Hi, I did a setup with 3.12.6-rt9 kernel on a Beaglebone board with an TI AM335x CPU running at fixed 720MHz. The kernel uses the omap2plus_defconfig configuration that comes with this kernel with additional CONFIG_PREEMPT_RT_FULL switched on. I am observing some strange max. latencies when comparing with kernel 3.2.45-rt66. The results of a similiar system can be found on the OSADL lab site (https://www.osadl.org/Latency-plot-of-system-in-rack-7-slot.qa-latencyplot-r7s5.0.html). With "cyclictest -p99 -m -n -i200 -h400 -q -l10000000" I get ... # Min Latencies: 00013 # Avg Latencies: 00062 # Max Latencies: 04664 ... Also the latency does not differ much from that when doing a flood ping against the system as simple stress. 3.2.45-rt66 comes with a max. latency of 118us! So I expect something to be wrong with 3.12.6-rt9 at least on armv7 or especially this device (max. lat. 4664us). I also noticed an issue with the on-chip ethernet when CONFIG_PREEMPT_RT_FULL is switched on with this kernel: I also noticed that the 3.12.6-rt9 with enabled [ 2.817024] net eth0: initializing cpsw version 1.12 (0) [ 2.825269] net eth0: phy found : id is : 0x7c0f1 [ 2.830713] libphy: PHY 4a101000.mdio:01 not found [ 2.835777] net eth0: phy 4a101000.mdio:01 not found on slave 1 [ 12.846394] Waiting up to 110 more seconds for network. [ 22.846147] Waiting up to 100 more seconds for network. ... Finally networking is fine. But I need to wait for 110 seconds :-) Prior testing with 3.12.6-rt9 I did the same test with 3.12.5-rt7. Same results. But without CONFIG_PREEMPT_RT_FULL I do not get this bringup delay. I could not do this test with rt9 because it does not build without CONFIG_PREEMPT_RT_FULL! Any ideas? Matthias -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
