RFR: 8368781: PerfMemory - make issues more transparent [v2]

[Matthias Baesken]

On Tue, 14 Oct 2025 03:41:05 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> Matthias Baesken has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Make EnhanceErrorWarningLogging DIAGNOSTIC
>
> We already have some UL in PerfMemory:
>> 8168122: Update logging in perfMemory to Unified Logging
> Summary: -XX:+PerfTraceMemOps replaced with -Xlog:perf+memops=debug, -XX:+PerfTraceDataCreation replaced with -Xlog:perf+datacreation=debug
> 
> but for some reason that effort completely ignored the PrintMiscellaneous/Verbose "warnings". To keep things simple I suggest the following pattern:
> 
> diff --git a/src/hotspot/os/posix/perfMemory_posix.cpp b/src/hotspot/os/posix/perfMemory_posix.cpp
> index ed83487265c..968f3cea249 100644
> --- a/src/hotspot/os/posix/perfMemory_posix.cpp
> +++ b/src/hotspot/os/posix/perfMemory_posix.cpp
> @@ -447,10 +447,11 @@ static char* get_user_name(uid_t uid) {
>    int result = getpwuid_r(uid, &pwent, pwbuf, (size_t)bufsize, &p);
>  
>    if (result != 0 || p == nullptr || p->pw_name == nullptr || *(p->pw_name) == '\0') {
> -    if (PrintMiscellaneous && Verbose) {
> +    if (log_is_enabled(Debug, perf)) {
> +      LogStreamHandle(Debug, perf) log;
>        if (result != 0) {
> -        warning("Could not retrieve passwd entry: %s\n",
> -                os::strerror(result));
> +        log.print_cr("Could not retrieve passwd entry: %s",
> +                     os::strerror(result));
>        }
>        else if (p == nullptr) {
>          // this check is added to protect against an observed problem
> @@ -463,13 +464,13 @@ static char* get_user_name(uid_t uid) {
>          // message may result in an erroneous message.
>          // Bug Id 89052 was opened with RedHat.
>          //
> -        warning("Could not retrieve passwd entry: %s\n",
> -                os::strerror(errno));
> +        log.print_cr("Could not retrieve passwd entry: %s",
> +                      os::strerror(errno));
>        }
>        else {
> -        warning("Could not determine user name: %s\n",
> -                 p->pw_name == nullptr ? "pw_name = null" :
> -                                      "pw_name zero length");
> +        log.print_cr("Could not determine user name: %s",
> +                      p->pw_name == nullptr ? "pw_name = null" :
> +                                              "pw_name zero length");
>        }
>      }
>      FREE_C_HEAP_ARRAY(char, pwbuf);
> 
> Of course you proposed changes as a potential consumer of this information, so how would you like to procure it?

@dholmes-ora while looking into the Windows perfmem code, I noticed we are using  SetSecurityDescriptorControl 
https://github.com/openjdk/jdk/blob/1653999871c8d7b1e61b44f8525e09b2cd0bdb6b/src/hotspot/os/windows/perfMemory_windows.cpp#L1016

but we use still handling for ancient Windows versions (we use dynamic handling of the function at runtime), should we remove this ?
Looking at the description
https://learn.microsoft.com/en-us/windows/win32/api/securitybaseapi/nf-securitybaseapi-setsecuritydescriptorcontrol
it is available at least since Win2003/XP so the function can be called directly  (would do this in a separate JBS issue).
What do you think ?

-------------

PR Comment: https://git.openjdk.org/jdk/pull/27602#issuecomment-3411049767