Skip to contentSkip to conversation

Tegra baseline test results for v3.14-rc1

DISCUSSION Paul Walmsley 1 participant 1 message Feb 27, 2014 Tegra architecture support

lore
Share & tools
lore
#0 lore raw

Here are some basic Tegra test results for Linux v3.14-rc1. Logs and other details at:

     http://nvt.pwsan.com/pub/linux/testlogs/test_v3.14-rc1/20140227022109/

Test summary
------------

Build: zImage:
Pass: ( 2/ 2): multi_v7_defconfig, tegra_defconfig

Boot to userspace: multi_v7_defconfig:

     Pass: ( 2/ 2): tegra114-dalmore-a04, tegra30-beaver

Boot to userspace: tegra_defconfig:

     Pass: ( 2/ 2): tegra114-dalmore-a04, tegra30-beaver

vmlinux object size
(delta in bytes from test_v3.13 (d8ec26d7f8287f5788a494f56e8814210f0e64be)):
text data bss total kernel
+1833734 +109664 -460 +1942938 multi_v7_defconfig
+170871 +23512 -6616 +187767 tegra_defconfig

Boot-time memory difference
(delta in bytes from test_v3.13 (d8ec26d7f8287f5788a494f56e8814210f0e64be))
avail rsrvd high freed board kconfig dtb
. . . . tegra114-dalmore-a04 multi_v7_defconfig tegra114-dalmore
. . . . tegra114-dalmore-a04 tegra_defconfig tegra114-dalmore

  -2070932k  -25196k -1317888k    -472k tegra114-dalmore-hea   multi_v7_defconfig     tegra114-dalmore
  -2053808k  -42320k -1317888k    -244k tegra114-dalmore-hea      tegra_defconfig     tegra114-dalmore
   -18320k   18320k        .      36k       tegra30-beaver   multi_v7_defconfig       tegra30-beaver
     -168k     168k        .       4k       tegra30-beaver      tegra_defconfig       tegra30-beaver

The tegra114-dalmore-headless board, an A03 Dalmore, was dropped out of the testbed since apparently NVIDIA folks don't maintain upstream support for any Dalmore revision other than A04. That's the cause of the large boot-time memory deltas above. It's been replaced by an A04 Dalmore. There should have been large boot-time memory deltas on the A04 lines too, but the current set of scripts don't handle that condition well.

I double-checked the +1.9MB size increase for the multi_v7_defconfig kernel, and it appears to be accurate :-(