This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
elision testing
- From: Andi Kleen <andi at firstfloor dot org>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: Torvald Riegel <triegel at redhat dot com>, Carlos O'Donell <carlos at redhat dot com>, GNU C Library <libc-alpha at sourceware dot org>, Ryan Arnold <rsa at us dot ibm dot com>, "Joseph S. Myers" <joseph at codesourcery dot com>, Siddhesh Poyarekar <siddhesh at redhat dot com>, andi <andi at firstfloor dot org>
- Date: Fri, 21 Jun 2013 02:57:03 +0200
- Subject: elision testing
- References: <20130610231903 dot D9C602C09B at topped-with-meat dot com> <51B66222 dot 1040300 at redhat dot com> <20130614224427 dot 4B2532C077 at topped-with-meat dot com> <51BF3CF8 dot 1010901 at redhat dot com> <1371494971 dot 16968 dot 21574 dot camel at triegel dot csb> <20130617193649 dot 7B5872C08D at topped-with-meat dot com> <1371503900 dot 16968 dot 21902 dot camel at triegel dot csb> <20130619224234 dot 5AC132C10E at topped-with-meat dot com> <1371733830 dot 964 dot 1089 dot camel at triegel dot csb> <20130620230930 dot 6F21E2C135 at topped-with-meat dot com>
I also wanted to clarify another thing. A lot of the discussions
in this thread were about compatibility with elision.
I think compatibility with existing programs is very important.
I don't think it is necessary defined by the POSIX standard
(although it is a useful guide as a starting point)
but more a "bug to bug" compatibility that is often
in undefined territory. This doesn't mean that every bug
needs to be faithfully handled, but at least common or
wide spread bugs should be.
It's not something that can be the theoretically determined.
Knowing about the compatibility picture for elision
is very important.
In my mind the only way to get this information is
to enable elision for some subset of users, and see
what (and if any) problems they see, and also what
performance they get.
Only after these reports are in can a useful decision be made
what the default should be.
One issue I have with this proposal with only allowing
recompiled binaries to use elision, is that it will be a big
hurddle to collecting this data effectively, as all existing
binaries will not be tested.
-Andi
--
ak@linux.intel.com -- Speaking for myself only.