RFR(s): 8144219: [posix] Remove redundant code around os::print_siginfo() (original) (raw)
Thomas Stüfe thomas.stuefe at gmail.com
Mon Nov 30 10:03:15 UTC 2015
- Previous message: [aarch64-port-dev ] [RFR] openjdk aarch64: jdk/test/com/sun/net/httpserver/Test6a.java fails with --enable-unlimited-crypto
- Next message: 8144223: Move j.l.invoke.{ForceInline, DontInline, Stable} to jdk.internal.vm.annotation package
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi all,
please review this small change:
Webrev: http://cr.openjdk.java.net/~stuefe/webrevs/8144219-posix-os-print_siginfo/webrev.00/webrev/ Bug Report: https://bugs.openjdk.java.net/browse/JDK-8144219
This is mainly a cleanup change for posix platforms. It gets rid of multiple copies of os::print_siginfo() in favour of the central os::Posix::print_siginfo() in os_posix.cpp.
It also extends os::Posix::print_siginfo() to write more details about the received signal which stems from our experiences with various Unices over the time. This will improve hs-err file printout.
Note that there was on all platforms code which examines the crash address for a possible CDS fault and prints a specialized hint. In my opinion that makes no sense in a general os::print_siginfo() function and I moved it to a separate error reporting step into vmError.cpp. This may be a point of contention because the way to extract the fault address from a siginfo/EXCEPTION_RECORD structure is platform dependend, but imho this is still the least ugly solution.
Kind regards, Thomas
- Previous message: [aarch64-port-dev ] [RFR] openjdk aarch64: jdk/test/com/sun/net/httpserver/Test6a.java fails with --enable-unlimited-crypto
- Next message: 8144223: Move j.l.invoke.{ForceInline, DontInline, Stable} to jdk.internal.vm.annotation package
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]