> -----Original Message----- > From: Koul, Vinod > Sent: Friday, February 13, 2015 4:37 PM > To: Jie, Yang > Cc: Andy Shevchenko; dmaengine@xxxxxxxxxxxxxxx; Girdwood, Liam R > Subject: Re: [PATCH] dmaengine: dw: don't handle interrupt when > dmaengine is not used > > On Wed, Feb 11, 2015 at 05:42:43PM +0530, Jie, Yang wrote: > > > -----Original Message----- > > > From: Koul, Vinod > > > Sent: Wednesday, February 11, 2015 9:19 AM > > > To: Andy Shevchenko > > > Cc: Jie, Yang; dmaengine@xxxxxxxxxxxxxxx; Girdwood, Liam R > > > Subject: Re: [PATCH] dmaengine: dw: don't handle interrupt when > > > dmaengine is not used > > > > > > On Tue, Jan 20, 2015 at 11:36:05AM +0200, Andy Shevchenko wrote: > > > > On Tue, 2015-01-20 at 17:12 +0800, Jie Yang wrote: > > > > > When dma controller is not used by any user and set off, we > > > > > should disble interrupt handler, at least the interrupt reset > > > > > part, for some subsystem, e.g. ADSP, may use the dma in its own > > > > > logic, here reset the interrupt may make this subsystem work > abnormally. > > > > > > > > I think like on Intel MID (Medfield) there should be a specific > > > > register to route interrupt pin from ADSP to CPU and vise versa. > > > > So, my understanding we have to utilize that register somehow. > > > > > > > > Since it's quite specific to ADSP HW I would consider that ADSP > > > > driver would take care of this. > > > > > > > > Vinod, it seems we already discussed this earlier, though I'm not > > > > quite sure I remember your point. Can you share your opinion? > > > Yes using those registers is recommended. Not handling a kernel irq > > > seems not a good option. > > > > > > Do we have an updated patch for this now? > > [Keyon] Andy sent out a update patch on 2015/1/27: > > [PATCH] dmaengine: dw: disable BLOCK interrupts in dw_dma_off() What's > > your opinion for that solution? > Thats not a right solution either. The _right_ solution is to use mask registers > so that interrupt is delivered to the target that needs them. [Keyon] just come back from CNY vacation, so which register do you mean we should mask: ISC/IMC, ISD/IMD, or STSINT(MSKERR, MSKDSTTRAN, MSKSRCTRAN, MSKBLOCK and MSKTFR)? Masking IMC can't really prevent DMA receive the interrupt, and it's not easy to mask STSINT in audio driver side. > > -- > ~Vinod -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
