> egcs-1.1.x
I have a linux box I've been actively working on over the past few
months, and everything is as the bleeding edge on it, and it seems to
function just fine.... here's the relevant versions of everything on it:
Kernel: 2.2.5 (been running the latest since the late 2.1.xxx)
Libc: glibc-2.1
Compiler: egcs-1.1.2 (and previously 1.1.1, 1.1b, etc...)
Binutils: 2.9.1.0.22b (19a until about a week or two ago)
I'm actually way past taking my life in my own hands, in that the
binutils, compiler, and kernel have all been built using egcs-1.1.2 with
the following optimization flags:
-march=pentiumpro -O6 -fomit-frame-pointer -malign-loops=0
-malign-functions=0 -malign-jumps=0
(yes, even the kernel at O6).
And if the problem is related to Linux network code... well, this box is
a firewall/router with two 3c905B interfaces pumping data at 100mbps
full duplex, complex ipchains for filtering and TOS, etc... The box
itself is a dual PII-450, 256MB SDRAM, on an ASUS P2B-D motherboard. It
has a couple of 6GB IDE drives, the two aforementioned network cards,
and a video card (which never gets used, the box is usually headless
unless there's a problem). I haven't experienced any egcs-related
kernel miscompiles in this setup, even when I expected to (like the
first time I tried O6). I guess I might also add for completeness that
I'm running the latest of Richard's devfs patches, the PIII/FXSR patch
(as ammended on the list for the struct size and that typo'd bitshift),
and the /proc/.config patch (modified a bit locally)... and I'm
compiling with no modules support at all, if that makes any difference.
Brandon
:.˛mkabzwmb˛mbz_^nrzh&z vy 杶i i