develooper Front page | perl.perl5.porters | Postings from June 2013

Re: Storable refactoring, was Re: [perl #118139] Storable in DESTROYblocks

Thread Previous | Thread Next
From:
Salvador Fandino
Date:
June 13, 2013 08:17
Subject:
Re: Storable refactoring, was Re: [perl #118139] Storable in DESTROYblocks
Message ID:
1371111444.60334.YahooMailNeo@web163106.mail.bf1.yahoo.com
On 06/12/2013 09:29 PM, Reini Urban wrote:

> And you cannot just remove the hints to fix gcc -O3 optimizer problems.

The Linux hints file says:

# gcc -O3 (and higher) can cause code produced from Storable.xs that
# dumps core immediately in recurse.t and retrieve.t, in is_storing()
# and last_op_in_netorder(), respectively.  In both cases the cxt is
# full of junk (and according to valgrind the cxt was never stack'd,
# malloc'd or free'd).  Observed in Debian 3.0 x86, with gccs 2.95.4
# 20011002 and 3.3, and in Redhat 7.1 with gcc 3.3.1. The failures
# happen only for unthreaded builds, threaded builds work okay.

"is_stored" and "last_op_in_netorder" have been completely rewritten,
and the old hairy context handling code is also gone. I don't see a
reason to keep the work around unless it is proven that compiling with
-O3 generates bad code yet.

On the other hand, it seems I overlooked the HP-UX hints.

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About