Date: Fri, 14 Sep 2001 15:17:04 -0700 From: Mike Smith <msmith@freebsd.org> To: Warner Losh <imp@harmony.village.org> Cc: Mike Smith <msmith@FreeBSD.ORG>, Yar Tikhiy <yar@FreeBSD.ORG>, hackers@FreeBSD.ORG, msmith@mass.dis.org, msmith@mass.dis.org Subject: Re: Driver structures & alignment Message-ID: <200109142217.f8EMH4504153@mass.dis.org> In-Reply-To: Message from Warner Losh <imp@harmony.village.org> of "Fri, 14 Sep 2001 15:02:45 MDT." <200109142102.f8EL2jt24452@harmony.village.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> In message <200109132309.f8DN9vX03967@mass.dis.org> Mike Smith writes:
> : Any form of packing is going to cause problems for items that are
> : located in illegal fashions.
>
> It is almost more appropriate to use bus_space_{read,write} than using
> memory mapped structures. The bus macros will work for otherwise
> unaligned thigns, while direct access won't. Also, the
> compiler/system may optimize things in strange ways that cause
> problems. Do yourself a big favor and don't go down this path.
The most common use for alignment-sensitive data structures nowadays
is in control structures exchanged with devices via busmaster DMA.
Needless to say, the bus_space functions aren't very useful for this, and
controlled alignment is necessary.
Even then, I really want a mechanical way of generating a stub function to
correct endianness in a structure based on the structure definition; we have
a lot of drivers that are explicitly little-endian due to all this.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200109142217.f8EMH4504153>