From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stefan Roese Subject: Re: [PATCH] [PPC32] ADMA support for PPC 440SPe processors. Date: Sat, 17 Mar 2007 19:43:27 +0100 Message-ID: <200703171943.27465.sr@denx.de> References: <20070315232956.37DAB353A6C@atlas.denx.de> <200703170909.13869.sr@denx.de> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Content-Disposition: inline Sender: linux-raid-owner@vger.kernel.org To: linuxppc-dev@ozlabs.org Cc: Dan Williams , linux-raid@vger.kernel.org List-Id: linux-raid.ids On Saturday 17 March 2007 19:17, Dan Williams wrote: > Yes, defaulting to 'y' is not necessary, but ASYNC_TX_DMA=y && > DMA_ENGINE=n is an explicit feature of the interface. When DMA_ENGINE > is not selected all the asynchronous paths in the API are compiled > out. This allows subsytems, like md-raid5, to be written in an > asynchronous fashion without regard for the architecture[1] or > availability of offload engines. The current implementation builds on my embedded PPC4xx system without any disks the objects async_tx.o and xor.o into the kernel which I definitely don't need and want. And I get something like: async_tx: api initialized (sync-only) xor: measuring software checksumming speed 8regs : 145.000 MB/sec 8regs_prefetch: 115.000 MB/sec 32regs : 176.000 MB/sec 32regs_prefetch: 135.000 MB/sec xor: using function: 32regs (176.000 MB/sec) upon bootup. Best regards, Stefan