RE: [suse-sparc] kernel compile fails on ultrasparc64

From: Leif Sawyer (lsawyer@gci.com)
Date: Wed Nov 01 2000 - 12:29:02 PST

  • Next message: Joshua Uziel: "Re: [suse-sparc] kernel compile fails on ultrasparc64"

    Message-ID: <BF9651D8732ED311A61D00105A9CA31502EF8054@berkeley.gci.com>
    From: Leif Sawyer <lsawyer@gci.com>
    Date: Wed, 1 Nov 2000 11:29:02 -0900 
    Subject: RE: [suse-sparc] kernel compile fails on ultrasparc64
    

    I'm consolidating my replies:

    Thorsten Kukuk [kukuk@suse.de] replies:
    > To what I wrote:
    >
    > > I've tried everything I can (editing makefiles,etc) in
    > > hopes of getting a new kernel compiled (like 2.4.0-pre9
    > > + reiserfs) so that I can do the things i'd like to do..
    >
    > A kernel with reiserfs cannot work on SPARC. The reiserfs people are
    > working on a 64bit version (1. step you need), but the whole code
    > does not compile on big-endian machines (second step you need).
    >

    Yeah, i realize that the sparc port isn't available yet, however
    the Alpha port is passing tests, so I thought perhaps I'd start
    looking into getting it running on the sparc. Just something to play
    around with and hopefully learn a little about sparc assembly.

    Joshua Uziel [uzi@suse.com] responds
    > to My rantings:
    > > Just joined the list and am wondering something:
    > >
    > > I've tried a couple of different distro's with my
    > > UltraSparc-2, and in no case am i able to successfully
    > > compile a kernel.
    >
    > You need to elaborate more on this. Are you not installing the 64-bit
    > compiler for building a sparc64 kernel? This is _absolutely_ required
    > if you're going to do so.

    Oh, so gcc out of the box is only 32 bit, and I've got down find the 64-bit
    version somewhere? Okay. Be nice if this was blatently documented, but I
    won't belabour the point. (i.e., the faq at ultralinux.org doesn't mention
    a specific requirement for a 64-bit gcc, only GCC version 2.7.x, or EGCS
    1.0.3 and up.
    Or the fact that when installing, there's no mention of sparc64-gcc in the
    default.
    But I am endeavoring to install off of the ISO images. I'll see how
    2.4.0-test10
    (released today) works out.
     
    > > Also, I receive kernel oops' when i try to initialize my hme
    > > interface. This is using the stock suse kernel that comes with the
    > > latest ISO.
    >
    > More info need on this as well, please.

     
    > > Forgive me, as I have yet to do a full suse install -- I
    > > have to have the cd burned.
    >
    > So you're using the SuSE kernel on the Mandrake distribution?
    > Nonetheless, you need to add technical information to bug reports.

    Just wiped the mandrake distro in order to install the Suse.
    I have solaris 8 on /dev/sda, and am installing linux to /dev/sdb

    During initial configuration, YaST reports that my SunHME will be
    configured, but a dialog box pops up:

    Device initialization failed, Error code 255, Module name sunhme

    tty10 shows:
    insmod: /lib/modules/2.2.16/net/sunhme.o: init_module: Device or resource
    busy
    insmod: /lib/modules/2.2.16/net/sunhme.o: insmod
    /lib/modules/2.2.16/net/sunhme.o failed
    insmod: /lib/modules/2.2.16/net/sunhme.o: insmod sunhme failed

    Please note that when I boot into solaris, my HME interface works fine.
    (well enough to download the suse ISO's)
    And as side note, i also have two QFE cards installed as well.

    ( It seems that if there are > 5 interfaces when ifconfig is run,
     a kernel oops happens: )

    So I reboot, and edit modules.conf to disable
    ax25, ipx, appletalk, netrom, ipv6, and rose.

    I don't see pcmcia, so i don't worry about trying to disable it. :-)

    Also, i disabled the pc_hardware parallel module. Looks like
    somebody missed that one.. :-)

    Upon reboot, i log in and ifconfig -a, and I get the standard output
    from /dev/lo

    # uname -rm
    2.2.16 sparc64

    # ifconfig hme0
    hme0: error fetching interface information: Device not found

    # modprobe sunhme
    /lib/modules/2.2.16/net/sunhme.o: init_module: Device or resource busy
    Hint: this error can be caused by incorrect [....]

    # ifconfig -a
    Killed

    and more kernel oops's.
    Here's the full output from ksymoops:

    Unable to handle kernel paging request at virtual address 000000000101e000
    tsk->mm->context = 0000000000000585
    tsk->mm->pgd = fffff8003e242000
                  \|/ ____ \|/
                  "@'/ .. \`@"
                  /_| \__/ |_\
                     \__U_/
    ifconfig(400): Oops
    TSTATE: 0000000000009601 TPC: 0000000000431258 TNPC: 00000000004312e0 Y:
    00000000
    Using defaults from ksymoops -t elf32-sparc -a sparc
    g0: fffff8003e1c8000 g1: 0000000000000000 g2: ffffffffffffffff g3:
    0000000000000000
    g4: fffff80000000000 g5: 0000000000000000 g6: fffff8003e1c8000 g7:
    0000000000000000
    o0: 000000000101e640 o1: 000000000101e000 o2: fffff800001fd0e0 o3:
    0000000070000000
    o4: 0000000000000080 o5: 0000000000595c26 sp: fffff8003e1cb171 ret_pc:
    00000000004310a0
    l0: fffff8003e3ecd40 l1: 0000000000000000 l2: fffff800001fd108 l3:
    0000000077000000
    l4: 0000000000000000 l5: 0000000000000000 l6: 0000000000000000 l7:
    0000000000000000
    i0: fffff8003e1cbac0 i1: 000000000101e000 i2: 0000000000000000 i3:
    0000000000595d10
    i4: 000000002e310047 i5: 0000000000595d10 i6: fffff8003e1cb241 i7:
    0000000000408ce0
    Caller[0000000000408ce0]
    Caller[00000000004a3190]
    Caller[00000000004a32b4]
    Caller[0000000000485360]
    Caller[0000000000452064]
    Caller[000000000040fc94]
    Caller[000000007007f158]
    Instruction DUMP: d65e2088 d05e2088 10680023 <e2020000> d059a7e0 9332f020
    920a6ffc d45a2018 d2028009

    >>PC; 00431258 <do_sparc64_fault+218/5a0> <=====
    >>O7; 004310a0 <do_sparc64_fault+60/5a0>
    >>I7; 00408ce0 <sparc64_realfault_common+18/28>
    Trace; 00408ce0 <sparc64_realfault_common+18/28>
    Trace; 004a3190 <sprintf_stats+10/100>
    Trace; 004a32b4 <dev_get_info+34/a0>
    Trace; 00485360 <proc_file_read+80/180>
    Trace; 00452064 <sys_read+e4/100>
    Trace; 0040fc94 <linux_sparc_syscall+34/40>
    Trace; 7007f158 <END_OF_CODE+6f0695b0/????>
    Code; 0043124c <do_sparc64_fault+20c/5a0>
    0000000000000000 <_PC>:
    Code; 0043124c <do_sparc64_fault+20c/5a0>
       0: d6 5e 20 88 unknown
    Code; 00431250 <do_sparc64_fault+210/5a0>
       4: d0 5e 20 88 unknown
    Code; 00431254 <do_sparc64_fault+214/5a0>
       8: 10 68 00 23 unknown
    Code; 00431258 <do_sparc64_fault+218/5a0> <=====
       c: e2 02 00 00 ld [ %o0 ], %l1 <=====
    Code; 0043125c <do_sparc64_fault+21c/5a0>
      10: d0 59 a7 e0 unknown
    Code; 00431260 <do_sparc64_fault+220/5a0>
      14: 93 32 f0 20 unknown
    Code; 00431264 <do_sparc64_fault+224/5a0>
      18: 92 0a 6f fc and %o1, 0xffc, %o1
    Code; 00431268 <do_sparc64_fault+228/5a0>
      1c: d4 5a 20 18 unknown
    Code; 0043126c <do_sparc64_fault+22c/5a0>
      20: d2 02 80 09 ld [ %o2 + %o1 ], %o1

    > > hoping the audience is listening, Leif
    >
    > It is.

    And thanks! There's much quicker (and less noisy) response
    here than any other sparc/linux list i've been on.

    I'm now more on my way to having a usable system than i was
    a couple of days ago. yay!



    This archive was generated by hypermail 2.1.0 : Wed Jun 06 2001 - 11:08:14 PDT