Review request for 5049299 (original) (raw)

Martin Buchholz martinrb at google.com
Sun May 24 01:20:40 UTC 2009


Two comments:


I would use _exit instead of standard exit in the process helper code, as in the existing code, although unlike the existing code, it doesn't seem to be a correctness issue.


I implemented

defaultPath

but should have tried to use confstr(_SC_PATH)

as in this example from the confstr manpage:

       char *pathbuf;
       size_t n;

       n = confstr(_CS_PATH,NULL,(size_t) 0);
       pathbuf = malloc(n);
       if (pathbuf == NULL)
           abort();
       confstr(_CS_PATH, pathbuf, n);

Of course, only within #ifdef _CS_PATH

That's a small portability bug that is worth fixing now that the jdk sources are targeting more platforms.

Martin

On Fri, May 22, 2009 at 03:05, Michael McMahon <Michael.McMahon at sun.com>wrote:

Hi,

I have just posted a webrev for 5049299: (process) Use posixspawn, not fork, on S10 to avoid swap exhaustion. webrev location: http://cr.openjdk.java.net/~michaelm/5049299/webrev.00/<http://cr.openjdk.java.net/%7Emichaelm/5049299/webrev.00/> -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.openjdk.java.net/pipermail/core-libs-dev/attachments/20090523/6d2cd33b/attachment.html>



More information about the core-libs-dev mailing list