Hi Linus, On Friday 29 July 2011 04:04 AM, Linus Walleij wrote: > On Thu, Jul 28, 2011 at 7:04 AM, Minkyu Kang<proms...@gmail.com> wrote: >> Dear Linus Walleij, >> >> On 18 July 2011 19:35, Linus Walleij<linus.wall...@linaro.org> wrote: >>> On Mon, Jul 18, 2011 at 9:23 AM, Lukasz Majewski<l.majew...@samsung.com> >>> wrote: >>> >>>> Disable D cache for goni target (s5p) - rationale: >>>> >>>> 1. Before change introduced in commit >>>> SHA1:c2dd0d45540397704de9b13287417d21049d34c6 the D-cache for >>>> GONI target was disabled. This change was supposed to >>>> preserve this state. >>> >>> I had exactly the same rationale for disabling cache on the Integrator - it >>> used to be off now it's default on and doesn't work - but that patch was >>> NAK:ed by Wolfgang. >>> >> >> Could you please give me the link of that thread? > > Me: > http://marc.info/?l=u-boot&m=131066804010454&w=2
I hadn't noticed this thread before. Here are my thoughts on your problem: I think the fundamental problem is that caching was enabled for ARM without putting the cache maintenance functions in place. At a minimum, we need a few flush and invalidate functions. I worked on this for armv7, but similar things need to be done for earlier architectures if you have to enable caching. Once we have these cache maintenance functions, they need to be used in the following scenarios to avoid coherency issues. 1. Drivers using DMA: i. Flush the buffer before initiating a memory to peripheral DMA. ii. Invalidate the buffer after a peripheral to memory DMA (before MPU reads it) 2. Cleanup before Linux: Flush the entire SDRAM and disable the cache. I think you are hit by not doing 2 properly. If you have dirty entries in the cache when Linux boots you will have coherency issues. I had faced similar issues when I had some issues in armv7 cleanup_before_linux(). I had to do an extra invalidate to solve it. Please see how this has been done for armv7 in arch/arm/cpu/armv7/cpu.c in function cleanup_before_linux(). In some other cases there are problems due to (1) above. In any case, the starting point really is to make sure that you have the required set of maintenance functions for your architecture revision. We may not need maintenance functions for each architecture revision depending on the backward compatibility of the successive revisions. I have not looked into those details though. best regards, Aneesh _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
