mips64 and GCC 5

["John D. Baker" <jdbaker%mylinuxisp.com@localhost>]

Some things I observed in my earlier trial of GCC 5 on my Lemote YeeLoong
(LOONGSON-derived kernel), now borne out with the official switch of
mips64 ports to GCC 5:

'ntpd' no longer complains about an invalid struct timex pointer.

Xorg module auto-loading now works.  I do not have to explicitly load
"vgahw", "exa", and "xaa" modules via "xorg.conf" for the siliconmotion
driver.  (Still have to explicitly set "Panel Size" and "DefaultDepth".)


Things still needing attention:

The changes to ext2fs have made it unstable.  It mostly works, but is
likely to corrupt the filesystem.  I don't know if this is cross-platform
or is only manifesting on mips64(el) as I don't have any other machines
using ext2fs.  This is not related to GCC 5.

amd still doesn't work, complaining about "invalid argument".
port-evbmips/50058


Oops.  I spoke too soon.  Letting the system sit overnight and run the
nightly sweep of files resulted in:

pid 1005(find): trap: cpu0, address error (load or I-fetch) in kernel mode
status=0x8400ffa3, cause=0x10, epc=0xffffffff804007a4, vaddr=0x98000000854a001a
tf=0x98000000812cba40 ksp=98000000812cbb80 ra=0xffffffff804007a4 pp1=0x4000
kernel: address error (load or I-fetch) trap
Stoped in pid 1005.1 (find) at ffffffff804007a4:      lwu     v0,0(s0)
db> bt
0x98000000812cbb80: 0+ffffffff804007a4 (98000000bf459418,0,98000000bf459208,98000000bf459248) ra 0 sz 0
User-level: pid 1005.1

I've seen this happen before, but hadn't recently until now.

A curious thing is that DDB output isn't rendered on the screen unless
one switches to another virtual terminal and back again.

-- 
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645