Code review request: 7186778 MachO decoder implementation for MacOSX

[Kevin Walls]

Hi Zhengyu

Thanks for taking this on. 8-)

os_bsd.cpp at line 1976 is my comment which needs to go:

1976   // Is this where we end up for a non-dynamic symbol?
1977   // If so, here's the brute force approach:
1978   // Hoping that dladdr() sets dlinfo.dli_fbase above, or do we 
need to look it up?
1979   // Perhaps verify it's not null.
1980   if (Decoder::decode(addr, localbuf, MACH_MAXSYMLEN, offset, 
dlinfo.dli_fbase)) {

...or at least change to:

// Handle non-dymanic manually:

Also there's a comment at 1957 which should go.  It seems that the 
dladdr() call at 1961 sets our dlinfo.dli_fbase, even when it can't 
return a symbol name.  (If any doubt, we should check that the dli_fbase 
address is at least non-null before calling Decoder::decode at 1980?...  
Actually, yes we should check as we are relying on dladdr() populating 
that field, but what if it failed for some reason other than it being a 
dynamic symbol?)

Thanks!
Kevin



On 25/07/12 18:47, Zhengyu Gu wrote:
> This is the implementation of MachO decoder for MacOSX. The 
> implementation is based on the patch provided by Kevin Walls, I 
> modified it to fit it into current decoder framework.
>
>
> CR: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7186778
> Webrev: http://cr.openjdk.java.net/~zgu/7186778/webrev.00/
>
> Tests:
>   The repo has ran through JPRT. I also used NMT queries to display 
> the decoded callsites.
>
>
> Thanks,
>
> -Zhengyu