From: cotulla@yandex.ua Subject: Re: [DISCUSSION] Hexagon code inside kernel Date: Sat, 16 Feb 2013 05:35:13 +0400 Message-ID: <1433341360978513@web2h.yandex.ru> References: <1093431360938494@web22d.yandex.ru> Mime-Version: 1.0 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=yandex.ru; s=mail; t=1360978513; bh=udiA8kPVRe09QHo26lvmKDV/JLaDp9MOmGLWa7QMaC0=; h=From:To:In-Reply-To:References:Subject:Date; b=IWjpfdCal7IJpa14TzBqv4ptcUp2xs6ixqD6F2QYuW4nUApO/A/k3tVqPJWEu9QIR mJkT54NlpV55MEfVIZhv2o7m7ebcI0W1z3H5Ja/8JTQPRw4XnEL+Acs+LSKH/32WIH jEIVSs8Q1PZ2ANm3kTfLWSDOXyCnUNYQwKnKpAG0= In-Reply-To: Sender: linux-hexagon-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="windows-1252" To: linux-hexagon@vger.kernel.org Hello, > =9AIts documented at https://developer.qualcomm.com/hexagon-processor > =9Ahttps://developer.qualcomm.com/download/80-nb419-3ahexagonvirtualm= achinespec.pdf Yes, I saw and active using that document. But it's only a specificatio= n. A lot of things are not really described there. And kernel implementati= on is rather messy -=20 for example, in some places it's using HVM interface for cache operatio= ns and in the another it is executing native commands.=20 > =9AI think it was supposed to be, I'm not sure why its not. =9AThere = had been some > =9Agrand plans for hexagon, but they never quite materialized. =9AIts= far more > =9Apower-efficient than even the most efficient ARM's (I'm guessing i= t might be > =9Athe most power-efficient (32/64-bit) processor ever?), and that wa= s going to > =9Abe its selling point. =9AI think they could not figure out how to = sell > =9Ait, or to whom. Interesting. But is it rather slow? Especially a memory access? > =9AWell, I'm not sure, but I'm guessing that if you prowl around on t= he ROM, > =9Ayou'll find a copy of it in there somewhere.=20 No, it doesn't present in any ROM. I already checked that. As well as it is useless inside without actual Linux image. >=9AAnd if you don't, its actually a=9Amini-VM, its maybe a few hundred= lines of assembly. =9AIt really did very little > =9Aother than enabling and dispatching first-level interrupts. Perhap= s I'm not > =9Athinking clearly, but I'm guessing that you can hack around having= it at all. > =9AI don't remember if there's some must-use-it-or-die undocumented =9A= register > =9Ain there or not. One important part is a software MMU implementation, because Hexagon do= esn't have hardware MMU, only standalone TLB entries. But HVM provides that functionality to the guest. As well as most supervisor mode registers are not really defined, like = "bit 18 of SSR is interrupt mask flag". Of course it's not a problem for myself, I already researched most thin= gs. It should be enough to bring native Hexagon kernel instead of fake hype= rvisor one. I am working with HTC LEO which has QSD8250B/QDSP6v2.=20 I already can run QDSP6 code in polling-bootloader mode. > =9AMost of the other vendors do it to. Certainly, most of the powerpc > =9Acode requires > =9Aan extremely large and complicated hypervisor -- something like mi= llions of > =9Alines of code -- and its extremely proprietary. Well, I guess they provide those hypervisors with hardware kinda like B= IOS? In Hexagon case it's actually dead code.=20 What is the point to include such thing inside mainline kernel? > =9AWhile I was at Qualcomm, I saw development boards that were suppos= ed > =9Ato be generally available for a few hundred $$. =9AThey were basic= ally > =9Acell-phones > =9Awith the radio disabled (to avoid complications with the carriers)= and some > =9Aextra headers soldered on for hardware debuggers. =9AQualcomm was = even > =9Afunding someone to build a very low-cost (sub $500) semi/mostly(?)= open-source > =9Ahardware debugger, =9Asomething that could read the JTAG bits and > =9Ahalt/single-step > =9Athe CPU. =9A=9AThe point of the low-cost was to allow hobbyists to= get > =9Ainto it .. =9Asince > =9Apaying $5K to debug a $200 board only makes sense for NDA-signing = corporations > =9Awho can afford it. Well, it sounds well. But what reason to push code to _mainline_ before= all that is done? Is it really good to include random bullshit into kernel to make it clu= ttered? As well in the neighboring mail list people tells that QCT camera code = can not be included to the mainline because it has private service for = userspace. -Cotulla