headers
Patrick Welche | 9 Jul 2009 19:48
Picon
Picon
Favicon

amd64 on SU9400

I can netboot NetBSD-current/i386 on a system with a SU9400 which according
to http://ark.intel.com/Product.aspx?id=36697 has a 64-bit instruction set,
but not NetBSD-current/amd64.

Despite having NFS, DIAGNOSTIC, DEBUG, LOCKDEBUG, DDB, DDB_ONPANIC=1 in the
kernel, after "Loading nfs" and the /|\ animation the laptop simply reboots.

Any ideas given that works with i386?

Cheers,

Patrick
Permalink | Reply | 6 comments |
headers
Joerg Sonnenberger | 9 Jul 2009 20:04
Picon

Re: amd64 on SU9400

On Thu, Jul 09, 2009 at 06:48:59PM +0100, Patrick Welche wrote:
> I can netboot NetBSD-current/i386 on a system with a SU9400 which according
> to http://ark.intel.com/Product.aspx?id=36697 has a 64-bit instruction set,
> but not NetBSD-current/amd64.

What is the output of
	cpuctl identify 0
on NetBSD/i386?

Joerg
Permalink | Reply | 1 comment |
headers
Patrick Welche | 9 Jul 2009 20:06
Picon
Picon
Favicon

Re: amd64 on SU9400

On Thu, Jul 09, 2009 at 08:04:07PM +0200, Joerg Sonnenberger wrote:
> On Thu, Jul 09, 2009 at 06:48:59PM +0100, Patrick Welche wrote:
> > I can netboot NetBSD-current/i386 on a system with a SU9400 which according
> > to http://ark.intel.com/Product.aspx?id=36697 has a 64-bit instruction set,
> > but not NetBSD-current/amd64.
> 
> What is the output of
> 	cpuctl identify 0
> on NetBSD/i386?

Luckily I grabbed a copy earlier :-)

cpu0: Intel Core 2 Extreme (686-class), 1396.63 MHz, id 0x10676
cpu0: features  0xbfebfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR>
cpu0: features  0xbfebfbff<PGE,MCA,CMOV,PAT,PSE36,CFLUSH,DS,ACPI,MMX,FXSR>
cpu0: features  0xbfebfbff<SSE,SSE2,SS,HTT,TM,SBF>
cpu0: features2 0x8e3fd<SSE3,DTES64,MONITOR,DS-CPL,VMX,SMX,EST,TM2,SSSE3>
cpu0: features2 0x8e3fd<CX16,xTPR,PDCM,SSE41>
cpu0: features3 0x20100000<XD,EM64T>
cpu0: features4 0x1<LAHF>
cpu0: "Intel(R) Core(TM)2 Duo CPU     U9400   <at>  1.40GHz"
cpu0: I-cache 32KB 64B/line 8-way, D-cache 32KB 64B/line 8-way
cpu0: L2 cache 3MB 64B/line 12-way
cpu0: ITLB 128 4KB entries 4-way
cpu0: DTLB 256 4KB entries 4-way, 16 4MB entries 4-way
cpu0: Initial APIC ID 0
cpu0: Cluster/Package ID 0
cpu0: Core ID 0
cpu0: family 06 model 07 extfamily 00 extmodel 01
(Continue reading)

Permalink | Reply | 0 comments |
headers
Patrick Welche | 10 Jul 2009 17:44
Picon
Picon
Favicon

Re: amd64 on SU9400

On Thu, Jul 09, 2009 at 06:48:59PM +0100, Patrick Welche wrote:
> I can netboot NetBSD-current/i386 on a system with a SU9400 which according
> to http://ark.intel.com/Product.aspx?id=36697 has a 64-bit instruction set,
> but not NetBSD-current/amd64.
> 
> Despite having NFS, DIAGNOSTIC, DEBUG, LOCKDEBUG, DDB, DDB_ONPANIC=1 in the
> kernel, after "Loading nfs" and the /|\ animation the laptop simply reboots.
> 
> Any ideas given that works with i386?

It works with a kernel compiled from today's source. Could

revision 1.31
date: 2009/07/09 21:43:16;  author: rmind;  state: Exp;  lines: +9 -1
- Fix rare crashe in the intr_lapic_tlb_bcast() handler: save and setup
  %fs on i386, %gs on amd64 registers, before using them.  Otherwise, it
  might be invalid/garbage, eg. IPI can interrupt userspace.

- Explicitly initialize per-CPU emap generation number.

Thanks <drochner> for reporting and testing of patch.

this have fixed it?!

Cheers,

Patrick
Permalink | Reply | 0 comments |
headers
neys | 16 Jul 2009 09:41

security.conf and xdm/xfs

hello,

i set up a box with netbsd 5.0 amd64. there are no X-components installed, 
therefore the files /etc/rc.d/xdm and /etc/rc.d/xfs are missing. so security 
checks from /etc/daily always claim:

Checking special files and directories.
missing: ./etc/rc.d/xdm
missing: ./etc/rc.d/xfs

because of this part from man security.conf(5):

The following mtree(8) tags modify how files are determined from 
/etc/mtree/special and /etc/mtree/special.local:
     exclude  The entry is ignored; no backups are made
              and the differences are not displayed.
              This includes dynamic or binary files such
              as /var/run/utmp.

     nodiff   The entry is backed up but the differences
              are not displayed because the contents of
              the file are sensitive.  This includes
              files such as /etc/master.passwd.

i tried to block the check of those files by adding the following two lines 
to /etc/mtree/special.local:

./etc/rc.d/xdm type=file mode=0555 tags=exclude
./etc/rc.d/xfs type=file mode=0555 tags=exclude
(Continue reading)

Permalink | Reply | 2 comments |
headers
David Brownlee | 24 Jul 2009 01:06
Gravatar

Switching from 5.0 i386 to amd64

Just done this on my main server and laptop and hit a few issues, but 
overall went quite well.

I ended up switching the swap partition to ffs, extracting amd64 into it 
and then booting it, having a quick test & then copying everything except 
/etc across and rerunning MAKEDEV & bootblocks.

Just some notes on the experience:
- needed to remove old i386 /etc/ld.so.conf to avoid a harmless libm 
warning
- devel/p5-Compress-Raw-Bzip2 had some really bizarre PLIST issues which 
now seem to have gone
- postgres needs 8 byte rather than 4 byte alignment (just dump in i386 & 
restore in a md64)
- cannot build jdk15 as gcc3 will not build on amd64
- alpine segvs whenever it tries to open a link or run an external program 
on an attachment
- wine-devel will not build

But otherwise I have gnome, bugzilla, trzc, apache, mediawiki, openoffice3, 
exim, dovecot, roundcube & many other apps running wonderfully.

Kudos to NetBSD, pkgsrc, and many third party developers :)

I had a stint in the 64bit world of alpha & sparc64 some (too many) years 
ago, but have languished in 32bit i386 for too long.
Permalink | Reply | 0 comments |
headers
Manuel Bouyer | 28 Jul 2009 20:59

Re: security.conf and xdm/xfs

On Thu, Jul 16, 2009 at 09:41:31AM +0200, neys wrote:
> hello,
> 
> i set up a box with netbsd 5.0 amd64. there are no X-components installed, 
> therefore the files /etc/rc.d/xdm and /etc/rc.d/xfs are missing. so 
> security checks from /etc/daily always claim:
> 
> Checking special files and directories.
> missing: ./etc/rc.d/xdm
> missing: ./etc/rc.d/xfs
> 
> 
> because of this part from man security.conf(5):
> 
> The following mtree(8) tags modify how files are determined from 
> /etc/mtree/special and /etc/mtree/special.local:
>     exclude  The entry is ignored; no backups are made
>              and the differences are not displayed.
>              This includes dynamic or binary files such
>              as /var/run/utmp.
> 
>     nodiff   The entry is backed up but the differences
>              are not displayed because the contents of
>              the file are sensitive.  This includes
>              files such as /etc/master.passwd.
> 
> i tried to block the check of those files by adding the following two lines 
> to /etc/mtree/special.local:
> 
> ./etc/rc.d/xdm type=file mode=0555 tags=exclude
(Continue reading)

Permalink | Reply | 1 comment |
headers
Torsten Harenberg | 29 Jul 2009 13:09
Picon
Favicon

Large disks with gpt on amd64

Dear list,

sorry, this maybe the wrong place to ask (I'm not sure if this is 
port-specific), but let me try it here first:

I have several HP DL380G5 with MSA 60 RAID arrays attached and try to 
use them under NetBSD. To be on the safe side, I tried the Jibbed live 
distribution before installing.

The first results were promising, all relevant hardware was detected 
with the shipped kernel, especially the logical devices on these RAID 
arrays with ~10TB each:

# dmesg | grep ^sd
sd0 at scsibus0 target 0 lun 0: <HP, LOGICAL VOLUME, 5.26> disk fixed
sd0: 136 GB, 35132 cyl, 255 head, 32 sec, 512 bytes/sect x 286677120 sectors
sd1 at scsibus1 target 0 lun 0: <HP, LOGICAL VOLUME, 5.26> disk fixed
sd1: 9314 GB, 65535 cyl, 255 head, 1168 sec, 512 bytes/sect x 
19534592432 sectors
sd1: mbr partition exceeds disk size
sd1: GPT GUID: 5de529bf-8519-4d58-a677-a195517f3159
sd2 at scsibus1 target 1 lun 0: <HP, LOGICAL VOLUME, 5.26> disk fixed
sd2: 9314 GB, 65535 cyl, 255 head, 1168 sec, 512 bytes/sect x 
19534592432 sectors
sd2: GPT GUID: e53773a0-dffc-4fea-8afa-f6af15d77ef2
sd3 at scsibus1 target 2 lun 0: <HP, LOGICAL VOLUME, 5.26> disk fixed
sd3: 9314 GB, 65535 cyl, 255 head, 1168 sec, 512 bytes/sect x 
19534592432 sectors
sd3: GPT GUID: 14019865-fbec-48a6-875a-e103510cba99
sd3: bad GPT partition array CRC
(Continue reading)

Permalink | Reply | 9 comments |
headers
Joerg Sonnenberger | 29 Jul 2009 13:15
Picon

Re: Large disks with gpt on amd64

On Wed, Jul 29, 2009 at 01:09:40PM +0200, Torsten Harenberg wrote:
> Now I want to use them inside NetBSD, but mbrlabel already shows strange  
> partition sizes:

You have to use GPT for disks > 2TB. You should have dk devices for all
GPT entries. The gpt(8) command can be used for changing the disk,
though you might run into some issues the first time you convert from
MBR/disklabel to GPT as in "might need a reboot".

> And speaking about large disks, what would be an appropreate file system?

UFS2 is the only option for > 2TB right now.

Joerg
Permalink | Reply | 8 comments |
headers
Torsten Harenberg | 29 Jul 2009 15:10
Picon
Favicon

Re: Large disks with gpt on amd64

Hi Joerg,

thanks a lot for your quick help, which worked perfectly..

I'm facing another problem now.

With this live cd, after I gpt'ed, newfs'ed, ...  my ~10 TB file system, 
the system get a hard reset a few seconds after I mounted it. Nothing is 
printed on the root console.

So I installed NetBSD properly on the machine, which consists of a

p400 raid controller (136 GB, recognized as "sd0" by sysinst) and a
p800 raid controller (with 4x10TB, recognized as "sd1".."sd4" by sysinst).

Obviously I wanted to install to the 136 GB device which I did.

sysinst finished without problems, but after re-booting, the boot loader 
seems not to find the partitions created anymore (I created one 136 GB 
partition on "sd0", and disklabeled it with a ~100 GB "/" - sd0a - and 
some swap as sd0b).

Messages are like:

booting hd0a:netbsd - starting in 0
open netbsd: bad partition
....
....
....
(Continue reading)

Permalink | Reply | 5 comments |

Gmane