Hello,
> On 24. Dec 2023, at 15:31, Camm Maguire <c...@maguirefamily.org> wrote:
>
> "Kirill A. Korinsky" <kir...@korins.ky> writes:
>
>> --- o/main.c
>> +++ o/main.c
>> @@ -432,7 +432,11 @@ gcl_cleanup(int gc) {
>> if (gc) {
>>
>> saving_system=TRUE;
>> +
>> +/* on macOS it leads to ERROR "Caught fatal error [memory may be damaged]"
>> */
>> +#ifndef __APPLE__
>> GBC(t_other);
>> +#endif
>> saving_system=FALSE;
>>
>> minimize_image();
>>
>> If I recall right I made this patch as result of git bisect that points that
>> this issue was introduced by cb842e61fd2d0ff5baf51918112003a3a1b69b45 when
>> this line was introduced.
>>
>
> OK we need to get to the bottom of this too. It seems the problem
> area is:
>
> if (saving_system) {
>
> struct pageinfo *v;
> void *x;
> struct typemanager *tm=tm_of(t_stream);
> unsigned j;
>
> for (v=cell_list_head;v;v=v->next)
> if (tm->tm_type==v->type)
> for (x=pagetochar(page(v)),j=tm->tm_nppage;j--;x+=tm->tm_size) {
> object o=x;
> if (type_of(o)==t_stream && !is_free(o) && o->sm.sm_fp &&
> o->sm.sm_fp!=stdin && o->sm.sm_fp!=stdout && o->sm.sm_fp!=stderr)
> close_stream(o);
> }
>
> gc_time = -1;
> }
>
> Problem should also go away if you remove the saving_system=TRUE
> line. I suspect mac is naming stdin,stdout,stderr in an unusual way
> and gcl is trying to close one of these.before I run almost 3 hours of build on my laptop :) let confirm your hypotesa about wired naming. How can I check it? -- wbr, Kirill
signature.asc
Description: Message signed with OpenPGP
