RFR(T) 8218566: NMT: missing memory tag for assert poison page

[zgu at redhat.com]

Hi Thomas,

On Wed, 2019-02-06 at 17:29 +0100, Thomas Stüfe wrote:
> Hmm...
> 
> the patch is fine, but we should fix os::reserve_memory:
> 
Thanks for reviewing.

> we have
> 
>   static char*  reserve_memory(size_t bytes, char* addr = 0,
>                                size_t alignment_hint = 0, int
> file_desc = -1);
>   static char*  reserve_memory(size_t bytes, char* addr,
>                                size_t alignment_hint, MEMFLAGS
> flags);
> 
> one version registers the memory with NMT, one does not. It would be
> better if both did, like os::malloc does in all its variants, instead
> of having to fix up the callsites...
Yes, they are inconsistent. IIRC, that was due to up/down calls between
os and pd methods. There is also related RFE (JDK-8198226) in this
area.

Thanks,

-Zhengyu


> 
> Cheers, Thomas
> 
> 
> 
> 
> On Wed, Feb 6, 2019 at 5:15 PM <zgu at redhat.com> wrote:
> > Please review the trivial change to tag assert poison page.
> > 
> > Bug: https://bugs.openjdk.java.net/browse/JDK-8218566
> > Webrev: http://cr.openjdk.java.net/~zgu/JDK-8218566/webrev.00/
> > 
> > Test:
> >   hotspot_nmt on Linux x64 (fastdebug and release),
> >   passed all except
> > runtime/NMT/CheckForProperDetailStackTrace.java  
> >   (JDK-8218458) with release build.
> > 
> > Thanks,
> > 
> > -Zhengyu