (font->frame)->n_fonts);

[Noam Postavsky]

close 16128 
quit

Dmitry Antipov <dmantipov@yandex.ru> writes:

> On 12/13/2013 06:24 PM, Nicolas Richard wrote:
>
>> Sorry I cannot confirm.
>
> Yes, this is somewhat more serious than I suspected before.  When the frame 
> is gone,
> 'struct frame' memory may be reused for other object and so we can't rely on 
> anything
> under 'font->frame' pointer.  This approach probably requires a kind of 
> reference
> counting system, which will trigger even more potentially dangerous changes.
> With all my regrets, I'm reverting 2013-12-12 font-related change - although 
> I believe
> in that this is Right Thing (TM) in general, this code needs more work :-(.

Since there was no further response, I guess that revert fixed it.