[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openjdk-hotspot-runtime-dev
Subject:    Re: RFR(S) 8055289: Internal Error: mallocTracker.cpp:146 fatal error: Should not use malloc for big
From:       Zhengyu Gu <zhengyu.gu () oracle ! com>
Date:       2014-09-04 18:47:49
Message-ID: 5408B3D5.7030205 () oracle ! com
[Download RAW message or body]

Thanks.

-Zhengyu

On 9/4/2014 2:43 PM, George Triantafillou wrote:
> Hi Zhengyu,
>
> This looks good.
>
> -George
>
> On 8/29/2014 2:23 PM, Zhengyu Gu wrote:
>> Native Memory Tracking uses bit field for malloc size, it uses 30 
>> bits (1GB-1) on 32-bit system and 62 bits ((1<<62)-1) bytes on 64-bit 
>> system, which is not an issue.
>>
>> Although, JVM does not have such large malloc, but it is possible to 
>> request >= 1GB malloc on 32-bit system via Unsafe API.
>>
>> The solution is to return NULL for such malloc request when NMT is 
>> enabled.
>>
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8055289
>> Webrev: http://cr.openjdk.java.net/~zgu/8055289/webrev.00/ 
>> <http://cr.openjdk.java.net/%7Ezgu/8055289/webrev.00/>
>>
>>
>> Thanks,
>>
>> -Zhengyu
>

[prev in list] [next in list] [prev in thread] [next in thread]