On Mon, 2005-08-29 at 11:49 +0100, Ross Burton wrote:
> On Sat, 2005-08-27 at 20:43 -0400, Owen Taylor wrote:
> > cpp is built into GCC these days, so is not lightweight by any means.
> >
> > I don't know if there is a 10k cpp implementation out there, but if
> > there is, then switching X.org to use that might by the right course in
> > this area.
>
> People have said this but I don't see it:
>
> $ which cpp
> /usr/bin/cpp
> $ ll /usr/bin/cpp
> lrwxrwxrwx 1 root root 7 2005-08-21 15:01 /usr/bin/cpp -> cpp-4.0
> $ ll /usr/bin/cpp-4.0
> -rwxr-xr-x 1 root root 90232 2005-08-21 19:51 /usr/bin/cpp-4.0
> $ ldd /usr/bin/cpp-4.0
> linux-gate.so.1 => (0xffffe000)
> libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0xb7dee000)
> /lib/ld-linux.so.2 (0xb7f2d000)
>
> Using "strace -e open" doesn't show cpp opening at runtime any other
> libraries.
>
> What is xrdb/cpp *doing* in the 10% of the startup time? It doesn't
> perform any I/O as gnome-settings-daemon does that instead, and streams
> the contents of the files into xrdb's stdin.
cpp and gcc are just wrapper scripts.
$ strace -f -e execve cpp foo.c > /dev/null
execve("/usr/bin/cpp", ["cpp", "foo.c"], [/* 38 vars */]) = 0
Process 16288 attached
Process 16287 suspended
[pid 16288] execve("/usr/libexec/gcc/i386-redhat-linux/4.0.1/cc1",
["/usr/libexec/gcc/i386-redhat-lin"..., "-E", "-quiet", "foo.c"],
[/* 40 vars */]) = 0
Process 16287 resumed
Process 16288 detached
--- SIGCHLD (Child exited) @ 0 (0) ---
Process 16287 detached
$ ls -l /usr/libexec/gcc/i386-redhat-linux/4.0.1/cc1
-rwxr-xr-x 1 root root 4576236 Aug 26 22:34 /usr/libexec/gcc/i386-redhat-linux/4.0.1/cc1
Taking a 1-2 seconds to read a binary of that size isn't great, and
likely object-file reordering could improve things quite a bit, but it
should be more believable than for a 90k executable.
Regards,
Owen
Attachment:
signature.asc
Description: This is a digitally signed message part