Front page | perl.perl5.porters |
Postings from January 2014
Matthew Horsfall wrote:
> ==28093== Invalid write of size 8
> ==28093== at 0xA472891: XS_Data__Clone_clone (Data-Clone.xs:395)
Could you get a Perl backtrace at that point?
E.g., set a breakpoint on XS_Data__Clone_clone and then use:
gdb) p Perl_warn("") /* no threads */
gdb) p Perl_warn(my_perl,"") /* threads */
Then edit the cited sourced file and put Carp::cluck() where
appropriate.
Then re-run the whole thing through valgrind.
The output of that would be much appreciated.
> ==28093== Address 0x592fd20 is 16 bytes inside a block of size 1,024 free'd
...
> ==28093== by 0x5285CC: Perl_stack_grow (scope.c:38)
> ==28093== by 0x55BDC2: Perl_do_kv (doop.c:1268)
That line is an EXTEND. So somehow the entersub call above is using a
reallocated stack, which is weird.
Thread Previous
|
Thread Next