Re: x86 math_error warning in Linux kernel 3.10

[Thomas Gleixner]

On Fri, 14 Nov 2014, Borislav Petkov wrote:
> On Fri, Nov 14, 2014 at 12:19:16PM -0800, Vinson Lee wrote:
> > Hi.
> > 
> > We hit this x86 math_error warning in Linux kernel 3.10.
> > 
> > ------------[ cut here ]------------
> > WARNING: at arch/x86/include/asm/fpu-internal.h:524 math_error+0xd1/0x219()
> > Modules linked in: cls_basic act_mirred cls_u32 veth sch_ingress
> > netconsole configfs ipv6 dm_multipath scsi_dh video sbs sbshc hed
> > acpi_pad acpi_ipmi acpi_i2c parport_pc lp parport tcp_diag inet_diag
> > ipmi_si ipmi_devintf ipmi_msghandler dell_rbu sg iTCO_wdt
> > iTCO_vendor_support dcdbas igb i2c_algo_bit ptp pps_core shpchp
> > lpc_ich i2c_i801 mfd_core i2c_core ioatdma dca i7core_edac edac_core
> > microcode freq_table mperf ahci libahci libata sd_mod scsi_mod
> > CPU: 2 PID: 53182 Comm: java Not tainted 3.10.50 #1
> >  0000000000000000 ffff8808e1f5be38 ffffffff8146cb74 ffff8808e1f5be70
> >  ffffffff8103cbf9 0000000000000000 ffff88090b2c9730 ffff8808e1f5bf58
> >  0000000000000010 0000000000000000 ffff8808e1f5be80 ffffffff8103ccbf
> > Call Trace:
> >  [<ffffffff8146cb74>] dump_stack+0x19/0x1b
> >  [<ffffffff8103cbf9>] warn_slowpath_common+0x65/0x7d
> >  [<ffffffff8103ccbf>] warn_slowpath_null+0x1a/0x1c
> >  [<ffffffff8100317b>] math_error+0xd1/0x219
> >  [<ffffffff81007f6c>] ? read_tsc+0x9/0x19
> >  [<ffffffff8107bab8>] ? timekeeping_get_ns.constprop.10+0x11/0x36
> >  [<ffffffff8107bf32>] ? ktime_get+0x68/0x76
> >  [<ffffffff81007f6c>] ? read_tsc+0x9/0x19
> >  [<ffffffff8107bab8>] ? timekeeping_get_ns.constprop.10+0x11/0x36
> >  [<ffffffff810032d6>] do_coprocessor_error+0x13/0x15
> >  [<ffffffff81478a38>] coprocessor_error+0x18/0x20
> > ---[ end trace 3e4a6532a67ba6d3 ]---
> 
> AFAICT, you're getting an FPU exception for a task which hasn't used the
> FPU or current is somehow pointing to the wrong task.
> 
> Can you trigger this with the latest kernel too, i.e., say, 3.18-rc4?

Also Vinson forgot to mention HOW that is triggered. Without an
explanation of the reproducer it's hard to tell what's going wrong.

Thanks,

	tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/