Failed boards:
Successful boards:
apc vt8500_v6_v7_defconfig : passed 1:14.72 arndale exynos_defconfig : passed 1:06.89 bbb omap2plus_defconfig : passed 1:36.67 bbb multi_v7_defconfig : passed 1:18.62 beaver tegra_defconfig : passed 0:52.98 beaver multi_v7_defconfig : passed 0:42.07 capri bcm_defconfig : passed 0:42.45 capri multi_v7_defconfig : passed 0:43.31 cubie sunxi_defconfig : passed 0:53.48 cubie multi_v7_defconfig : passed 0:57.67 cubie2 sunxi_defconfig : passed 0:50.21 cubie2 multi_v7_defconfig : passed 1:04.19 cubie2 multi_lpae_defconfig : passed 0:51.22 cubox dove_defconfig : passed 0:56.71 dalmore tegra_defconfig : passed 1:21.36 dalmore multi_v7_defconfig : passed 1:04.17 dalmore multi_lpae_defconfig : passed 1:02.16 panda omap2plus_defconfig : passed 1:16.82 panda multi_v7_defconfig : passed 1:07.62 sama5 sama5_defconfig : passed 1:56.48 seaboard tegra_defconfig : passed 1:02.07 seaboard multi_v7_defconfig : passed 0:56.20 snow exynos_defconfig : passed 1:18.31 snowball u8500_defconfig : passed 1:30.90 snowball multi_v7_defconfig : passed 1:04.34 trimslice tegra_defconfig : passed 0:58.93 trimslice multi_v7_defconfig : passed 0:48.86 wandboard imx_v6_v7_defconfig : passed 0:55.28 wandboard multi_v7_defconfig : passed 0:54.45
Board legend is available at http://arm-soc.lixom.net/boards.html
On Thu, Jan 16, 2014 at 05:01:31PM -0800, Olof's autobooter wrote:
bbb omap2plus_defconfig : passed 1:36.67 bbb multi_v7_defconfig : passed 1:18.62
Here's an interesting question...
beaver tegra_defconfig : passed 0:52.98 beaver multi_v7_defconfig : passed 0:42.07
Why should a board...
cubie sunxi_defconfig : passed 0:53.48 cubie multi_v7_defconfig : passed 0:57.67
booting its own kernel...
cubie2 sunxi_defconfig : passed 0:50.21 cubie2 multi_v7_defconfig : passed 1:04.19 cubie2 multi_lpae_defconfig : passed 0:51.22
vs booting a multiplatform kernel...
dalmore tegra_defconfig : passed 1:21.36 dalmore multi_v7_defconfig : passed 1:04.17 dalmore multi_lpae_defconfig : passed 1:02.16
take significantly different...
snowball u8500_defconfig : passed 1:30.90 snowball multi_v7_defconfig : passed 1:04.34
amounts of time?
It may be that multi_v7_defconfig doesn't contain everything that the platform needs, but some of these figures are significantly different, such as the last one where there's something like a 50% increase over multi_v7_defconfig, which I doubt can be attributed to different numbers of drivers being probed.
On Fri, Jan 17, 2014 at 01:05:58AM +0000, Russell King - ARM Linux wrote:
On Thu, Jan 16, 2014 at 05:01:31PM -0800, Olof's autobooter wrote:
bbb omap2plus_defconfig : passed 1:36.67 bbb multi_v7_defconfig : passed 1:18.62Here's an interesting question...
beaver tegra_defconfig : passed 0:52.98 beaver multi_v7_defconfig : passed 0:42.07Why should a board...
cubie sunxi_defconfig : passed 0:53.48 cubie multi_v7_defconfig : passed 0:57.67booting its own kernel...
cubie2 sunxi_defconfig : passed 0:50.21 cubie2 multi_v7_defconfig : passed 1:04.19 cubie2 multi_lpae_defconfig : passed 0:51.22vs booting a multiplatform kernel...
dalmore tegra_defconfig : passed 1:21.36 dalmore multi_v7_defconfig : passed 1:04.17 dalmore multi_lpae_defconfig : passed 1:02.16take significantly different...
snowball u8500_defconfig : passed 1:30.90 snowball multi_v7_defconfig : passed 1:04.34amounts of time?
It may be that multi_v7_defconfig doesn't contain everything that the platform needs, but some of these figures are significantly different, such as the last one where there's something like a 50% increase over multi_v7_defconfig, which I doubt can be attributed to different numbers of drivers being probed.
There's a massive amount of noise in these numbers, since they include the whole time from power cycle, to breaking into u-boot, to setting variables and tftp:ing the kernel, etc.
I should probably break apart the time reporting a bit more so it's easier to tell what is what. It's not really relevant whether tftp:ing the image takes longer in some cases (and it removes some of the value of reporting boot times, really).
This explains the few-seconds changes. multiplatform configs being faster to boot is quite likely due to missing drivers in many cases -- tegra has a set of patches for 3.14 to fix theirs, others should also be addressed at some point.
-Olof
kernel-build-reports@lists.linaro.org
-
Olof Johansson -
Olof's autobooter -
Russell King - ARM Linux