In the past we've only used this where it makes sense.
That's because in the past -Os has often resulted in either mis-compiled code or has resulted in larger binaries than w/o it.
I suppose it is OK, but I'm leery since this touches every single architecture.

Check also D7600 - I did try to take on -Os there, but it is a bit mixed up with crypto stuff. Maybe some ideas are useful in your work.

In D9430#194891, @imp wrote:

In the past we've only used this where it makes sense.
That's because in the past -Os has often resulted in either mis-compiled code or has resulted in larger binaries than w/o it.
I suppose it is OK, but I'm leery since this touches every single architecture.

That is good point as well.

Having thought about it, I'd go farther: absolutely do not commit this change unless you can test every single boot loader that this change touches. So does it work on mips? does it work on powerpc? Do these binaries you made actually work?

In D9430#194891, @imp wrote:

In the past we've only used this where it makes sense.
That's because in the past -Os has often resulted in either mis-compiled code or has resulted in larger binaries than w/o it.
I suppose it is OK, but I'm leery since this touches every single architecture.

-Os miscompiling binaries is a potential issue, but if that was the case that's a compiler bug to care about and fix.

In D9430#194902, @imp wrote:

Having thought about it, I'd go farther: absolutely do not commit this change unless you can test every single boot loader that this change touches. So does it work on mips? does it work on powerpc? Do these binaries you made actually work?

If only I had working QEMU emulators to play with...

In D9430#194903, @ngie wrote:

In D9430#194891, @imp wrote:

In the past we've only used this where it makes sense.
That's because in the past -Os has often resulted in either mis-compiled code or has resulted in larger binaries than w/o it.
I suppose it is OK, but I'm leery since this touches every single architecture.

-Os miscompiling binaries is a potential issue, but if that was the case that's a compiler bug to care about and fix.

While that's true, that's no reason to purposely break something in the tree.

In D9430#194904, @ngie wrote:

In D9430#194902, @imp wrote:

Having thought about it, I'd go farther: absolutely do not commit this change unless you can test every single boot loader that this change touches. So does it work on mips? does it work on powerpc? Do these binaries you made actually work?

If only I had working QEMU emulators to play with...

Ask for help from others. There's several folks that have qemu environments that work for all our ports except maybe sparc64.

In D9430#194914, @imp wrote:

...

Ask for help from others. There's several folks that have qemu environments that work for all our ports except maybe sparc64.

Yeah, I'll try to make this work again, and if successful document it.

In D9430#194891, @imp wrote:

In the past we've only used this where it makes sense.
That's because in the past -Os has often resulted in either mis-compiled code or has resulted in larger binaries than w/o it.
I suppose it is OK, but I'm leery since this touches every single architecture.

That is good point as well.

In D9430#194915, @ngie wrote:

In D9430#194914, @imp wrote:

...

Ask for help from others. There's several folks that have qemu environments that work for all our ports except maybe sparc64.

Yeah, I'll try to make this work again, and if successful document it.

Still, even if we just build everything with -Os, actually even x86 bits are built without. I did simple test - built ficl, geli, zfs, libstand and libi386 with -Os

-rw-r--r-- 1 root wheel 368640 Feb 4 12:24 zfsloader
-r-xr-xr-x 1 root wheel 417792 Feb 1 14:34 /boot/zfsloader

In D9430#194902, @imp wrote:

Having thought about it, I'd go farther: absolutely do not commit this change unless you can test every single boot loader that this change touches. So does it work on mips? does it work on powerpc? Do these binaries you made actually work?

I don't have a working example to go off of for MIPS, but I was able to boot a cd image on ppc64 and USB image on arm64 using qemu.

• Couldn't verify ppc because qemu support is busted.
• Couldn't verify mips* because I lack a qemu recipe that uses PXE/NFS roots and release/ doesn't support mips*.
• I'll try sparc64 in a bit..

Ask on #bsdmips for help with MIPS. Adrian and Baldwin should be able to help. Might be able to get direction from Robert Watson or Brooks Davis since they do something with some VM for their CHERRI work.