[opensuse-arm] Getting the correct kernel
I know one of the reasons we chose a single target for hardware was to make the job of getting the right kernel and associated modules somewhat easier. Now (un)fortunately, we have some kind sponsors of hardware Genesi with the Efika MX and Pandaboard.org with the PandaBoard. How difficult would it be to have a single kernel that supports both these devices? I've had a sniff on the web and have found a kernel repo for EfikaMX[0] and PandaBoard[1], there is also the Linaro kernel[2] which I believe supports them; nice to have would be the BeagleBoard too. I believe they should all be supported in upstream but I can't say for certain, saying that though I believe the EfikaMX needs the siihdmi driver pushing upstream for the 3.x kernel. I have a contact to push about this and will chase him up later today. Regards, Andy 0=http://gitorious.org/efikamx/linux-kernel 1=http://gitorious.org/pandaboard/kernel-mainline 2=http://git.linaro.org/gitweb?p=kernel/linux-linaro-3.1.git;a=summary -- Andrew Wafaa IRC: FunkyPenguin GPG: 0x3A36312F -- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
On Thu, October 13, 2011 7:05 am, Andrew Wafaa wrote:
I know one of the reasons we chose a single target for hardware was to make the job of getting the right kernel and associated modules somewhat easier.
Now (un)fortunately, we have some kind sponsors of hardware Genesi with the Efika MX and Pandaboard.org with the PandaBoard. How difficult would it be to have a single kernel that supports both these devices? I've had a sniff on the web and have found a kernel repo for EfikaMX[0] and PandaBoard[1], there is also the Linaro kernel[2] which I believe supports them; nice to have would be the BeagleBoard too.
Is the Efika MX this device? http://www.genesi-usa.com/products/smartbook If this is the case, this one has a Freescale i.MX515 (ARM Cortex-A8 800MHz). In that case I think we might have another problem. We compile with option. -march=armv7-a -mtune=cortex-a9 I don't know if this will cause problems on Cortex A8 hardware? Or do we need to change to: -mtune=cortex-a8 ?
I believe they should all be supported in upstream but I can't say for certain, saying that though I believe the EfikaMX needs the siihdmi driver pushing upstream for the 3.x kernel. I have a contact to push about this and will chase him up later today.
Regards,
Andy
0=http://gitorious.org/efikamx/linux-kernel 1=http://gitorious.org/pandaboard/kernel-mainline 2=http://git.linaro.org/gitweb?p=kernel/linux-linaro-3.1.git;a=summary
-- Andrew Wafaa IRC: FunkyPenguin GPG: 0x3A36312F
-- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
Joop. -- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
On Thu, 2011-10-13 at 12:45 +0200, Joop Boonen wrote:
Is the Efika MX this device? http://www.genesi-usa.com/products/smartbook If this is the case, this one has a Freescale i.MX515 (ARM Cortex-A8 800MHz).
In that case I think we might have another problem. We compile with option. -march=armv7-a -mtune=cortex-a9
I don't know if this will cause problems on Cortex A8 hardware? Or do we need to change to: -mtune=cortex-a8 ?
It is indeed that device. I'm not 100% sure, as I know Michal and Pavol have a BeagleBoard which is also a Cortex A8 device. I'll admit that my knowledge on the SoC side of things is limited, but I was under the impression that there wasn't a huge difference between the A8 and the A9. I've also just checked with the guys and gals in #linaro and the message was "it will be just fine". Regards, Andy -- Andrew Wafaa IRC: FunkyPenguin GPG: 0x3A36312F -- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
On 10/13/2011 01:37 PM, Andrew Wafaa wrote:
On Thu, 2011-10-13 at 12:45 +0200, Joop Boonen wrote:
Is the Efika MX this device? http://www.genesi-usa.com/products/smartbook If this is the case, this one has a Freescale i.MX515 (ARM Cortex-A8 800MHz).
In that case I think we might have another problem. We compile with option. -march=armv7-a -mtune=cortex-a9
I don't know if this will cause problems on Cortex A8 hardware? Or do we need to change to: -mtune=cortex-a8 ?
It is indeed that device. I'm not 100% sure, as I know Michal and Pavol have a BeagleBoard which is also a Cortex A8 device.
I'll admit that my knowledge on the SoC side of things is limited, but I was under the impression that there wasn't a huge difference between the A8 and the A9. I've also just checked with the guys and gals in #linaro and the message was "it will be just fine".
Yes, -mtune only makes sure that we use gcc's instruction optimizer with A9 numbers. So if an instruction is fast on A9, but slow on A8, gcc will still generate code that runs fast on A9 but slower on A8. However, the code will run. And it won't be a major loss in performance either. Alex -- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
On Thu, October 13, 2011 1:51 pm, Alexander Graf wrote:
On 10/13/2011 01:37 PM, Andrew Wafaa wrote:
On Thu, 2011-10-13 at 12:45 +0200, Joop Boonen wrote:
Is the Efika MX this device? http://www.genesi-usa.com/products/smartbook If this is the case, this one has a Freescale i.MX515 (ARM Cortex-A8 800MHz).
In that case I think we might have another problem. We compile with option. -march=armv7-a -mtune=cortex-a9
I don't know if this will cause problems on Cortex A8 hardware? Or do we need to change to: -mtune=cortex-a8 ?
It is indeed that device. I'm not 100% sure, as I know Michal and Pavol have a BeagleBoard which is also a Cortex A8 device.
I'll admit that my knowledge on the SoC side of things is limited, but I was under the impression that there wasn't a huge difference between the A8 and the A9. I've also just checked with the guys and gals in #linaro and the message was "it will be just fine".
Yes, -mtune only makes sure that we use gcc's instruction optimizer with A9 numbers. So if an instruction is fast on A9, but slow on A8, gcc will still generate code that runs fast on A9 but slower on A8. However, the code will run. And it won't be a major loss in performance either.
Great to hear that it'll also work on Cortex A8. I was under the impression that it would use code that might use instructions that only work on the Cortex A9.
Alex
-- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
Joop. -- To unsubscribe, e-mail: opensuse-arm+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-arm+owner@opensuse.org
participants (3)
-
Alexander Graf -
Andrew Wafaa -
Joop Boonen