Linux 4.7: Reported regressions as of Sunday, 2016-07-17

From: Thorsten Leemhuis
Date: Sun Jul 17 2016 - 12:52:39 EST

Hi! Here is my sixth regression report for Linux 4.7. It lists 8
regressions I'm currently aware of; 2 of them are new.

The report also mentions 3 regressions that were fixed since the last
report(Â). There were a few ones that were reported to me in the past
week (many thx for that!) and fixed already, which I did not add to
the list.

As usual: Please let me know about any regression missing on the list
or if it contains something which shouldn't be there. I'm afk next
weekend, so this likely will be the last regression report before 4.7
gets released. But I guess I'll write at least one or two more reports
after that to make sure regressions found after the release get
collected somewhere.

HTH, CU, Thorsten

(Â) previous reports can be found at

== Current regressions ==

Desc: 795ae7a0de: pixz.throughput -9.1% regression
Repo: 2016-06-02
Stat: 2016-06-22
Note: stalled; hannes can't reproduce and waits for reporter to post results from more tests

Desc: System hang when plug/un-plug USB 3.1 key via thunderbolt port on Dell XPS 13
Repo: 2016-06-14
Stat: 2016-07-16
Note: wip, but root cause hard to track down

Desc: regression in 8250 uart driver
Repo: 2016-06-14
Stat: 2016-07-02
Note: andy has it still on his todo list, but something else has a higher priority

Desc: Multi-thread udp 4.7 regression, bisected to 71d8c47fc653; performance decreased or complete failure with a test
Repo: 2016-06-27
Stat: 2016-07-12
Note: patch "netfilter: conntrack: skip clash resolution if nat is in place" in the net tree fixes this

Desc: kmemcheck: Caught 64-bit read from uninitialized memory; iptables/nf_register_net_hooks
Repo: 2016-06-19
Stat: 2016-07-11
Note: reporter wanted to recheck, nothing new since then

Desc: UBSAN splat in drivers/acpi/acpica/dsutils.c:641:16
Repo: 2016-06-15
Stat: 2016-06-22
Note: wonder if this is important enough to investigate further; related, similar bugs: Not sure which of them are regressions

Desc: After commit f21a21983ef13a, the i915 display is turned off
Repo: 2016-06-16
Stat: 2016-07-15
Note: fix in the works:

Desc: legacy gamma table updates are completely broken for Intel on Linux-4.7-rc7
Repo: 2016-07-12
Stat: 2016-07-14
Note: seems (drm/i915: add missing condition for committing planes on crtc) fixes this

== Fixed since last report ==

Desc: Bad flicker on skylake HQD due to code in the 4.7 merge window
Repo: 2016-05-30
Stat: 2016-07-08
Note: Fixed by vswing issue, maybe primary a Firmware problem; a few patches improved situation, but did not fix the problem

Desc: Kernel does not boot with 7ed18e2d1b6782989eb399ef79a8cc1a1b583b3c ( acpi-4.7-rc7 aka All of these fix recent regressions in ACPICA, in the ACPI PCI IRQ management code and in the ACPI AML debugger.)
Repo: 2016-07-08
Stat: 2016-07-10
Note: fixed by

Desc: ACPI EC problems/ ACPI / EC: Fix an order issue in ec_remove_handlers()
Repo: 2016-07-06
Stat: 2016-07-08
Note: fixed by