Linux 4.17 Stable Has Been Settling Well, Coming Soon To Fedora

[nuetzel]

It should also arrive in Tumbleweed, it is being tested.

What, should...?
Kernel:stable is your friend.

since 4.17.0

[nuetzel]

'...if you have run into any upsets or issues with Linux 4.17...'

Since 4.16.0 kernel-pae (yes, for 'old' 32 bit x86) boxes:

[ 0.000000] BUG: Bad page state in process swapper pfn:3c000
[ 0.000000] page:f49d2000 count:0 mapcount:1 mapping:00000000 index:0x0
[ 0.000000] flags: 0x0()
[ 0.000000] raw: 00000000 00000000 00000000 00000000 00000000 00000100 00000200 00000000
[ 0.000000] raw: 00000000
[ 0.000000] page dumped because: nonzero mapcount
[ 0.000000] Modules linked in:
[ 0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 4.17.1-1.ge721478-pae #1 openSUSE Tumbleweed (unreleased)
[ 0.000000] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./PV530, BIOS P1.00 06/30/2010
[ 0.000000] Call Trace:
[ 0.000000] dump_stack+0x7d/0xbd
[ 0.000000] bad_page.cold.111+0x90/0xc7
[ 0.000000] free_pages_check_bad+0x52/0x70
[ 0.000000] free_pcppages_bulk+0x37d/0x570
[ 0.000000] free_unref_page_commit+0x9a/0xc0
[ 0.000000] free_unref_page+0x6a/0xa0
[ 0.000000] __free_pages+0x17/0x30
[ 0.000000] free_highmem_page+0x1e/0x50
[ 0.000000] add_highpages_with_active_regions+0xd6/0x113
[ 0.000000] set_highmem_pages_init+0x67/0x7d
[ 0.000000] mem_init+0x23/0x1d9
[ 0.000000] start_kernel+0x1c2/0x437
[ 0.000000] i386_start_kernel+0x98/0x9c
[ 0.000000] startup_32_smp+0x164/0x168

[euler271]

Well, hasn't been running well on my desktop. The AMDGPU driver is now totally broken for my R9 390 - refuses to boot (or atleast, the screen is completely blank a second after kernel startup - modeset doesn't succeed).

I think I should just quit trying to be an Android developer, and focus entirely on fixing this crappy AMDGPU driver, or write my own driver.

that doesn't look like a bad option at all. does it? you should go for it. if more ppl work on the low level stuff, you(an many more) wouldnt be having this kind of issues in the first place.

[JeansenVaars]

Can't wait to check this out in Fedora for battery life on my Yoga 920

[grigi]

I have a problem with the 4.17 kernel on my Dell Latitude E5470 (skylake notebook).
When I suspend, it appears to suspend correctly. When I resume, it will most of the time (¾) reboot instead of resume.

Downgraded to 4.16 and problem went away.

This is my daily driver system, so I can't spend a lot of time debugging it, unfortunately.

I think I saw some patches lined up for 4.19 that talk about resume issues using the intel driver.

[rycodge]

Hmmm when writing any value to pp_power_profile_mode with Polaris or Vega on Linux 4.17 or 4.18 Git, I just get a write error for invalid argument... Tried upgrading the firmware too, no luck. Need to dig back through the patches to see what I am missing. bridgman is there something silly being overlooked? Following the comment from one of the patches, but looks like I must have overlooked something for trying to get the POWER_SAVING mode.

I've made some progress on 4.17.1.

This is required to get around the "invalid argument error":

echo manual >  /sys/class/drm/card0/device/power_dpm_force_performance_level

After setting it to manual you can change modes like so:

echo 4 >  /sys/class/drm/card0/device/pp_power_profile_mode

cat /sys/class/drm/card1/device/pp_power_profile_mode

As you can see, I've set it to compute mode:

NUM        MODE_NAME     SCLK_UP_HYST   SCLK_DOWN_HYST SCLK_ACTIVE_LEVEL     MCLK_UP_HYST   MCLK_DOWN_HYST MCLK_ACTIVE_LEVEL
  0   3D_FULL_SCREEN:        0              100               30                0              100               10
  1     POWER_SAVING:       10                0               30                -                -                -
  2            VIDEO:        -                -                -               10               16               31
  3               VR:        0               11               50                0              100               10
  4        COMPUTE *:        0                5               30                0              100               10
  5           CUSTOM:        -                -                -                -                -                -

However, I'm not able to set custom values such as SCLK_DOWN, etc. Hitting a wall right now.

[rycodge]

Figured out the custom entry:

echo "5 1 5 0 30 1 10 1 1" > /sys/class/drm/card1/device/pp_power_profile_mode

Results in:

NUM        MODE_NAME     SCLK_UP_HYST   SCLK_DOWN_HYST SCLK_ACTIVE_LEVEL     MCLK_UP_HYST   MCLK_DOWN_HYST MCLK_ACTIVE_LEVEL
  0   3D_FULL_SCREEN:        0              100               30                0              100               10
  1     POWER_SAVING:       10                0               30                -                -                -
  2            VIDEO:        -                -                -               10               16               31
  3               VR:        0               11               50                0              100               10
  4          COMPUTE:        0                5               30                -                -                -
  5         CUSTOM *:        5                0               30               10                1                1

credit to: https://git.kernel.org/pub/scm/linux...35b904e34b3fce

and https://lists.freedesktop.org/archiv...il/021576.html

[nanonyme]

Phoronix: Linux 4.17 Stable Has Been Settling Well, Coming Soon To Fedora

Since the release of Linux 4.17 almost two weeks ago, I haven't heard of any horror stories, Linux 4.17 continues running excellent on all of my test systems, the 4.17.1 point release was quite small, and more distributions are gearing up to ship this latest kernel release...

http://www.phoronix.com/scan.php?pag...ion-Doing-Well

How about starting to link to kernel bugzilla in this kind of posts?

[Charlie68]

What, should...?
Kernel:stable is your friend.

since 4.17.0

What is not clear? I just said it's also coming in Tumbleweed and right now it's being tested openQA !

[StefanBruens]

What, should...?
Kernel:stable is your friend.

since 4.17.0

Kernel:stable is not Tumbleweed. Kernel:stable is the source for Factory and, after openQA testing, Tumbleweed. Having a specific kernel version currently being tested by openQA implies it has been in Kernel:stable before.