A test with the POSIX S_ISREG macro on HISTFILE will determine if it, or the file it points to in the case of a symlink, is a regular file.
Just looked through the source, and it looks like general.c:file_exists() does not do any special handling of non-regular files, and lib/readline/histfile.c:history_do_write() calls open() and rename() on HISTFILE without checking if it is a non-regular file, which I imagine could lead to various "bad things" in the case of pipes, char and block devices, etc. such as what the OP pointed about about "/dev/null". I haven't verified, but I think it also will not dereference links and may potentially be made to clobber or otherwise mess with files ending in hyphen (/etc/shadow-, /etc/passwd- and /etc/group- come to mind) because of the way lib/readline/histfile.c:history_backupfile() constructs the backup file name. Someone with more knowledge than myself might want to look at this closer. It seems like it might be a vector for abuse. -Jonathan Hankins On Fri, Jan 30, 2015 at 9:06 AM, Greg Wooledge <wool...@eeg.ccf.org> wrote: > On Fri, Jan 30, 2015 at 09:58:43AM -0500, Chet Ramey wrote: > > On 1/30/15 4:36 AM, crocket wrote: > > > It turns out that tramp on emacs 24.4 sets $HISTFILE to /dev/null and > > > makes bash delete /dev/null when I kill emacs. > > > > > > When /dev/null is not a character device but a regular file, a lot of > > > programs freeze. > > > > This indicates a permissions or file system problem. > > My guess was that (s)he's running bash as root, which accounts for the > ability to delete /dev/null and recreate it as an ordinary file. > > -- ------------------------------------------------------------------------ Jonathan Hankins Homewood City Schools The simplest thought, like the concept of the number one, has an elaborate logical underpinning. - Carl Sagan jhank...@homewood.k12.al.us ------------------------------------------------------------------------
