various updates, thanks to those who provided feedback.

- powerpc ports work
- m68k ports work, x68k has a new caveat
- hppa works
- mips64eb vs UVMHIST has been fix
- i386 reboot problem fixed (cardbus issue, not GCC related)
This commit is contained in:
mrg 2021-04-17 01:59:41 +00:00
parent f405a45af6
commit ddd88a608a
1 changed files with 17 additions and 31 deletions

View File

@ -1,4 +1,4 @@
$NetBSD: README.gcc10,v 1.9 2021/04/15 05:12:57 mrg Exp $
$NetBSD: README.gcc10,v 1.10 2021/04/17 01:59:41 mrg Exp $
new stuff:
@ -49,17 +49,17 @@ earmv7 y b y y y ? ? ?
earmv7eb y b y y y ? ? ?
earmv7hf y y y y y y y n
earmv7hfeb y b y y y ? ? ?
hppa y b y y y ? ? ?
hppa y y y y y y ? ?
i386 y y y y y y n[8] ?
ia64 y y y y y ? N/A n
m68000 y b y y n[1] ? ? ?
m68k y y y y y[9] y ? ?
m68k y y y y y[9] y n n
mipseb y y y y y y ? ?
mipsel y y y y y y ? ?
mips64eb y y y y y n[4] ? ?
mips64eb y y y y y y ? ?
mips64el y b y y y y ? ?
powerpc y b y y y n[2] ? ?
powerpc64 y b y y y N/A N/A ?
powerpc y b y y y y ? n
powerpc64 y b y y y N/A N/A n
sh3eb y b y y y ? ? ?
sh3el y y y y y y y[7] ?
sparc y y y y y y y n
@ -74,36 +74,22 @@ coldfire ? N/A ? ? ? N/A N/A
architecture tools kernels libgcc native-gcc make release runs atf
[1] - ramdisk.fs is too large, needs fixing.
[2] - powerpc ports have regressions; fixes incoming from rin@
[3] - missing __aarch64_swp1_acq in kernels. missing lse.S in libgcc.
[4] - edgerouter lite3 crashes early in boot:
[ 1.0000000] u-boot bootmem desc @ 0x24108 version 3.0
[ 1.0000000] phys segment: 0xfc000 @ 0x102000
[ 1.0000000] adding 0xfc000 @ 0x102000 to freelist 3
[ 1.0000000] phys segment: 0x7276000 @ 0xe8a000
[ 1.0000000] adding 0x7276000 @ 0xe8a000 to freelist 3
[ 1.0000000] phys segment: 0x7efa000 @ 0x8102000
[ 1.0000000] adding 0x7efa000 @ 0x8102000 to freelist 3
[ 1.0000000] phys segment: 0xff00000 @ 0x410000000
[ 1.0000000] adding 0xff00000 @ 0x410000000 to freelist 0
[ 1.0000000] pid 0(system): trap: cpu0, TLB miss (load or instr. fetch) in kernel mode
[ 1.0000000] status=0xa3, cause=0x8, epc=0xffffffff802564d4, vaddr=0
[ 1.0000000] tf=0x9800000000165bb0 ksp=0x9800000000165cf0 ra=0xffffffff80256470 ppl=0
[ 1.0000000] kernel: TLB miss (load or instr. fetch) trap
Stopped in pid 0.0 (system) at ffffffff802564d4: sw zero,16(s0)
0x9800000000165cf0: 0xffffffff802563+0xd8 (0xffffffff80000000,0,0,0xc000000008000000) ra 0xffffffff803edd18 sz 0
0x9800000000165cf0: 0xffffffff803edc+0x84 (0xffffffff80000000,0,0,0xc000000008000000) ra 0 sz 0
- OCTEON kernel seems OK, something in my private configuration?
[6] - vax vs c++ exceptions issue, same as it ever was
[7] - fails just as poorly in gxemul/landisk as GCC 9
[8] - i386 seems to have a signal delivery issue. GCC 9 or 10 kernels are
unable to reboot properly, and GCC 10 atf runs hang in signal delivery.
bisection so far has reduced bug introduction between 2020-05-08 and
2020-05-15. possibly pthread or condvar changes in this period.
[8] - i386 seems to have a signal delivery issue. pthread tests hang and then
complain with eg:
threads_and_exec: q[ 627.6700846] sorry, pid 3154 was killed: orphaned traced process
this problem occurs with GCC 9 as well.
[9] - x68k/stand fails with:
link loadbsd/loadbsd.x
loadbsd1: program header #1 is not loadable
may be due to it pulling in libc sources that get TLS marked?
problem is understood (thanks tsutsui@): previously loadbsd was linked
against normal userland libc components for m68k, which allows it to
detect running on m68000 CPU and sanely fail, instead of an barf error.
new gcc built libc includes TLS support, that isn't compatible with this
and the conversion process fails. can be fixed by using libkern instead,
and accepting the poor error on m68000.
@ -114,7 +100,7 @@ CPU platform list
--- -------- ----
acorn32 cats epoc32 evbarm-eb evbarm-el hpcarm iyonix netwinder shark zaurus
earmv4: y y y y y y y y
earm: r y y y
earm: y y y y
earmhf: y y
earmv6: y y
earmv6hf: y y