[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: store_open() unforgiving in bootstrap fs
From: |
John Tobey |
Subject: |
Re: store_open() unforgiving in bootstrap fs |
Date: |
Mon, 26 Nov 2001 07:28:12 -0500 |
User-agent: |
Mutt/1.2.5i |
On Mon, Nov 26, 2001 at 06:48:42AM -0500, Roland McGrath wrote:
> > The crash came in file_name_lookup() presumably because there is no
> > working directory port yet.
>
> Don't presume, debug! There is no reason why file_name_lookup should ever
> produce a crash (unless given a bad string argument), and if it does then
> that is the bug (in libc).
Okay, but I will have to wait until after work. :-) Debugging user
code is still rather tedious under my setup. The problem should be
easily reproducible - just delete "device:" from your root string and
boot. If it works or prints a meaningful diagnostic, then I am on
crack and should be ignored. But I think I can get you the source
line where it happens. (It's actually in _hurd_ports_use() doing
"xchg %eax,(%edx)" with %edx==0.)
I am not so sure this should be considered a bug in libc, since by
assuming the existence of the usual environment (including current
directory) it can be faster the 99.999999% of the time that that
environment exists.
-John