[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFH: zfs-fuse



On Mon, Feb 02, 2009 at 07:52:03PM +0100, Mike Hommey wrote:
Could someone still give me current policy regarding sparc64 code (see
below for a reminder of the original question)
Mike Hommey wrote:
> Hi,
> > I'm currently giving a try to zfs-fuse (ITP #419746) and it has support
> for sparc64. My understanding of the debian port is that it is mostly a
> 32 bits port, with a 64 bits kernel. The zfs-fuse build scripts only
> deal with sparc64 as returned by uname -m, but builds as 32-bits
> application with the current toolchain. Also, it builds with
> -mcpu=ultrasparc, which I'm unsure is allowed for the debian sparc port
> (as well as expecting uname -m to always return sparc64), but the code
> does use ultrasparc specific opcodes, most notably "cas".

It is my understanding (the porters will correct me if I'm wrong) that
lenny will no longer support SPARC v8 (support for v7 was removed long
ago) because the 32-bit SPARC kernel is no longer supported upstream.
Since v9 systems are UltraSPARC, this shouldn't be a problem.

This also has the pleasant side-effects that multiplication and division
(and other UltraSPARC instructions) are now assumed to be available for
all sparc machines.

However, I suggest you use -mcpu=v9.  -mcpu=ultrasparc chooses v9
instructions, but then optimizes the code for UltraSPARC I/II/IIi; this
may not be optimal for newer UltraSPARC machines, such as III, IV, T1,
and T2.

--
brian m. carlson / brian with sandals: Houston, Texas, US
+1 713 440 7475 | http://crustytoothpaste.ath.cx/~bmc | My opinion only
troff on top of XML: http://crustytoothpaste.ath.cx/~bmc/code/thwack
OpenPGP: RSA v4 4096b 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187

Attachment: signature.asc
Description: Digital signature


Reply to: