Re: [PATCHES] SuperH Build fixes

[Guy Harris <gharris () sonic net>]

On Wed, Mar 19, 2003 at 04:16:31PM -0500, Paul Mundt wrote:
> Presently libpcap and tcpdump fail in the unaligned access test.. the
> attached patches disable the test for both.

I've checked that in, but, when I asked, you said that the test failed
because the test program dumped core; the script *should*, if the test
program dumps core, define LBL_ALIGN to indicate that unaligned
accesses aren't allowed - i.e., the test program is *supposed* to crash
on those platforms, that's how the configure script figures out that
unaligned accesses aren't OK!

So presumably there's some other problem with the configure script where
the core dump from the test program doesn't cause the script to see an
empty output file from the test program, and thus doesn't cause the
script to set ac_cv_lbl_unaligned_fail to "yes".  What is it not doing
correctly?

This brings up another issue - the "compile this program and try running
it" scheme doesn't work if you're cross-compiling.  Should we just
define a set of CPUs where unaligned accesses doesn't work, and set
ac_cv_lbl_unaligned_fail to "yes" on those, define *another* set of CPUs
where it *does* work (on all implementations - I think some POWER-family
CPUs support them and others don't, for example, but I'd treat that as a
"doesn't work" platform, for binary compatibility) and set
ac_cv_lbl_unaligned_fail to "no" on those, and have the script fail on
others and, when people report the failure, add the new CPU to the
appropriate category?
-
This is the TCPDUMP workers list. It is archived at
http://www.tcpdump.org/lists/workers/index.html
To unsubscribe use mailto:tcpdump-workers-request () tcpdump org?body=unsubscribe