6-CORE Dunnington - FreeBSD

This is a discussion on 6-CORE Dunnington - FreeBSD ; Hiya Has anyone successfully tried installing and running FreeBSD on Intel's newest 6-CORE Dunnington? I was able to install 200809-CURRENT on it but after recompiling the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out that I could ...

+ Reply to Thread
Results 1 to 9 of 9

Thread: 6-CORE Dunnington

  1. 6-CORE Dunnington

    Hiya

    Has anyone successfully tried installing and running FreeBSD on
    Intel's newest 6-CORE Dunnington?

    I was able to install 200809-CURRENT on it but after recompiling
    the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out
    that I could not boot it anymore. What's weird is that I could not
    boot the same 200809-CURRENT CD that I used the first time. Just hangs
    on
    ....
    uhci0: [GIANT-LOCKED]
    uhci0: [ITHREAD]
    ....

    I've also tried installing via PXE and it always hangs on the above
    spot, even using a -CURRENT from Sept 26 and PXE-booting it. No go :-(

    Any thoughts? Thanks.


    --
    cheers
    mars
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  2. Re: 6-CORE Dunnington

    Mars, good day.

    Wed, Oct 01, 2008 at 11:36:43PM +0800, Mars G Miro wrote:
    > I was able to install 200809-CURRENT on it but after recompiling
    > the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out
    > that I could not boot it anymore. What's weird is that I could not
    > boot the same 200809-CURRENT CD that I used the first time.


    [Assuming you had not touched any hardware or BIOS configuration
    since last good boot from -CURRENT CD.]

    Was is the totally cold boot (with power-off via unplugging the power
    cord/switching off the master power input on the power supply) or just a
    some sort of a warm or semi-cold boot with power button? I had seen
    the cases where hardware was in a such bad state, that only totally
    cold boot was helping to recover.

    Just hangs
    > on
    > ....
    > uhci0: [GIANT-LOCKED]
    > uhci0: [ITHREAD]


    Any ways to disable (at least partially) USB stuff via BIOS? May
    be disabling other devices will help too -- try to play with the
    disabling the various controllers.
    --
    Eygene
    _ ___ _.--. #
    \`.|\..----...-'` `-._.-'_.-'` # Remember that it is hard
    / ' ` , __.--' # to read the on-line manual
    )/' _/ \ `-_, / # while single-stepping the kernel.
    `-'" `"\_ ,_.-;_.-\_ ', fsc/as #
    _.-'_./ {_.' ; / # -- FreeBSD Developers handbook
    {_.-``-' {_/ #

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v2.0.9 (FreeBSD)

    iEYEARECAAYFAkjkD9EACgkQthUKNsbL7YhMbACgk/Ioh0rR07UtEZg3c3E7agLU
    bqYAnjccR7FlMVAx6NqkiSkEsaqZPbOt
    =e+r6
    -----END PGP SIGNATURE-----


  3. Re: 6-CORE Dunnington

    Hello!

    Mars G Miro schrieb:
    > ....
    > uhci0: [GIANT-LOCKED]
    > uhci0: [ITHREAD]
    > ....
    >
    > I've also tried installing via PXE and it always hangs on the above
    > spot, even using a -CURRENT from Sept 26 and PXE-booting it. No go :-(


    Have you connected a self-powered USB hub to the system? I also
    noticed (on some other computers) major problems during cold boot
    when the USB hub provides power not only to its downstream ports but
    also to the upstream host port. This way the mainboard will get some
    power even when it should be completely turned off.

    It took me quite a long time to find out that nowadays cheap USB hubs
    do not have proper power supplies (e.g. a Schottky diode on the upstream
    port) which prevent such behaviour.

    Also some self-powered USB devices behave strange during enumeration
    done by the BIOS which also caused a hanging system during early boot
    stages.

    One of my computers (Asus P5Q Pro mainboard) didn't boot in all cases
    when the mouse was connected to a USB hub integrated into one of the
    TFT monitors. Connecting it directly to the PC (resp. it root hub)
    solved the problem. After a BIOS update the problem vanished.

    So you should try to disconnect all(!) USB devices from the computer
    in order to find out if they cause the problems.

    Regards
    Andreas Schweigstill


    --
    Dipl.-Phys. Andreas Schweigstill
    Schweigstill IT | Embedded Systems
    Schauenburgerstraße 116, D-24118 Kiel, Germany
    Phone: (+49) 431 5606-435, Fax: (+49) 431 5606-436
    Mobile: (+49) 171 6921973, Web: http://www.schweigstill.de/
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  4. Re: 6-CORE Dunnington

    On Thu, Oct 2, 2008 at 12:15 AM, Gavin Atkinson wrote:
    > On Wed, 2008-10-01 at 23:36 +0800, Mars G Miro wrote:
    >> Hiya
    >>
    >> Has anyone successfully tried installing and running FreeBSD on
    >> Intel's newest 6-CORE Dunnington?
    >>
    >> I was able to install 200809-CURRENT on it but after recompiling
    >> the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out
    >> that I could not boot it anymore. What's weird is that I could not
    >> boot the same 200809-CURRENT CD that I used the first time. Just hangs
    >> on
    >> ....
    >> uhci0: [GIANT-LOCKED]
    >> uhci0: [ITHREAD]
    >> ....
    >>
    >> I've also tried installing via PXE and it always hangs on the above
    >> spot, even using a -CURRENT from Sept 26 and PXE-booting it. No go :-(

    >
    > Firstly, does it seem to be detecting all six CPUs correctly? Are you
    > able to reboot with the old kernel and put a verbose dmesg and the
    > output of "acpidump -dt" up on the web somewhere? It would also be
    > interesting to know if anything else is printed after the above when it
    > hangs on a verbose boot.
    >



    I can't be sure that it detects the 6 cores. I did get a dmesg prior
    to rebuilding the world on it but I didnt get it off the box as I was
    working on rebuilding the world. I tried rebooting w/ the old kernel
    as well as, like I said, the 200809-CURRENT CD but it just hangs.

    btw, this is a 4-CPU 6-core (4 Physical CPUs, 6 Cores each)


    > There was a report back in July about FreeBSD not discovering all the
    > CPUs on the board, but there was no mention of hanging on boot.
    >


    Yes. I read that thread. The OP said he had a chance to test the
    6-core for a few hours. This box is in my hands for quite an
    indefinite amount of time ;-)

    Thanks.

    > Gavin
    >




    --
    cheers
    mars
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  5. Re: 6-CORE Dunnington

    Gavin Atkinson wrote:
    > On Wed, 2008-10-01 at 23:36 +0800, Mars G Miro wrote:
    >> Hiya
    >>
    >> Has anyone successfully tried installing and running FreeBSD on
    >> Intel's newest 6-CORE Dunnington?
    >>
    >> I was able to install 200809-CURRENT on it but after recompiling
    >> the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out
    >> that I could not boot it anymore. What's weird is that I could not
    >> boot the same 200809-CURRENT CD that I used the first time. Just hangs
    >> on
    >> ....
    >> uhci0: [GIANT-LOCKED]
    >> uhci0: [ITHREAD]
    >> ....
    >>
    >> I've also tried installing via PXE and it always hangs on the above
    >> spot, even using a -CURRENT from Sept 26 and PXE-booting it. No go :-(

    >
    > Firstly, does it seem to be detecting all six CPUs correctly? Are you
    > able to reboot with the old kernel and put a verbose dmesg and the
    > output of "acpidump -dt" up on the web somewhere? It would also be
    > interesting to know if anything else is printed after the above when it
    > hangs on a verbose boot.
    >
    > There was a report back in July about FreeBSD not discovering all the
    > CPUs on the board, but there was no mention of hanging on boot.


    That was me and I actually did discover a similar boot problem later.
    For me, it turned out that booting FreeBSD after booting Linux worked
    fine, but booting FreeBSD from cold-boot hung in a similar way.

    I didn't discover what the problem was with detecting all 6 cores but I
    might have been a buggy/development BIOS.


    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.6 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFI5IbUldnAQVacBcgRAgcaAJ9QdmfpSskfhgaefp2bkt KPWW3P5ACgkyxT
    2j6//Z8HV/N8SObcZSjtJJI=
    =WMmS
    -----END PGP SIGNATURE-----


  6. Re: 6-CORE Dunnington

    On Thu, Oct 2, 2008 at 4:25 PM, Mars G Miro wrote:
    > On Thu, Oct 2, 2008 at 8:03 AM, Eygene Ryabinkin wrote:
    >> Mars, good day.
    >>

    >
    > Yo
    >
    >> Wed, Oct 01, 2008 at 11:36:43PM +0800, Mars G Miro wrote:
    >>> I was able to install 200809-CURRENT on it but after recompiling
    >>> the kernel (taking out WITNESS, INVARIANTS KGDB et al) I found out
    >>> that I could not boot it anymore. What's weird is that I could not
    >>> boot the same 200809-CURRENT CD that I used the first time.

    >>
    >> [Assuming you had not touched any hardware or BIOS configuration
    >> since last good boot from -CURRENT CD.]
    >>
    >> Was is the totally cold boot (with power-off via unplugging the power
    >> cord/switching off the master power input on the power supply) or just a
    >> some sort of a warm or semi-cold boot with power button? I had seen
    >> the cases where hardware was in a such bad state, that only totally
    >> cold boot was helping to recover.
    >>

    >
    > Yes, I do this from time to time when I encounter hardware problems
    > like these. I've tried various BIOS settings and configurations,
    > changing to default BIOS values, I'd think I've scoured through all
    > the BIOS settings, frustratingly because the BIOS takes quite some
    > time to load (well this is a test platform, whaddya expect :-p)
    >
    > Cold-booting does not help either.
    >
    >
    >> Just hangs
    >>> on
    >>> ....
    >>> uhci0: [GIANT-LOCKED]
    >>> uhci0: [ITHREAD]

    >>
    >> Any ways to disable (at least partially) USB stuff via BIOS? May
    >> be disabling other devices will help too -- try to play with the
    >> disabling the various controllers.

    >
    > I've tried that one too.
    >
    >


    Someone mentioned the same boot hangup problems I've had and the 'fix'
    was to boot to some other OS first, like Linux. I tried Slackware 12.1
    (also hangs somewhere during boot) and then Ubuntu 8.04 (am able to go
    into the install screen) but rebooting back to freebsd is no go. Then
    I recalled we were able to successfully install 64-bit Windows 2003
    Server on it. So I booted it to the windows 2003 server just to the
    point where it asks to install it, then rebooted it to the already
    installed FreeBSD.

    verbose dmesgs
    of FreeBSD-200809 CURRENT: http://pastebin.com/f367b3203
    of FreeBSD-20080925 CURRENT: http://pastebin.com/f4338460a

    I have also an output of acpidump but its quite huge, pastebin has limits:
    -rw-r--r-- 1 mars staff - 398105 Oct 9 16:38 DUNNINGTON-acpidump-td.txt
    Email me privately if you want this (or I can give this to someone w/
    ample bandwidth and a web server).

    Also it detects only 16CPUs instead of 24. Win2k3 detects all 24.

    make -j16 buildworld: 4442.721u 3236.681s 16:29.50 776.0%
    6064+6203k 539+8258io 15117pf+0w

    But I have an HP DL380 Quad-Core Xeon that rebuilds the world in just
    13 minutes :-p

    Thanks.

    >> --
    >> Eygene
    >> _ ___ _.--. #
    >> \`.|\..----...-'` `-._.-'_.-'` # Remember that it is hard
    >> / ' ` , __.--' # to read the on-line manual
    >> )/' _/ \ `-_, / # while single-stepping the kernel.
    >> `-'" `"\_ ,_.-;_.-\_ ', fsc/as #
    >> _.-'_./ {_.' ; / # -- FreeBSD Developers handbook
    >> {_.-``-' {_/ #
    >>

    >
    >
    >
    > --
    > cheers
    > mars
    >




    --
    cheers
    mars
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  7. Re: 6-CORE Dunnington

    Mars G Miro wrote:

    > Someone mentioned the same boot hangup problems I've had and the 'fix'
    > was to boot to some other OS first, like Linux. I tried Slackware 12.1


    That was me...

    > (also hangs somewhere during boot) and then Ubuntu 8.04 (am able to go
    > into the install screen) but rebooting back to freebsd is no go. Then


    Hmm, Ubuntu (64-bit) worked for me.

    > I recalled we were able to successfully install 64-bit Windows 2003
    > Server on it. So I booted it to the windows 2003 server just to the
    > point where it asks to install it, then rebooted it to the already
    > installed FreeBSD.


    So the problem looks like it has something to do with hardware
    initialization.

    > Also it detects only 16CPUs instead of 24. Win2k3 detects all 24.


    On Wed Oct 1 21:59:04 2008 John Baldwin increased the limit on CPUs to
    32 for both AMD64 and i386. There is ongoing work to increase this to 64
    CPUs in the 64-bit version.

    >
    > make -j16 buildworld: 4442.721u 3236.681s 16:29.50 776.0%
    > 6064+6203k 539+8258io 15117pf+0w
    >
    > But I have an HP DL380 Quad-Core Xeon that rebuilds the world in just
    > 13 minutes :-p


    21 minutes with 2x4-core 1.86 GHz Xeon



    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.6 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFI7fUbldnAQVacBcgRAo1WAKDNTOGju5mfQyr9dqbOiy 1w/JhY4QCghM0E
    WDZK+Ca82sWEIoaHOb1lE94=
    =OwpD
    -----END PGP SIGNATURE-----


  8. Re: 6-CORE Dunnington

    On Thu, Oct 9, 2008 at 7:32 PM, Max Laier wrote:
    [snip]
    >>
    >> Also it detects only 16CPUs instead of 24. Win2k3 detects all 24.

    >
    > This should be fixed with the following commit:
    > http://svn.freebsd.org/changeset/base/183525
    >


    Cool. http://pastebin.com/f509d9e94

    [in a separate email ..]
    >> I recalled we were able to successfully install 64-bit Windows 2003
    >> Server on it. So I booted it to the windows 2003 server just to the
    >> point where it asks to install it, then rebooted it to the already
    >> installed FreeBSD.


    >So the problem looks like it has something to do with hardware initialization.



    So it seems.


    >> make -j16 buildworld: 4442.721u 3236.681s 16:29.50 776.0%
    >> 6064+6203k 539+8258io 15117pf+0w
    >>
    >> But I have an HP DL380 Quad-Core Xeon that rebuilds the world in just
    >> 13 minutes :-p
    >>


    > 21 minutes with 2x4-core 1.86 GHz Xeon


    new buildworld stats on 4CPU 6-Core DUNNINGTON (24 CPUs):
    make -j8 buildworld: 4157.424u 2675.440s 19:43.60 577.2%
    6010+6465k 498+8312io 15123pf+0w
    make -j16 buildworld: 4680.518u 6933.141s 19:24.25 997.5%
    5794+5492k 436+8312io 15123pf+0w
    make -j24 buildworld: 5243.507u 10805.515s 20:59.02 1274.7%
    5686+5060k 16761+8455io 16821pf+0w

    Thanks guys.

    >> Thanks.
    >>




    --
    cheers
    mars
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  9. Re: 6-CORE Dunnington

    On Fri, Oct 10, 2008 at 4:36 PM, Ivan Voras wrote:
    > 2008/10/10 Mars G Miro :
    >
    >>> 21 minutes with 2x4-core 1.86 GHz Xeon

    >>
    >> new buildworld stats on 4CPU 6-Core DUNNINGTON (24 CPUs):

    >
    >> make -j8 buildworld: 4157.424u 2675.440s 19:43.60 577.2%
    >> 6010+6465k 498+8312io 15123pf+0w

    >
    >> make -j16 buildworld: 4680.518u 6933.141s 19:24.25 997.5%
    >> 5794+5492k 436+8312io 15123pf+0w

    >
    >> make -j24 buildworld: 5243.507u 10805.515s 20:59.02 1274.7%
    >> 5686+5060k 16761+8455io 16821pf+0w

    >
    > If I'm parsing this correctly the time slightly increases with the
    > number of CPUs used?
    >


    That's what it says.

    Also, the 16-minute buildworld clock was from a rebuild of
    Sep25-CURRENT (csup'd to around this time, rebooted, then clock
    rebuilding the world) , so I don't know how much code from then up
    until the recent that it takes ~ 4 minutes longer now.


    --
    cheers
    mars
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/lis...reebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


+ Reply to Thread