RFR(S): 8195691: AIX build broken after 8194312

[Doerr, Martin]

Hi Kim and Götz,

thanks for reviewing.

Thanks for your support, Kim. Maybe you can push it together with the fix for JDK-8195690?

Best regards,
Martin


-----Original Message-----
From: Lindenmaier, Goetz 
Sent: Dienstag, 23. Januar 2018 11:56
To: Doerr, Martin <martin.doerr at sap.com>; Kim Barrett <kim.barrett at oracle.com>
Cc: hotspot-runtime-dev at openjdk.java.net
Subject: RE: RFR(S): 8195691: AIX build broken after 8194312

Hi Martin, 

the change looks good. Thanks. 

Best regards,
  Goetz.

> -----Original Message-----
> From: Doerr, Martin
> Sent: Montag, 22. Januar 2018 12:29
> To: Kim Barrett <kim.barrett at oracle.com>
> Cc: hotspot-runtime-dev at openjdk.java.net; Lindenmaier, Goetz
> <goetz.lindenmaier at sap.com>
> Subject: RE: RFR(S): 8195691: AIX build broken after 8194312
> 
> Hi Kim,
> 
> 1. The friend proposal still doesn't help and I'm fine with making the
> embedded classes public on AIX. I've added the "private" behind them.
> 2. The const fix works with changed type OopBlock*. I usually prefer const
> over macros, but in this case, a macro would have been fine for me, too,
> because EXPECT_EQ dives into the macro world, anyway. But if you prefer
> the const, here it is:
> http://cr.openjdk.java.net/~mdoerr/8195691_fix_AIX_build/webrev.02/
> 
> Best regards,
> Martin
> 
> 
> -----Original Message-----
> From: Kim Barrett [mailto:kim.barrett at oracle.com]
> Sent: Freitag, 19. Januar 2018 19:20
> To: Doerr, Martin <martin.doerr at sap.com>
> Cc: hotspot-runtime-dev at openjdk.java.net; Lindenmaier, Goetz
> <goetz.lindenmaier at sap.com>
> Subject: Re: RFR(S): 8195691: AIX build broken after 8194312
> 
> > On Jan 19, 2018, at 8:45 AM, Doerr, Martin <martin.doerr at sap.com>
> wrote:
> >
> > Hi Kim,
> >
> > thanks a lot for the detailed explanations.
> >
> > 1. Unfortunately, the friend proposal didn't help.
> 
> Drat.  But wait. The suggestion I sent you was probably insufficient.  Probably
> also need (AIX conditional) “friend class TestAccess;” after the declaration of
> TestAccess.
> 
> If that doesn’t work (or even if it does, but you decide to use the conditional
> non-private approach anyway), please add (unconditional) “private:” before
> the data member declarations in OopStorage, so that we’re not making the
> entire OopStorage class public on AIX, but only the parts that need to be.
> 
> > 2. The const proposal didn't work, either:
> > test/hotspot/gtest/gc/shared/test_oopStorage.cpp", line 1203.13: 1540-
> 0274 (S) The name lookup for "NULL_BLOCK" did not find a declaration.
> 
> I don’t understand this.  Was there some earlier error message on the
> declaration of NULL_BLOCK?
> 
> > You're right, that OopBlock* should be used. I have changed that and
> moved the macro below the typedef.
> 
> Okay, subject to understanding why the const declaration didn’t work.
> 
> > I have also added some comments which you have suggested.
> 
> Okay.
> 
> > I have uploaded a new webrev here:
> > http://cr.openjdk.java.net/~mdoerr/8195691_fix_AIX_build/webrev.01/
> >
> > Please review and push it when it's ok.
> 
> I’ll sponsor.