RFR (S) Make tenuring threshold unsigned
John Cuthbertson
john.cuthbertson at oracle.com
Tue Sep 25 20:53:52 UTC 2012
Hi Jesper,
Looks OK except for the nits below:
cmsAdaptiveSizePolicy.cpp:
Line 1318: Change the format descriptor.
g1CollectorPolicy.hpp:
Looks like you missed this file:
Line 844:
// Current tenuring threshold, set to 0 if the collector reaches the
// maximum amount of suvivors regions.
int _tenuring_threshold;
Line 865:
age should be unsigned too.
psScavenge.cpp:
Line 528: Change the format descriptor.
psScavenge.hpp:
Line 91: Missed
static int tenuring_threshold() { return _tenuring_threshold; }
adaptiveSizePolicy.cpp
Looks like you missed this file:
Line 643:
bool AdaptiveSizePolicy::print_adaptive_size_policy_on(
outputStream* st,
int tenuring_threshold_arg)
const {
Line 666: Change format descriptor
adaptiveSizePolic.hpp:
Line 493:
Another missed:
bool print_adaptive_size_policy_on(outputStream* st, int
tenuring_threshold) const;
ageTable.cpp:
Line 99: Change format descriptors.
defNewGeneration.hpp:
Another file missed:
Line 46:
int _tenuring_threshold; // Tenuring threshold for next
collection.
Line 328:
int tenuring_threshold() { return _tenuring_threshold; }
vmStructs.cpp:
Another miss:
Line 511:
nonstatic_field(DefNewGeneration,
_tenuring_threshold,
int) \
arguments.cpp:
Line 1158:
tenuring_default should be uintx.
oop.hpp:
Line 330: Shouldn't age() return an unsigned?
JohnC
On 09/24/12 06:57, Jesper Wilhelmsson wrote:
> Hi,
>
> While working on a different change I noticed that the tenuring
> threshold was stored in a signed int. This struck me as odd, and since
> I wanted to avoid some unintuitive casts in my other change I would
> prefer to change the threshold to an unsigned.
>
> The change is available here:
>
> http://cr.openjdk.java.net/~jwilhelm/TenuringThreshold/
>
> Several files are touched but there's just type changes in a few
> places per file.
>
> In this change I also change the declared type of
> -XX:MaxTenuringThreshold and -XX:InitialTenuringThreshold to uintx.
> This should not be a problem since we don't allow negative values for
> these flags anyway.
> /Jesper
More information about the hotspot-gc-dev
mailing list