Hi Dave/Adam,

Could you give me some references/examples for this type of debugging. Much
appreciation for you great help.

Thanks,
John

2007/9/18, David Powell <[EMAIL PROTECTED]>:
>
> Adam Leventhal wrote:
> > On Mon, Sep 17, 2007 at 06:45:58PM +0800, xiong xiaoyi wrote:
> >> Hi  experts,
> >>
> >> I have a simple problem on mdb debugging, I use mdb to debug a core,
> >>
> >>> $C
> >> ffbfde18 libc.so.1`realfree+0x40(90c58, ef012eef, ff000000, a90fc, 0,
> >> ff362a00)
> >> ffbfde78 libc.so.1`cleanfree+0x58 (0, 4, ff01b6b8, ff012620, a885c, 1)
> >> ffbfded8 libc.so.1`_malloc_unlocked+0x104(8, ef012eec, 90c50, ef012eef,
> >> 1c00,
> >> ff19e5c0)
> >> ffbfdf38 libc.so.1`malloc+0x48(8, 1cc4, 1, ff012620, a9824, ff01ca2c)
> >> ffbfdf98 libc.so.1`calloc+0x58 (8, 8, 8, 10, 30, 30)
> >> ffbfdff8 libnsl.so.1`thr_get_storage+0x40(ff22e440, 8, 5ce70, 59c2c,
> 0, 0)
> >>
> >> In above stack, it prints function's argument, but the number of the
> >> arguments is not same as the function definition.
> >> For example, in mdb, we get:
> >>
> >>        ffbfdf38 libc.so.1`malloc+0x48(8, 1cc4, 1, ff012620, a9824,
> ff01ca2c)
> >>
> >> But the molloc definition is :
> >>
> >>        malloc(size_t size)
> >>
> >> So from the mdb output above, how can I know what's the size in malloc?
> >> Thanks in advance.
> >
> > This question is probably more appropriate for mdb-discuss, but to
> answer:
> >
> > On SPARC we always give you the values of the 8 input registers in order
> > from 0 to 7. You should just pay attention to the first value in this
> case.
>
>    Keep in mind that the registers listed in the stack trace are the %i
>    registers stored at the point the next function was called (in this
>    case, at malloc+0x48).  You will need to verify (with ::dis) that the
>    register you're interested in hasn't been modified since the top of
>    the function.  If it has, you'll either need to find where the
>    argument was stashed, or reconstruct the argument using the caller's
>    stack frame and the code around calloc+0x58.
>
>    Dave
>
>
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to