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

List:       openjdk-2d-dev
Subject:    [OpenJDK 2D-Dev] 7044285: 64 bit VM crashes in Java_sun_java2d_loops_MaskFill_MaskFill
From:       suenaga.yasumasa () oss ! ntt ! co ! jp (Yasumasa Suenaga)
Date:       2011-07-08 1:58:05
Message-ID: 4E16642D.8080801 () oss ! ntt ! co ! jp
[Download RAW message or body]

Hi,

I posted the email attached to this email.
How about the situation of this bug?

According to the BugDatabase,
 - Synopsis has changed
 - Category has changed to "java:classes_2d"
 - State has changed to "3-Accepted"

So, I guess that this bug is fixing in Oracle Corp. now.


If cause of this bug is integer overflow, we have only the use of data type
which size is same as the pointer. So, I suggested the cast to ptrdiff_t.

Another solution, data type are changed into jlong from jint "rasInfo" and
"rasInfo.bounds" . However, to change the data type affects another code widely.


When I checked coredump image, related members' data type was "jint" (32 bits).

(gdb) ptype rasInfo
type = struct {
    SurfaceDataBounds bounds;
    void *rasBase;
    jint pixelBitOffset;
    jint pixelStride;  <-
    jint scanStride;   <-
    unsigned int lutSize;
    jint *lutBase;
    unsigned char *invColorTable;
    char *redErrTable;
    char *grnErrTable;
    char *bluErrTable;
    int *invGrayTable;
    union {
        void *align;
        char data[64];
    } priv;
}
(gdb) ptype rasInfo.bounds
type = struct {
    jint x1;  <-
    jint y1;  <-
    jint x2;
    jint y2;
}

In the LP64 architecture, the size of pointer is 64 bits.


Thanks,
Yasumasa
-------------- next part --------------
An embedded message was scrubbed...
From: Yasumasa Suenaga <suenaga.yasumasa at oss.ntt.co.jp>
Subject: 7044285: VM crashes in server app
Date: Mon, 06 Jun 2011 20:24:08 +0900
Size: 7486
Url: http://mail.openjdk.java.net/pipermail/2d-dev/attachments/20110708/b1650347/attachment.eml 

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