[8u] RFR: fix inconsistency of backport of 8219677: "-XX:OnOutOfMemoryError" uses fork instead of vfork

[Hohensee, Paul]

Thanks for the review, Volker.

Paul

From: Volker Simonis <volker.simonis at gmail.com>
Date: Sunday, November 17, 2019 at 4:43 AM
To: "Hohensee, Paul" <hohensee at amazon.com>
Cc: "Yangfei (Felix)" <felix.yang at huawei.com>, jdk8u-dev <jdk8u-dev at openjdk.java.net>
Subject: Re: [8u] RFR: fix inconsistency of backport of 8219677: "-XX:OnOutOfMemoryError" uses fork instead of vfork

Hohensee, Paul <hohensee at amazon.com<mailto:hohensee at amazon.com>> schrieb am Fr., 15. Nov. 2019, 21:09:
May I have another review please?

Webrev: http://cr.openjdk.java.net/~phh/8234264/webrev.8u.00/

The fix is trivial and the change looks good. Thumbs up from me.

As a side note, I don't understand why the call to fork_and_exec() in void VMError::report_and_die (which was wrongly patched by the downport) hasn't been lifted by the initial change 8047434 to use vfork() as well? But that's a question not relevant for this fix of the downport and should probably be addressed in the head revision.

Best regards,
Volker

Thanks,

Paul

On 11/15/19, 8:49 AM, "jdk8u-dev on behalf of Hohensee, Paul" <jdk8u-dev-bounces at openjdk.java.net<mailto:jdk8u-dev-bounces at openjdk.java.net> on behalf of hohensee at amazon.com<mailto:hohensee at amazon.com>> wrote:

    https://bugs.openjdk.java.net/browse/JDK-8234264

    On 11/15/19, 8:33 AM, "jdk8u-dev on behalf of Hohensee, Paul" <jdk8u-dev-bounces at openjdk.java.net<mailto:jdk8u-dev-bounces at openjdk.java.net> on behalf of hohensee at amazon.com<mailto:hohensee at amazon.com>> wrote:

        You are correct. I'll file an issue to fix it.

        Thanks,

        Paul

        On 11/14/19, 11:05 PM, "jdk8u-dev on behalf of Yangfei (Felix)" <jdk8u-dev-bounces at openjdk.java.net<mailto:jdk8u-dev-bounces at openjdk.java.net> on behalf of felix.yang at huawei.com<mailto:felix.yang at huawei.com>> wrote:

            Hi,

              We found one inconsistency of backport of 8219677.
              The upstream patch set parameter use_vfork_if_available to true when call os::fork_and_exec in VM_ReportJavaOutOfMemory::doit().
              But looks like the 8u backport did something inconsistent: the change was not made in the correct place.
              The 8u backport patch: http://hg.openjdk.java.net/jdk8u/jdk8u/hotspot/rev/97d605522fcb

            Patch for jdk8u-dev:

            diff -r 775e2bf92114 src/share/vm/utilities/vmError.cpp
            --- a/src/share/vm/utilities/vmError.cpp        Wed Aug 07 17:00:19 2019 +0800
            +++ b/src/share/vm/utilities/vmError.cpp        Fri Nov 15 14:23:50 2019 +0800
            @@ -1060,7 +1060,7 @@
                   out.print_raw   (cmd);
                   out.print_raw_cr("\" ...");

            -      if (os::fork_and_exec(cmd, true) < 0) {
            +      if (os::fork_and_exec(cmd) < 0) {
                     out.print_cr("os::fork_and_exec failed: %s (%d)", strerror(errno), errno);
                   }
                 }
            @@ -1147,7 +1147,7 @@
            #endif
                 tty->print_cr("\"%s\"...", cmd);

            -    if (os::fork_and_exec(cmd) < 0) {
            +    if (os::fork_and_exec(cmd, true) < 0) {
                   tty->print_cr("os::fork_and_exec failed: %s (%d)", strerror(errno), errno);
                 }
               }

            Please confirm.

            Thanks,
            Felix