From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Shimoda, Yoshihiro" Date: Thu, 12 Jan 2012 01:09:11 +0000 Subject: Re: [PATCH v4 2/6] sh: add a resource name for shdma Message-Id: <4F0E32B7.7060902@renesas.com> List-Id: References: <4F0D39F5.90208@renesas.com> In-Reply-To: <4F0D39F5.90208@renesas.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-sh@vger.kernel.org 2012/01/11 18:05, Guennadi Liakhovetski wrote: > Unrelated to your patch, just something that caught my eye: > [ snip ] > > The above two platforms seem incorrect. Back then, as I switched sh_dma.c > to use resources for interrupts [1], I just reproduced the behaviour of > the original driver, but a verification with the header or the datasheet > shows, that sh7780 and sh7785 have separate interrupts per channel, so, if > anyone cares about those platforms, they might need a fix. The interrupts of SH7780 DMAC and Sh7785 DMAC are strange. Would you check the commit bdaa6e8062 "sh: multiple vectors per irq - base"? - The SH7780 and SH7785 have separate INTEVT per channel, but the INT2MSKR for the DMAC has one bit. -- In this case, the commit log says: "If multiple vectors shar the same enum then only the first vector will be avilable as a linux irq." - I think that the DMTE4_IRQ of SH7780 and SH7785 in the "arch/sh/include/cpu-sh4/cpu/dma-sh4a.h" is breaking. But, the old DMA API driver doesn't use this actually because the get_dmte_irq() in dma-sh.c will change the IRQ number... Best regards, Yoshihiro Shimoda > Thanks > Guennadi > > [1] http://thread.gmane.org/gmane.linux.ports.sh.devel/7269/focusr70 > --- > Guennadi Liakhovetski, Ph.D. > Freelance Open-Source Software Developer > http://www.open-technology.de/ > -- Yoshihiro Shimoda EC No.