This technique won't work in a PV domU since as I mentioned we don't
allow inb/outb or access to the real physical hardware. It would still
work in an HVM domU since in that case there is a simulated real time
clock device.
Margot Miller wrote:
> Yes, we are using outb()'s. We are using the real time clock to generate a
> random number to help generate a random number for hostid.
>
> + static int32_t
> + set_soft_hostid(void)
> + {
> + struct _buf *file;
> + char tokbuf[MAXNAMELEN];
> + token_t token;
> + int done = 0;
> + int i;
> + u_longlong_t tmp;
> + int32_t hostid = -1;
> + unsigned char reg, *c;
> + struct rtc_t rtc;
> + todinfo_t tod;
> + + /*
> + * If /etc/hostid file not found, we'd like to get a pseudo
> + * random number to use at the hostid. A nice way to do this
> + * is to read the real time clock, but if it's invalid, we
> + * just punt, and use the value of tenmicrodata, hoping it
> + * is random enough.
> + */
> + + if ((file = kobj_open_file(hostid_file)) == (struct _buf
> *)-1) {
> + /* hostid file not found */
> + + /* stolen from todpc_rtcget() and friends */
> + mutex_enter(&tod_lock);
> + outb(RTC_ADDR, RTC_D);
> + reg = inb(RTC_DATA);
> + if ((reg & RTC_VRT) == 0) { /* clock invalid */
> + hostid = (int32_t)tenmicrodata; /* punt */
> + } else {
> + for (i = 0, c = (unsigned char *)&rtc;
> + i < RTC_NREG; i++) {
> + outb(RTC_ADDR, i);
> + *c++ = inb(RTC_DATA);
> + }
> + + tod.tod_year = BCD_TO_BYTE(rtc.rtc_yr);
> + if (tod.tod_year < 69)
> + tod.tod_year += 100;
> + + tod.tod_month = BCD_TO_BYTE(rtc.rtc_mon);
> + tod.tod_day = BCD_TO_BYTE(rtc.rtc_dom);
> + /* day of week < 10; no conversion needed */
> + tod.tod_dow = rtc.rtc_dow;
> + tod.tod_hour = BCD_TO_BYTE(rtc.rtc_hr);
> + tod.tod_min = BCD_TO_BYTE(rtc.rtc_min);
> + tod.tod_sec = BCD_TO_BYTE(rtc.rtc_sec);
> + hostid = (int32_t)tod_to_utc(tod);
> + }
> + mutex_exit(&tod_lock);
> + } else {
> + /* hostid file found */
> ...
>
>
>
>
>
> Stuart Maybee wrote:
>>
>> Solaris currently does not support assigning devices to domU and thus
>> also at this time does not permit domU to perform in/out instructions.
>> You crashed because you attempted an out instruction.
>> Did your changes include adding i/o instructions to Solaris domU?
>> There is an outb in startup_modules() but it is ifdef'd out for an
>> i86xpv build. Are you sure you attempted to boot an i86xpv kernel in
>> your domU?
>>
>> Margot H. Miller wrote:
>>> We based our Solaris build on build 80 and made some minor kernel
>>> changes for our project
>>> and in the install consolidation and wanted to test them with xen to
>>> ensure we didn't break anything. I installed our modified build 80
>>> and then tried to install that same iso on a domU. It crashed as
>>> seen below. Reproducible. Then used the build 86 iso from internal
>>> Sun server nana
>>> and the domU install is working now. We are going to resync with the
>>> latest xen changes and rebuild but wanted to know if anything below
>>> jumps out or if anyone has any trouble shooting tips.
>>> Thanks,
>>> Margot
>>>
>>> # virt-install
>>> What is the name of your virtual machine? dom-1
>>> How much RAM should be allocated (in megabytes)? 500
>>> What would you like to use as the disk (path)? /export/home/dom1.img
>>> How large would you like the disk (/export/home/dom1.img) to be (in
>>> gigabytes)? 10
>>> Would you like to enable graphics support? (yes or no) no
>>> What is the install location? /export/home/solarisdvd.iso
>>>
>>>
>>> Starting install...
>>> Creating storage file... 100% |=========================| 10 GB
>>> 00:00 Creating
>>> domain... 0 B 00:20
>>> v3.0.4-1-xvm chgset 'Tue Dec 04 09:56:10 2007 +0000 13231:11a40bfd0ad8'
>>> SunOS Release 5.11 Version psylocke:onnv-hostid:01-25-2008 32-bit
>>> Copyright 1983-2007 Sun Microsystems, Inc. All rights reserved.
>>> Use is subject to license terms.
>>> DEBUG enabled
>>>
>>> panic[cpu0]/thread=f502c620: BAD TRAP: type=d (#gp General
>>> protection) rp=f50571d0 addr=f502c620
>>>
>>> #gp General protection
>>> addr=0xf502c620
>>> pid=0, pc=0xf4c4b219, sp=0xf4c2e0e1, eflags=0x10086
>>> cr0: 80050033<pg,wp,ne,et,mp,pe> cr4: 660<xmme,fxsr,mce,pae>
>>> cr2: 0
>>> gs: 1b1 fs: f5050000 es: f5050161 ds: f5050161
>>> edi: 0 esi: f502c620 ebp: f505738c esp: f5057208
>>> ebx: ffffffff edx: f5020070 ecx: f50b0258 eax: d
>>> trp: d err: 0 eip: f4c4b219 cs: 159
>>> efl: 10086 usp: f4c2e0e1 ss: 70
>>>
>>> cpu address timestamp type vc handler pc
>>> 0 f505b20c 179d4e29b6d8 trap 400 ast cmntrap+183
>>> 0 f505b164 179d4e29a188 trap d #gp outb+9
>>>
>>> f505711c unix:die+105 (d, f50571d0, f502c6)
>>> f50571bc unix:trap+3b5 (f50571d0, f502c620,)
>>> f50571d0 unix:cmntrap+18b (1b1, f5050000, f505)
>>> f505738c unix:outb+9 (0, f502c620, f502bd)
>>> f50573b8 unix:startup_modules+d7 (f50573e0, f4d9438b,)
>>> f50573c0 unix:startup+35 (f4c00010, f50531e8,)
>>> f50573e0 genunix:main+5b ()
>>>
>>> skipping system dump - no dump device configured
>>> rebooting...
>>>
>>>
>>> This message posted from opensolaris.org
>>> _______________________________________________
>>> xen-discuss mailing list
>>> [email protected]
>>
>
_______________________________________________
xen-discuss mailing list
[email protected]
