RFR(xs): 8145114: const-correctness for ucontext_t* reading functions

[David Holmes]

Hi Thomas,

In:

ExtendedPC os::fetch_frame_from_context(const void* ucVoid,
                     intptr_t** ret_sp, intptr_t** ret_fp) {
   ExtendedPC  epc;
   ucontext_t* uc = (ucontext_t*)ucVoid;


is there some reason these ucontext_t's were not made const?

Thanks,
David

On 11/12/2015 7:57 PM, Thomas Stüfe wrote:
> Hi David,
>
> On Fri, Dec 11, 2015 at 10:35 AM, David Holmes <david.holmes at oracle.com
> <mailto:david.holmes at oracle.com>> wrote:
>
>     Hi Thomas,
>
>     This seems okay. To be honest I'm not sure of the usefulness but it
>     certainly isn't harmful :)
>
>     I'll prepare similar changes to our internal platforms and run it
>     all through JPRT.
>
>     Thanks,
>     David
>
>
>     On 11/12/2015 5:16 PM, Thomas Stüfe wrote:
>
>         Hi all,
>
>         please review this large-but-trivial change:
>
>         Bug: https://bugs.openjdk.java.net/browse/JDK-8145114
>         webrev:
>         http://cr.openjdk.java.net/~stuefe/webrevs/8145114-const-correctness-ucontext/webrev.00/webrev/index.html
>
>         Even though many files are involved, the changes are very
>         trivial. The
>         change takes care of declaring ucontext_t pointers as const for
>         all utility
>         functions which do not change the context (which are most of them).
>
>         This fix was motivated by me wasting time in an error situation
>         on AIX
>         where I accidentally changed the register values in the context
>         before
>         printing them to the hs-err file.
>
>         And, like pulling a thread from a ball of yarn, the change got
>         bigger and
>         bigger... like it is once one starts to fix const issues. But
>         the fixes in
>         this case involve the bottom functions and should not force any
>         new casts
>         upon anyone.
>
>         I did build the change on the following platforms:
>         AIX
>         Linux x64 (with and without ZERO)
>         Linux ppc
>         OSX
>         Solaris Sparc and x64
>
>         and all did build successfully.
>
>         Kind Regards, Thomas
>
>