perlembed and PERL_SYS_INIT3
Hi there all, Recently we got a bug report on freeradius that is not yet confirmed, but seems likely to be right. The short version is that rlm_perl does not use the recommended perl macros to set up the environment correctly, resulting in perl_parse() hanging in a futex lock on at least hppa. The slightly longer version is viewable at http://bugs.debian.org/495073 and the bug report linked from there. There are tentative patches in our git tree viewable at http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=e651a5b6... and http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=4eca65e0... (the second one cleans up the compiler complaints the first introduced). Thanks, -- -------------------------------------------------------------------------- | Stephen Gran | Blue paint today. [Funny to Jack | | steve@lobefin.net | Slingwine, Guy Harris and Hal Pierson. | | http://www.lobefin.net/~steve | Ed.] | --------------------------------------------------------------------------
Stephen Gran wrote:
There are tentative patches in our git tree viewable at http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=e651a5b6... and http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=4eca65e0... (the second one cleans up the compiler complaints the first introduced).
Which calls malloc() for 4 pointers... and never calls free. :) Alan DeKok.
On Thu, Aug 14, 2008 at 05:11:34PM +0200, Alan DeKok said:
Stephen Gran wrote:
There are tentative patches in our git tree viewable at http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=e651a5b6... and http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=4eca65e0... (the second one cleans up the compiler complaints the first introduced).
Which calls malloc() for 4 pointers... and never calls free. :)
Yeah, you are right about that. Since we're in effect emulating argv, I wasn't so concerned about it, but I'll see what I can do. -- -------------------------------------------------------------------------- | Stephen Gran | "A power so great, it can only be used | | steve@lobefin.net | for Good or Evil!" -- Firesign | | http://www.lobefin.net/~steve | Theatre, "The Giant Rat of Summatra" | --------------------------------------------------------------------------
On Thu, Aug 14, 2008 at 05:43:20PM +0100, Stephen Gran said:
On Thu, Aug 14, 2008 at 05:11:34PM +0200, Alan DeKok said:
Stephen Gran wrote:
http://git.debian.org/?p=collab-maint/freeradius.git;a=commitdiff;h=4eca65e0...
Which calls malloc() for 4 pointers... and never calls free. :)
Yeah, you are right about that. Since we're in effect emulating argv, I wasn't so concerned about it, but I'll see what I can do.
It looks like I can free(embed) just above return 0 at the end of perl_instantiate(), but I wanted to make sure I'm not overlooking anything obvious before committing that. The reason I'm hesitant is because of the argv emulation, and I have no idea if something else deep in the perl startup routines will still have a reference to it. Thanks, -- -------------------------------------------------------------------------- | Stephen Gran | You knew the job was dangerous when you | | steve@lobefin.net | took it, Fred. -- Superchicken | | http://www.lobefin.net/~steve | | --------------------------------------------------------------------------
participants (2)
-
Alan DeKok -
Stephen Gran