David Lee wrote:
In what follows the term "Solaris" is probably wider than that. A better
reading might turn out to be:
"OSes and/or architectures in which multibyte operations (e.g. assign a
16-bit quantity) require a multibyte alignment (e.g. word, doubleword,
etc)."
So understand my use of the word "Solaris" as a shorthand for that!
Firstly: a general query to all: Have any "Solaris" users successfully
run "ipctest" (under "lib/clplumbing")? Success/failure?
For me, it fails, as the client is trying to write "String-2". After much
tedious insertion of debugging statements, I have pinned it down. The
"client" process of ipctest crashes in "socket_resume_io_write()" at the
statement:
head->msg_len = msg->msg_len
Sure enough "head", on this occasion, is not cleanly aligned for an "int".
(I suspect that there will also be similar assignments with similar
problems: for instance I see one in the line following the one above!)
Further when I run this on Linux (although "Intel" might be a better
technical interpretation) all is well: these unaligned assignments are
still present, but work fine.
Secondly: a specific query to the developers: Could you confirm that the
above analysis is reasonable? If so, how do we set about addressing this
problem, and its instances? Should we be using "memcpy" for such
instances?
Malloc is supposed to ALWAYS return storage which is sufficiently
aligned for any purpose. It sounds like it's not. (didn't this come up
before on Solaris?)
What libc are you using?
--
Alan Robertson <[EMAIL PROTECTED]>
"Openness is the foundation and preservative of friendship... Let me
claim from you at all times your undisguised opinions." - William
Wilberforce
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
