Re: [CRASH] in GC, in compact_string_chars()
Kirill M Katsnelson <kkm(a)kis.ru> writes:
I see a corruption in a string_char_block struture. I remember that I
have
already been debugging this place:
if (FREE_STRUCT_P (from_s_chars))
{
fullsize = ((struct unused_string_chars *) from_s_chars)->fullsize;
from_pos += fullsize;
continue;
}
string = from_s_chars->string;
assert (!(FREE_STRUCT_P (string)));
If memory serves, this is near the same place where Andrew Cosgriff
was getting crashes or inflooping.
In the above assert statement, string variable has an ascii value,
like
0x4C282020 (this is from modeline: " (L" isp interaction), or something
else, but always characters. This no doubt ends up with GPF, or SEGV,
whatever you call this sweetie on your system.
I spent my whole life and borrowed another one catching the reason
for
that, and all to no avail. Does anybody remember something related to
such a crash?
I have no reproduction. This is the third time in the last 2 days
I'm getting this, and the first time in debug build.
Catching corruption problems like this in GC definitely makes it more
difficult to debug.