This is the mail archive of the
cygwin-patches@sources.redhat.com
mailing list for the Cygwin project.
Re: Just checked in a major change to cygwin
- To: Chris Faylor <cygwin-patches at sources dot redhat dot com>
- Subject: Re: Just checked in a major change to cygwin
- From: Chris Faylor <cgf at cygnus dot com>
- Date: Mon, 4 Sep 2000 23:19:36 -0400
- References: <20000903003055.A14834@cygnus.com> <128221669273.20000904115006@logos-m.ru>
On Mon, Sep 04, 2000 at 11:50:06AM +0400, Egor Duda wrote:
>Hi!
>
>Sunday, 03 September, 2000 Chris Faylor cgf@cygnus.com wrote:
>
>CF> For the last several weeks, I've been working on giving cygwin it's own
>CF> "heap". I use this heap to allocate memory that should be common to a
>CF> process's children. This was one of the things that I desperately wanted
>CF> copy-on-write for but I ended up implementing my own crude version instead.
>
>stat("f:\\tmp\\bla-bla",...) causes exception with your latest
>changes.
>
>in this line (path.cc:1025)
>
>rc = normalize_win32_path (cwd_win32 (TMPCWD), src_path, dst);
>
>alloca() is called after src_path and dst are put on stack. patch
>attached.
Hmm. I don't exactly understand why this is a problem. It doesn't
cause me any problems but maybe this is because I am using a different
version of the compiler.
Anyway, I took a look at cwd handling again and did what I should have
done the first time. I think that path.cc now is a little more sane
with respect to its handling of the current working directory.
I made enough changes that I probably broke something new but at least
I won't have to puzzle over 37 different places where getcwd_inner was
called.
cgf