Oops. Please ignore my last mail. This one uses the correct mailing lists.
- Ioi
=============================================
On 6/15/16 3:06 PM, Ioi Lam wrote:
Please review a small fix:
Bug: https://bugs.openjdk.java.net/browse/JDK-8154820
Desc: JDWP: -agentlib:jdwp=help assertion error
Fix: http://cr.openjdk.java.net/~iklam/jdk9/8154820_jdwp_help_exit.v02/
The were two issues:
[1] MAXPATHLEN is really small on Windows (~256), so MAX_MESSAGE_LEN
was around 1024, but the jdwp help message is about 1600 bytes.
The fix would expand it to abut 2500 bytes which is more than enough.
[2] The meaning of the "outputMaxLen" parameter to the utf8ToPlatform()
and utf8FromPlatform() functions was ambiguous. The assert
UTF_ASSERT(outputMaxLen > len);
suggest that the trailing zero is included in the buffer size, so
at most outputMaxLenbytes are written.
However, the implementation actually would write outputMaxLen+1 bytes
in the worst case.
Luckily, this does not seem to be a problem as all calls to
utf8ToPlatform/utf8FromPlatform allocate way more space than
needed. The only exception was vprint_message, which chose to pass
outputMaxLen := sizeof(buf) -1
So in the worst case, the VM would exit due to the above
UTF_ASSERT, but
we would not silently write one byte past the end of the buffer.
The fix is:
[a] Clarify the meaning of the outputMaxLen parameter
[b] Make sure we don't write past the end of the buffer
[c] Fix vprint_message() to pass in the correct size of the
buffer to include space for the trailing zero.
Thanks
Ioi
