Bug #931
closedFXRSTR: illegal FP MXCSR ffff002f didinit = 0
0%
Description
Still no good. I deleted /usr/obj and updated to latest -HEAD this
morning, recompiled/installed both world and GENERIC kernel. Once
again, I took out the SIGFPE so I could run things. I booted single
user and mounted my fs. Both fsck and mount caused the error. Idle
in single-user the error doesn't occur. Other commands do too, but
some don't or sometimes do. 'ls' is one example. By itself, it does
not produce the error, but ls -l does. ls -R does not but ls -lR
does. dmesg by itself doesn't, but dmesg with an invalid option
(dmesg -v) does. Hopefully this is helpful.
Eric
Updated by corecode about 16 years ago
What if you use kernel + world from a live cd? What if you only use the
kernel or only use the world from the live cd?
cheers
simon
Updated by eric.j.christeson about 16 years ago
On Jan 25, 2008 3:12 PM, Simon 'corecode' Schubert
<corecode@fs.ei.tum.de> wrote:
ejc wrote:
Still no good. I deleted /usr/obj and updated to latest -HEAD this
morning, recompiled/installed both world and GENERIC kernel.What if you use kernel + world from a live cd? What if you only use the
kernel or only use the world from the live cd?
I have no problems with a kernel from live cd. That's how I fixed it
in the first place.
I booted from a live cd and copied the kernel to my hard drive. Then
I could reboot with that kernel and rebuild my world/kernel.
This would seem to indicate it's a kernel problem, not app/library, no?
eric
Updated by corecode about 16 years ago
definitely.
are you sure you have fresh kernel sources? I couldn't explain else.
cheers
simon
Updated by eric.j.christeson about 16 years ago
On Jan 25, 2008 5:33 PM, Simon 'corecode' Schubert
<bugs@lists.dragonflybsd.org> wrote:
Simon 'corecode' Schubert <corecode@fs.ei.tum.de> added the comment:
definitely.
are you sure you have fresh kernel sources? I couldn't explain else.
I did a cvsup using the cvs-supfile to and empty directory, removed
/usr/src and /usr/obj/usr,
and did a cvs co from the new local cvs directory. The only change I
made to the code was
commenting out the FPE which would cause a panic. I'll put that back
in so I can debug, though
it wasn't very helpful last time. I don't see how that could bring in
any old stuff, unless the build
process is messed up.
Eric
Updated by dillon about 16 years ago
:..
:commenting out the FPE which would cause a panic. I'll put that back
:in so I can debug, though
:it wasn't very helpful last time. I don't see how that could bring in
:any old stuff, unless the build
:process is messed up.
:
:Eric
Maybe you did something to your /etc/make.conf that is blowing up the
build. High bits of ffff can only happen when userland overwrites
the signal stack fp save area with a fnsave instruction, and only the
old threaded libc_r did that.
-Matt
Matthew Dillon
<dillon@backplane.com>
Updated by eric.j.christeson about 16 years ago
I resync'd my cvs repo and rebuilt world and kernel with the /etc/make.conf from
a recent snapshot image. Same problem. This morning I downloaded
2CSNAP-20080128-2330-Devel.iso Booting from it results in the same error. This
can't be a problem with my build env if a snapshot CD exhibits the error.
Eric
Updated by dillon about 16 years ago
:I resync'd my cvs repo and rebuilt world and kernel with the /etc/make.conf from
:a recent snapshot image. Same problem. This morning I downloaded
:2CSNAP-20080128-2330-Devel.iso Booting from it results in the same error. This
:can't be a problem with my build env if a snapshot CD exhibits the error.
:
:Eric
Hmm. Maybe there's a bug with the OS related to machines without SSE
support. Could you post the CPU lines from your dmesg ? For example,
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ (2009.79-MHz 686-class CPU)
Origin = "AuthenticAMD" Id = 0x20f32 Stepping = 2
Features=0x178bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2,HTT>
Features2=0x1<SSE3>
Hyperthreading: 2 logical CPUs
AMD Features=0xe0500000<<b20>,AMIE,<b29>,DSP,3DNow!>
AMD: Rerouting HyperTransport EXTINT/NMI to APIC
-Matt
Matthew Dillon
<dillon@backplane.com>
Updated by dillon about 16 years ago
Well well well. I do believe I see a bug in the code.
The kernel error message it is spouting it is not conditionalized
on the existance of SSE.
I also see a bug in the vkernel ... it isn't supported SSE extensions
like it should.
-Matt
Matthew Dillon
<dillon@backplane.com>
Updated by dillon about 16 years ago
Ok, I believe the problem has now been fixed on HEAD. It was a bug
in the kernel when running on machines without SSE support.
Can you get the latest HEAD or are you in a situation now where you
can't compile anything at all due to the bug?
-Matt
Matthew Dillon
<dillon@backplane.com>
Updated by eric.j.christeson about 16 years ago
Yes, I've got a couple of spare kernels on the box so I'll reboot and update.
Here's my CPU lines from dmesg FWIW:
CPU: AMD Athlon(tm) Processor (995.56-MHz 686-class CPU)
Origin = "AuthenticAMD" Id = 0x642 Stepping = 2
Features=0x183f9ff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXSR>
AMD Features=0xc0440000<RSVD,AMIE,DSP,3DNow!>
Thanks,
Eric
Updated by eric.j.christeson about 16 years ago
That fixed the problem. Thanks
Eric