[8u-dev] Request for approval for CR 8069124: runtime/NMT/MallocSiteHashOverflow.java failing in nightlies
Rob McKenna
rob.mckenna at oracle.com
Mon Jun 25 14:52:49 UTC 2018
Thanks for the clarification Kevin.
Approved.
-Rob
On 25/06/18 15:07, Kevin Walls wrote:
> Hi,
>
> I'd like to request approval to backport from 9 to 8u:
>
> 8069124: runtime/NMT/MallocSiteHashOverflow.java failing in nightlies
> JBS: https://bugs.openjdk.java.net/browse/JDK-8069124
>
> 9 changeset:
> URL: http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/e0c6eb5fce97
> http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/e0c6eb5fce97
>
> 9 review thread:
> http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2015-March/014113.html
>
>
> This is a "clean backport, or imports with minimal/trivial changes".
> Specifically, this is a clean import except for
> src/share/vm/services/mallocSiteTable.cpp where the change of "int index" to
> "unsigned int index" failed to happen automatically: on a neighbouring line,
> the jdk9 change also removes an assert, which we don't have here in 8u.
> The rejected 9 change and proposed 8u change are pasted in below (will
> webrev if not clear...?).
>
> The bug title is a test failure, but this is wanted in 8u as it cleans up
> the types and avoids a compile error on later Windows
> compilers, i.e.:
> ...hotspot\src\share\vm\utilities\nativeCallStack.cpp(78): warning C4311:
> 'type cast': pointer truncation from 'const address' to 'long'
> ...hotspot\src\share\vm\utilities\nativeCallStack.cpp(78): warning C4302:
> 'type cast': truncation from 'const address' to 'long'
>
>
> Many thanks!
> Kevin
>
>
> 9 change rejected for 8u:
>
> bash-4.2$ cat src/share/vm/services/mallocSiteTable.cpp.rej
> --- mallocSiteTable.cpp
> +++ mallocSiteTable.cpp
> @@ -135,8 +135,7 @@
> */
> MallocSite* MallocSiteTable::lookup_or_add(const NativeCallStack& key,
> size_t* bucket_idx,
> size_t* pos_idx) {
> - int index = hash_to_index(key.hash());
> - assert(index >= 0, err_msg("Negative index %d", index));
> + unsigned int index = hash_to_index(key.hash());
> *bucket_idx = (size_t)index;
> *pos_idx = 0;
>
>
> 8u change:
>
> bash-4.2$ hg diff src/share/vm/services/mallocSiteTable.cpp
> diff -r 0fa4c2b668b9 src/share/vm/services/mallocSiteTable.cpp
> --- a/src/share/vm/services/mallocSiteTable.cpp Fri Jun 22 01:55:23 2018
> -0700
> +++ b/src/share/vm/services/mallocSiteTable.cpp Mon Jun 25 06:54:46 2018
> -0700
> @@ -136,7 +136,7 @@
> MallocSite* MallocSiteTable::lookup_or_add(const NativeCallStack& key,
> size_t* bucket_idx,
> size_t* pos_idx, MEMFLAGS flags) {
> assert(flags != mtNone, "Should have a real memory type");
> - int index = hash_to_index(key.hash());
> + unsigned int index = hash_to_index(key.hash());
> assert(index >= 0, "Negative index");
> *bucket_idx = (size_t)index;
> *pos_idx = 0;
> bash-4.2$
>
More information about the jdk8u-dev
mailing list