Thank you for the reply. I've read your patch on a webrev. Your patch will work on AMD64 architecture. However, this patch will not work on x86 (32bit) architecture. Preprocessor in GCC for i386 does not define "_LP64" .
----------------- [r...@rhel4-4 test]# cat /etc/redhat-release Red Hat Enterprise Linux ES release 4 (Nahant Update 4) [r...@rhel4-4 test]# uname -a Linux RHEL4-4 2.6.9-42.EL #1 Wed Jul 12 23:16:43 EDT 2006 i686 i686 i386 GNU/Linux [r...@rhel4-4 test]# rpm -q glibc glibc-2.3.4-2.25 [r...@rhel4-4 test]# rpm -q gcc gcc-3.4.6-3 [r...@rhel4-4 test]# touch test.h [r...@rhel4-4 test]# cpp -dM test.h | grep _LP64 [r...@rhel4-4 test]# ----------------- So, you need to modify the patch as follows: ----------------- [r...@fedora13 OpenJDK7]# diff -u b118/openjdk/hotspot/agent/src/os/linux/ps_proc.c trunk/ openjdk/hotspot/agent/src/os/linux/ps_proc.c --- a/agent/src/os/linux/ps_proc.c 2010-11-12 05:43:12.000000000 +0900 +++ b/agent/src/os/linux/ps_proc.c 2010-12-02 10:45:09.117050388 +0900 @@ -124,6 +124,8 @@ #ifdef _LP64 #ifdef PTRACE_GETREGS64 #define PTRACE_GETREGS_REQ PTRACE_GETREGS64 +#elif defined(PTRACE_GETREGS) +#define PTRACE_GETREGS_REQ PTRACE_GETREGS #endif #else #if defined(PTRACE_GETREGS) || defined(PT_GETREGS) ----------------- Thanks. (2010/12/02 1:41), Kevin Walls wrote: > > Sorry, there was actually a typo in that diff, although it was good > enough to be a solution on my system. > > A webrev also: > > http://cr.openjdk.java.net/~kevinw/7003789/webrev.00/ > > Thanks > Kevin > > > On 01/12/2010 16:01, Kevin Walls wrote: >> Hi -- >> >> I've just been hitting that as well.... I think it needs to be dealt >> with as a separate bug as there may be a few which are similar - and >> 6359295 is marked fixed in 5.0 long ago... >> >> I just found your message after I'd done some investigating... The patch >> I was working with is to split one of the existing "if defined" >> statements, as currently the bug is: if we have PT_GETREGS defined, we >> use PTRACE_GETREGS to define PTRACE_GETREGS_REQ... We soon realise that >> may NOT be defined and give the "unsupported" message. So I can log and >> bug and get this done if it sounds good: >> >> [ke...@oldbox make]$ hg diff ../src/os/linux/ps_proc.c >> diff --git a/agent/src/os/linux/ps_proc.c b/agent/src/os/linux/ps_proc.c >> --- a/agent/src/os/linux/ps_proc.c >> +++ b/agent/src/os/linux/ps_proc.c >> @@ -124,9 +124,9 @@ >> #ifdef _LP64 >> #ifdef PTRACE_GETREGS64 >> #define PTRACE_GETREGS_REQ PTRACE_GETREGS64 >> -#endif >> -#else >> -#if defined(PTRACE_GETREGS) || defined(PT_GETREGS) >> +#elif defined(PTRACE_GETREGS) >> +#define PTRACE_GETREGS_REQ PTRACE_GETREGS >> +#elif defined (PT_GETREGS) >> #define PTRACE_GETREGS_REQ PTRACE_GETREGS >> #endif >> #endif /* _LP64 */ >> >> >> >> Regards >> Kevin >> >> >> >> On 15/11/2010 06:46, Yasumasa Suenaga wrote: >>> Hi. >>> >>> I and co-worker use jstack for various trouble shooting. >>> >>> We mainly use Java on Linux with AMD64 architecture. >>> However, jstack -F option doesn't work our platform. >>> >>> I ran jstack -F with LIBSAPROC_DEBUG=1 (environment variable), >>> I got following messages: >>> >>> >>> /***************/ >>> Thread 31933: (state = BLOCKED) >>> libsaproc DEBUG: ptrace(PTRACE_GETREGS, ...) not supported >>> Error occurred during stack walking: >>> sun.jvm.hotspot.debugger.DebuggerException: >>> sun.jvm.hotspot.debugger.DebuggerException: get_thread_regs failed for a lwp >>> /***************/ >>> >>> >>> In order to fix this problem, I made a patch for preprocessor >>> macro in "agent/src/os/linux/ps_proc.c" . >>> The patch that attached this mail works well on Fedora 13 x86_64. >>> >>> Please merge this patch if you don't fix this problem yet. >>> >>> >>> Best regards. >>> >>> > -- 日本電信電話株式会社 研究企画部門 OSS センタ 応用技術ユニット Webグループ 末永 恭正（すえなが やすまさ） TEL: 03-5860-5105 (直通 5069) E-mail: suenaga.yasum...@oss.ntt.co.jp
