RFR 8014431 : cleanup warnings indicated by the -Wunused-value compiler option on linux

David Holmes david.holmes at oracle.com
Thu May 23 05:45:13 PDT 2013


Hi Calvin,

Mostly looks okay to me. A few comments

src/share/vm/memory/universe.cpp

Seems like a real bug!

--

src/share/vm/opto/memnode.cpp:

I think it is cleaner to add the (void) to the macro definition as in 
other cases:
#define verify_memory_slice(m,i,n) (void)(0)  // PRODUCT version is no-op

---

src/share/vm/runtime/perfData.hpp

This can't be right:

  773       return create_string_variable(ns, name, (int)0, s, CHECK_NULL);

why do we have to cast 0 to int ????

---

src/share/vm/services/diagnosticArgument.cpp

Another real bug!

---

src/share/vm/utilities/taskqueue.hpp

I do not understand this code:

!     // g++ complains if the volatile result of the assignment is unused.
!     const_cast<E&>(_elems[localBot] = t);

why do we even need the const cast? How is the assignment not used ???

I understand your fix of adding (void) but I'm baffled by the original code.

---

Thanks,
David
-----

On 22/05/2013 5:32 AM, Calvin Cheung wrote:
> This change is for fixing the errors resulting from enabling the
> -Wunused-value compiler option on linux.
>
> The patch was initiated by Jeremy (jeremymanson at google.com).
> More changes are needed for it to pass the build.
>
> Tests performed:
>     jprt
>     jtreg on linux_x64
>     vm.quick on linux_x64
>
> Webrev: http://cr.openjdk.java.net/~ccheung/8014431/webrev/
> Bug: http://bugs.sun.com/view_bug.do?bug_id=8014431
>
> thanks,
> Calvin


More information about the hotspot-runtime-dev mailing list