Re: [BUG] malloc inside signal handler

[Peter Stephenson <p.w.stephenson@ntlworld.com>]

> On 17 August 2021 at 05:20 Bart Schaefer <schaefer@brasslantern.com> wrote:
> Or maybe shingetline() doesn't actually need to call winch_unblock()
> when loop(0,...) is called and isatty(bshin) is false ... which means
> passing the value of toplevel down from loop().  But that probably
> just fixes adjustwinsize() and leaves other signal handlers
> unprotected.

Might be worth doing anyway, since adjustwinsize() is the one that's doing
the big allocation within the signal code.  I haven't checked, but I
doubt there's anything like that in the others.

Avoiding stdio for input of this sort doesn't actually look infeasible,
either --- mentions of bshin (buffered shell input from files) are
restricted to only a couple of files.  But it's kind of annoying as this
is exactly the job stdio is designed for.

pws