8212110: Build of saproc.dll broken on Windows 32 bit after JDK-8210647

[Doerr, Martin]

Hi Severin,

thank you for providing this fix so quickly. It looks good to me.

I was able to build jdk11u with this patch.

Just for information:
I'm using Visual Studio 2013 and configure with --disable-warnings-as-errors.
(Otherwise I got
Creating support/modules_cmds/jdk.hotspot.agent/jhsdb.exe from 1 file(s)
jdk11u/src/jdk.crypto.mscapi/windows/native/libsunmscapi/security.cpp(773) : error C2220: warning treated as error - no 'object' file generated
jdk11u/src/jdk.crypto.mscapi/windows/native/libsunmscapi/security.cpp(773) : warning C4244: 'argument' : conversion from 'jlong' to 'HCRYPTKEY', possible loss of data
jdk11u/src/jdk.crypto.mscapi/windows/native/libsunmscapi/security.cpp(773) : warning C4244: 'argument' : conversion from 'jlong' to 'HCRYPTPROV', possible loss of data
jdk11u/src/jdk.crypto.mscapi/windows/native/libsunmscapi/security.cpp(967) : warning C4244: 'argument' : conversion from 'jlong' to 'HCRYPTKEY', possible loss of data
jdk11u/src/jdk.crypto.mscapi/windows/native/libsunmscapi/security.cpp(967) : warning C4244: 'argument' : conversion from 'jlong' to 'HCRYPTPROV', possible loss of data
)

But that's unrelated to your fix. It works fine.

Best regards,
Martin


-----Original Message-----
From: Severin Gehwolf <sgehwolf at redhat.com> 
Sent: Freitag, 12. Oktober 2018 11:20
To: build-dev <build-dev at openjdk.java.net>
Cc: Doerr, Martin <martin.doerr at sap.com>; Baesken, Matthias <matthias.baesken at sap.com>
Subject: RFR: 8212110: Build of saproc.dll broken on Windows 32 bit after JDK-8210647

Hi,

Please review this fix for a build failure of saproc.dll on Windows 32
bit. For 32 bit Windows builds flag -RTC1 gets added unconditionally.
Yet, as per compiler doc[1] it should only be added for development
builds. This causes a build failure post JDK-8210647 as in
release/fastdebug variant builds of the JDK saproc.dll will get
optimization flags producing an incompatible set of flags. This used to
work prior JDK-8210647, because saproc.dll was never optimized (not
even in release/fastdebug variants).

The proposed fix is to only add the flag for slowdebug builds, which
corresponds to development builds in OpenJDK. Thoughts?

Bug: https://bugs.openjdk.java.net/browse/JDK-8212110
webrev: http://cr.openjdk.java.net/~sgehwolf/webrevs/JDK-8212110/webrev.01/

Unfortunately, I don't have a system to actually test this on. So will
rely on Martin or Matthias to test whether this fixes the build failure
[2] they are seeing. AFAIK, Oracle no longer builds on Windows 32 bit
either.

Thanks,
Severin

[1] https://msdn.microsoft.com/en-us/library/8wtf2dfz.aspx
[2] http://mail.openjdk.java.net/pipermail/jdk-updates-dev/2018-October/000213.html