🚀 go-pugleaf

In article <1990Jul6.165254.26878@bbt.se> pgd@bbt.se (P.Garbha) writes:
>I have problem with emacs. 
>The problem is that two quick C-G's give "fatal error (1)".
>Where "quick" is dependant on the system loading.
>Does anyone have a fix for this?
>
>(My operating system is Xenix/386)
>
>

I got one RTFM answer to this post, and since i thought that was quit
inappropriate, since it did not solve the problem at all, i hacked up
emacs to fix the problem. But i don't know if it will break something
else.

The problem is that after one C-G, the SIGINT trap is reset by the
SysV kernel, and when the next comes, it will abort the process.

There seems to be one simple fix, though. Since it is so simple, i
cannot understand why it is not there in emacs in the first place.
The fixes are to use the sigset() and sigrelse() system calls
instead of the signal() system call, in the keyboard.c file.
Then i get the advertised behavior, of emacs asking if i want to
abort/dump core, etc.

I suspect that this can be a problem for SysV users, on other
operating systems than Xenix.

To force the error, try the following:
Edit a very big file (i used a 3MB file), insert a character in the
beginning of the file, and then go to the bottom, and insert a
character. While emacs is shuffling core, give a few fast C-G's.
If you get "Fatal error (1)", you have the bug. If nothing happens,
try a bigger file, if you get the question about auto-save/dump-core
your version seems to work.