bug#52446: 28.0.90; Infinite loop in add_row_entry

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

> Now I'm confused: if row_table[i] is NULL for each i, then how did we
> enter that loop?  Its condition checks for entry being non-NULL.

The disassembly checks out, so that condition is not being mis-compiled.
But that doesn't rule out the possibility of a miscompilation elsewhere.

I will try to reproduce this under a memory checker, and with
optimizations turned off.