On Thu, May 09, 2013 at 10:13:15AM -0400, Jim Ohlstein wrote: > # sysctl hw.model > hw.model: AMD FX(tm)-8350 Eight-Core Processor Ahh, so it seems that this is a CPU with the LWP. Please try the patch at the end of message.
As another workaround, which does not disable AVX support, you
could try loader tunable hw.xsave_mask=0x7.
> #0 0x00007ffffffff1e3 in ?? ()
> (gdb) info registers
> rax 0x16 22
> rbx 0x101 257
> rcx 0x7ffffffff1e3 140737488351715
> rdx 0x7fffffffc980 140737488341376
> rsi 0x1 1
> rdi 0x7fffffffc980 140737488341376
> rbp 0x7fffffffd000 0x7fffffffd000
> rsp 0x7fffffffc968 0x7fffffffc968
> r8 0x0 0
> r9 0x19 25
> r10 0x50 80
> r11 0x203 515
> r12 0x801460400 34381104128
> r13 0x0 0
> r14 0x7fffffffcfb0 140737488342960
> r15 0x7fffffffcfd0 140737488342992
> rip 0x7ffffffff1e3 0x7ffffffff1e3
> eflags 0x10203 66051
> cs 0x43 67
> ss 0x3b 59
> ds 0x0 0
> es 0x0 0
> fs 0x0 0
> gs 0x0 0
> (gdb) x/2xg 0x7fffffffc980+0x300
> 0x7fffffffcc80: 0x00007fffffffcd40 0x00000000000001c0
> (gdb) x/8xg 0x00007fffffffcd40
> 0x7fffffffcd40: 0xffffffffffffffff 0x0000000000000000
> 0x7fffffffcd50: 0x0000000000000000 0x0000000000000000
> 0x7fffffffcd60: 0x0000000000000000 0x0000000000000000
> 0x7fffffffcd70: 0x0000000000000000 0x0000000000000000
> (kgdb) x/1xw use_xsave
> 0x1: Error accessing memory address 0x1: Bad address.
> (kgdb) x/1xg xsave_mask
> 0x4000000000000007: Error accessing memory address 0x4000000000000007: Bad
> address.
diff --git a/sys/amd64/amd64/fpu.c b/sys/amd64/amd64/fpu.c
index de79baa..61ce94d 100644
--- a/sys/amd64/amd64/fpu.c
+++ b/sys/amd64/amd64/fpu.c
@@ -687,7 +687,7 @@ fpugetregs(struct thread *td)
offsetof(struct xstate_hdr, xstate_bv));
max_ext_n = flsl(xsave_mask);
for (i = 0; i < max_ext_n; i++) {
- bit = 1 << i;
+ bit = 1ULL << i;
if ((*xstate_bv & bit) != 0)
continue;
bcopy((char *)fpu_initialstate +
pgpT0LzYN0QsN.pgp
Description: PGP signature
