oops with reiserfs and kernel 2.5.4-pre1 on sparc64

oops with reiserfs and kernel 2.5.4-pre1 on sparc64

[Luigi Genoni]

HI,
I was trying 2.5.4-pre1 on sparc64 sun4u ultra2,
with 1GB memory and scsi controller ess.
This system has just one CPU.

I get this oops while trying to mount the /usr filesystem: which is on
reiserFS:

mount(40): Kernel bad sw trap 5
TSTATE: 0000000011009607 TPC: 000000000046cf7c TNPC: 000000000046cf80 Y: 07000000    Not tainted
g0: fffff8006ff057a0 g1: 000000000003775d g2: 0000000000000018 g3: 0000000000000001
g4: fffff80000000000 g5: 0000000000000001 g6: fffff8006f074000 g7: 0000000000000001
o0: 000000000ffffe00 o1: 0000000000001e00 o2: ffffffffffffffff o3: 0000000000001fff
o4: 0000000000040a00 o5: fffff8006fe09908 sp: fffff8006f076d81 ret_pc: 000000000046cf38
l0: 0000000000009908 l1: fffff8006f79e2a0 l2: 00000000005c44d8 l3: 00000000005c4908
l4: 00000000005cc800 l5: fffff8006ef05e60 l6: 000000007002aa38 l7: 0000000070029ef0
i0: fffff8006f79e2a0 i1: 0000000000002012 i2: 0000000010000000 i3: 0000000000000000
i4: fffff8006f808f78 i5: 00000001fffda000 i6: fffff8006f076e81 i7: 000000000046b260
Caller[000000000046b260]
Caller[000000000046b56c]
Caller[00000000004d1374]
Caller[00000000004b3540]
Caller[000000000046f124]
Caller[000000000046f3cc]
Caller[00000000004828cc]
Caller[0000000000482c08]
Caller[000000000042fa28]
Caller[0000000000410974]
Caller[000000000001274c]
Instruction DUMP: 9006be00  92126200  80a20009 <99d02005> 9002a001
953a2000  932e800a  80a2400b  084ffffd
Error (Oops_bfd_perror): scan_arch for specified architecture Success

>>TPC; 0046cf7c <grow_buffers+5c/e0>   <=====
>>O7;  0046cf38 <grow_buffers+18/e0>
>>I7;  0046b260 <__getblk+20/60>
Trace; 0046b260 <__getblk+20/60>
Trace; 0046b56c <__bread+c/a0>
Trace; 004d1374 <journal_init+214/8e0>
Trace; 004b3540 <reiserfs_read_super+140/600>
Trace; 0046f124 <get_sb_bdev+224/2c0>
Trace; 0046f3cc <do_kern_mount+4c/160>
Trace; 004828cc <do_add_mount+6c/1a0>
Trace; 00482c08 <do_mount+168/1a0>
Trace; 0042fa28 <sys32_mount+108/160>
Trace; 00410974 <linux_sparc_syscall32+34/40>
Trace; 0001274c Before first symbol

I am willing to test any patch.
I hope this helps

Luigi Genoni

Re: [reiserfs-dev] oops with reiserfs and kernel 2.5.4-pre1 on sparc64

[Oleg Drokin]

Hello!

On Wed, Feb 06, 2002 at 06:53:47PM +0100, Luigi Genoni wrote:

> I was trying 2.5.4-pre1 on sparc64 sun4u ultra2,
> with 1GB memory and scsi controller ess.
> This system has just one CPU.
> >>TPC; 0046cf7c <grow_buffers+5c/e0>   <=====
> >>O7;  0046cf38 <grow_buffers+18/e0>
> >>I7;  0046b260 <__getblk+20/60>
> Trace; 0046b260 <__getblk+20/60>
> Trace; 0046b56c <__bread+c/a0>
> Trace; 004d1374 <journal_init+214/8e0>
This looks like VFS have exploded for whatever reson, I think
because one ot these checks failed:
linux/fs/buffer.c:
        if (size & (get_hardsect_size(to_kdev_t(bdev->bd_dev))-1))
                BUG();
        /* Size must be within 512 bytes and PAGE_SIZE */
        if (size < 512 || size > PAGE_SIZE)
                BUG();

Can any of sparc64 people comment on what kind of exception will one
get for __builtin_trap?
Second BUG() seems to be out of the question, though.
Do you have CONFIG_DEBUG_BUGVERBOSE enabled?

Bye,
    Oleg

Re: [reiserfs-dev] oops with reiserfs and kernel 2.5.4-pre1 on sparc64

[Luigi Genoni]

On Thu, 7 Feb 2002, Oleg Drokin wrote:

> Hello!
>
> On Wed, Feb 06, 2002 at 06:53:47PM +0100, Luigi Genoni wrote:
>
> > I was trying 2.5.4-pre1 on sparc64 sun4u ultra2,
> > with 1GB memory and scsi controller ess.
> > This system has just one CPU.
> > >>TPC; 0046cf7c <grow_buffers+5c/e0>   <=====
> > >>O7;  0046cf38 <grow_buffers+18/e0>
> > >>I7;  0046b260 <__getblk+20/60>
> > Trace; 0046b260 <__getblk+20/60>
> > Trace; 0046b56c <__bread+c/a0>
> > Trace; 004d1374 <journal_init+214/8e0>
> This looks like VFS have exploded for whatever reson, I think
> because one ot these checks failed:
> linux/fs/buffer.c:
>         if (size & (get_hardsect_size(to_kdev_t(bdev->bd_dev))-1))
>                 BUG();
>         /* Size must be within 512 bytes and PAGE_SIZE */
>         if (size < 512 || size > PAGE_SIZE)
>                 BUG();
>
> Can any of sparc64 people comment on what kind of exception will one
> get for __builtin_trap?
> Second BUG() seems to be out of the question, though.
> Do you have CONFIG_DEBUG_BUGVERBOSE enabled?
No, I have not

Re: [reiserfs-dev] oops with reiserfs and kernel 2.5.4-pre1 on sparc64

[Oleg Drokin]

Hello!

On Thu, Feb 07, 2002 at 10:58:27AM +0100, Luigi Genoni wrote:

> > Can any of sparc64 people comment on what kind of exception will one
> > get for __builtin_trap?
> > Second BUG() seems to be out of the question, though.
> > Do you have CONFIG_DEBUG_BUGVERBOSE enabled?
> No, I have not
If you'd enable it, you'd get a message on a next crash, where this BUG()
happened exactly (so that we can verify my guess)

Bye,
    Oleg

Re: [reiserfs-dev] oops with reiserfs and kernel 2.5.4-pre1 on sparc64

[Luigi Genoni]

IT says that the kernel BUG is at buffer.c at line 2243; that is:

 /* Size must be within 512 bytes and PAGE_SIZE */
        if (size < 512 || size > PAGE_SIZE)
                BUG();

Luigi

On Thu, 7 Feb 2002, Oleg Drokin wrote:

> Hello!
>
> On Thu, Feb 07, 2002 at 10:58:27AM +0100, Luigi Genoni wrote:
>
> > > Can any of sparc64 people comment on what kind of exception will one
> > > get for __builtin_trap?
> > > Second BUG() seems to be out of the question, though.
> > > Do you have CONFIG_DEBUG_BUGVERBOSE enabled?
> > No, I have not
> If you'd enable it, you'd get a message on a next crash, where this BUG()
> happened exactly (so that we can verify my guess)
>
> Bye,
>     Oleg
>