Front page | perl.perl5.porters |
Postings from October 2003
On Thu, Oct 23, 2003 at 09:47:21PM +0100, Dave Mitchell wrote:
> On Wed, Oct 22, 2003 at 10:15:52PM +0200, Elizabeth Mattijsen wrote:
> > Well, maybe someone can figure out what is going on here and maybe
> > we'll get a fix for 5.8.2?
I meant to reply to this "not likely unless someone does it real soon"
> Whether this will be integrated in 5.8.2 is up to Nick.
I confess I don't understand the sharing system well enough to be a good
judge of this.
> The gory details:
>
> While in shared interpeter context, the call to sv_setsv() that updates
> the (shared interpreter) RV at $a[0] to refer to the new (shared
> interpreter) [] also causes the old thing the RV was pointing to to be
> freed. But a cunning bit of code in sv_unref_flags() doesn't free the old
> referent immediately; instead it is converted into a mortal, and so freed
> later. This is apparently to handle stuff like $a = $a->[0]. However, this
> means that the mortal was added to the tmps stack of the *shared
> interpreter*, which never gets freed. Adding a SAVETMPS/FREETMPS within
> the shared context block of code solved the problem.
Is it safe to add to maint?
[I know that it's desirable to stem the leak, but is it unlikely to add
new bugs?]
> + if (SvTYPE(ssv) == SVt_PVAV) /* #24061 */
> + AvREAL_on(ssv);
Is AvREAL documented anywhere?
> CALLER_CONTEXT;
> }
>
> @@ -436,6 +438,12 @@
> if (target) {
> SV *tmp;
> SHARED_CONTEXT;
> + /* #24255: sv_setsv() (via sv_unref_flags()) may cause a
> + * deferred free with sv_2mortal(). Ensure that the free_tmps
> + * is done within this inpterpreter. DAPM.
> + */
> + ENTER;
> + SAVETMPS;
> tmp = newRV(SHAREDSvPTR(target));
> sv_setsv_nomg(SHAREDSvPTR(shared), tmp);
> SvREFCNT_dec(tmp);
> @@ -444,6 +452,8 @@
> SvOBJECT_on(SHAREDSvPTR(target));
> SvSTASH(SHAREDSvPTR(target)) = (HV*)fake_stash;
> }
> + FREETMPS;
> + LEAVE;
> CALLER_CONTEXT;
> }
> else {
This bit seems safe to me. (But who am I to judge?)
Nicholas Clark