RFR(trivial): JDK-8224568: minimal and zero build fails after JDK-8213084

Schmidt, Lutz lutz.schmidt at sap.com
Wed May 22 16:16:40 UTC 2019


Hi,

maybe rephrase the line to 
    // might prove cumbersome on platforms where instr_len is hard to find out

Thanks,
Lutz

On 22.05.19, 17:50, "Ao Qi" <aoqi at loongson.cn> wrote:

    On Wed, May 22, 2019 at 11:37 PM Schmidt, Lutz <lutz.schmidt at sap.com> wrote:
    >
    > Hi all,
    >
    > sorry for reacting only with delay.
    >
    > I agree with Aleksey's idea to unconditionally set _show_bytes=false.
    
    Should the comment for x86/arm/aarch64 at line 49 be kept ?
    
      47 #if defined(X86) || defined(ARM) || defined(AARCH64)
      48 bool AbstractDisassembler::_show_bytes  = false; // set "true" to
    see what's in memory bit by bit
      49                                                  // might prove
    cumbersome because instr_len is hard to find on x86 or arm
      50 #else
      51 bool AbstractDisassembler::_show_bytes  = false; // set "true" to
    see what's in memory bit by bit
      52 #endif
    
    Cheers,
    Ao Qi
    
    >
    > The platform-specific initialization has become far less important - if not obsolete. It is now possible to toggle _show_bytes via -XX:PrintAssemblyOptions=show_bytes.
    >
    > Thanks for fixing this!
    >
    > Regards,
    > Lutz
    >
    >
    > On 22.05.19, 16:19, "Aleksey Shipilev" <shade at redhat.com> wrote:
    >
    >     On 5/22/19 10:41 AM, Ao Qi wrote:
    >     > JBS:
    >     > https://bugs.openjdk.java.net/browse/JDK-8224568
    >     >
    >     > Webrev:
    >     > http://cr.openjdk.java.net/~aoqi/8224568/webrev.00/
    >
    >     Header addition looks trivial and good.
    >
    >     Rewiring platform-specific ifdefs raises a lot of questions (mostly to the original patch): since we
    >     are always defining to the same value, why do we even have platform-specific defines there? Also,
    >     that means that new platform would break without new platform-specific block there. I think we are
    >     better off just dropping the platform-specific defines and just unconditionally set _show_bytes=false.
    >
    >     Lutz, please advise here.
    >
    >     -Aleksey
    >
    >
    >
    



More information about the hotspot-compiler-dev mailing list