Date: Wed, 6 Mar 1996 17:46:07 -0700 From: Nate Williams <nate@sri.MT.net> To: Dave Boynton <dboynton@clark.net> Cc: questions@freebsd.org Subject: Re: FAQ erroneous Message-ID: <199603070046.RAA29317@rocky.sri.MT.net> In-Reply-To: <199603062248.RAA07920@mail.Clark.Net> References: <199603062248.RAA07920@mail.Clark.Net>
next in thread | previous in thread | raw e-mail | index | archive | help
> Answer 7.5 of the FAQ seems to suggest that the most common cause of > signal 11 errors is faulty memory. This is true. > Anyone who knows unix, knows that > segmentation violations are caused by dereferencing an invalid pointer, > by accessing beyond the end of your process' allocated memory, or > possibly by writing over your stack and returning to an invalid address. Yes, and this is caused when faulty memory (or cache) writes invalid data all over the place, which causes signal 11's. This is *extremely* typical with x86 hardware. > To suggest to possibly novice users that all of their signal 11 > errors are due to faulty memory is a disservice to the FreeBSD community, > and may prevent users from sending otherwise valid bug reports in which > involve segmentation violations. Perhaps that's why I'm experiencing a > bug involving signal 11 with both version 2.05-RELEASE and 2.1-RELEASE on > two different machines and the same program - no one sent in a bug report > because they thought it was their memory! Generally speaking, if you get re-producible signal 11's, it's probably not memory. The FAQ should be updated to reflect the difference between reproducible signal 11's (which always occur at the same place) and non-reproducible ones which cause compiles to fail. When you restart the compile everything works again until a later point, etc... Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603070046.RAA29317>