Messages in this thread | | | From | Rob Herring <> | Date | Thu, 26 Jun 2014 17:56:43 -0500 | Subject | Re: [PATCH 0/5] ARM: Centralize the access to the SCU register |
| |
On Thu, Jun 26, 2014 at 5:43 PM, Gregory CLEMENT <gregory.clement@free-electrons.com> wrote: > Hello, > > Following the feedback I go on the patch "ARM: mvebu: Enable SCU > Speculative linefills to L2 for Armada 375/38x" : > http://thread.gmane.org/gmane.linux.ports.arm.kernel/335961/focus=335993 > > I take the opportunity of adding new functions in smp_scu.c to > centralize the other access done on the SCU register from C file in > this file. > > The first patch is a preliminary clean-up in smp_scu.c. > > The second and the third patches add functions to manipulate the SCU > control register. > > The forth patch use the new scu_spec_linefills_enable() > function. Enabling SCU Speculative linefills to L2 for Armada 375/38x > was the reason of this series. > > The last patch removed a direct access to the SCU register by an > access through the new scu_standby_enable() function. For this one I > have just checked that the kernel can be built using the > imx_v6_v7_defconfig config, but I didn't test it on an imx6 hardware.
Why would we not just turn on these 2 features unconditionally? If we don't know of any platform where they are broken, then we should just enable them. We can add these functions only if necessary later.
Rob
| |