Re: NetBSD/sparc under qemu ?

[Artyom Tarasenko <atar4qemu%googlemail.com@localhost>]

Hi Manuel,

> I have an aging SS10 still in production, for which qemu would
> be a good remplacement (it has to run NetBSD/sparc as it runs
> some SunOS 4 binaries). A NetBSD 2.0 kernel (this is whats runs on the SS10
> at this time) fails with:
> scsibus0: waiting 2 seconds for devices to settle...
> sd0 at scsibus0 target 0 lun 0: <, , > disk fixed
> cpu0: NMI: system interrupts: 40040000<VME=0,SBUS=0,SC,ME>
...
> cpu_Debugger(0xf01b6a18, 0xf0147458, 0xf8c00e00, 0xf01ccd40, 0x100, 
> 0xf01d5800)
> at netbsd:mem_access_fault4m+0x148
> mem_access_fault4m(0x29, 0x36, 0xf8c00e00, 0xf01ccde8, 0x0, 0xf0147458) at
> 0xf0006254
> 0xf0006254(0xf01ac400, 0x0, 0x0, 0x0, 0x0, 0x3b) at
> netbsd:viking_module_error+0x70
> viking_module_error(0x0, 0x0, 0xf01ccf08, 0xf01ccf08, 0x24, 0x0) at
> netbsd:nmi_hard+0x108
> nmi_hard(0x0, 0x8000, 0x80000000, 0xf01698a4, 0x40400fc3, 0xf01f0a58) at
> netbsd:sparc_interrupt44c+0x2b4
> sparc_interrupt44c(0xf04f2800, 0xc, 0xc2, 0xf054b000, 0x0, 0xf0545748) at
> netbsd:ncr53c9x_select+0x234


it's a very nice stack trace! If you had posted it on the qemu mailing
list I wouldn't have to spend the last weekend trying to find out why
NetBSD 1.6-3.1 crashes under qemu.

Unfortunately I couldn't google for this message because the install
kernel crashed differently till I fixed NMI handling in qemu a couple
of weeks ago.

But I really wonder why did it work on the real SS10: there is an
obvious bug in ncr53c9x.c introduced in NetBSD 1.6 and fixed in 4.0. I
wonder if there is a counterpart bug in the hardware or documentation,
or it worked just due to a lucky timing, or it worked only in PIO
mode.

Do you still have the SS-10 machine with the installation? Can we do
some tests using a slightly modified ncr53c9x.c ?

Artyom

-- 
Regards,
Artyom Tarasenko

solaris/sparc under qemu blog: http://tyom.blogspot.com/