Comments

56 Comments
BZ#1759358 uboot-tools-2019.10 is available
karma

Got bitten by this badly. This fix seems to fix it, as confirmed by not seeing hundreds of zombie processes after upgrading to it. Thanks!

BZ#1656939 Incron causes defunct processes

I hit bug #1631970. Verifying that this update resolves it.

BZ#1631970 openssh host identification changed - without actually changing
karma

This change seems out of line with what I understood to be the plan of record.

@whot Yeah. As soon as you touch it, it jumps to the top left corner, and no manipulation of the pointer causes other movement. However, using the Logitech mouse works fine.

It's reproducable on multiple systems (by plugging in the USB keyboard). On one, I actually had three mice: there's one on the keyboard of the laptop in the docking station plus the external-keyboard mouse plus an additional Logitech mouse I plugged in to debug. Only the USB Keyboard Mouse exhibited the problem. I'll file a bug but it won't be until tomorrow evening as I have a lot of meetings during the day.

Also note that the trackpoint device built into the laptop works just fine. It's only the external device that's affected.

karma

Yeah, 1.10.902-1 has the same failure. Note that evtest seems to show completely reasonable

type 2 (EV_REL), code 1 (REL_Y), value -12
type 2 (EV_REL), code 0 (REL_X), value 1

events, even as the cursor fails to move.

I have an external keyboard, Lenovo ThinkPad Compact USB Keyboard with TrackPoint, and with libinput-1.10.901-1.fc28, merely touching the pointer makes the cursor jump to the top left and not move. Testing now with 1.10.902-1.fc28

karma

Confirming that this works with a test password with an initial capital.

karma

This works for me. Can we push it out more quickly than waiting for the batch, due to the high profile of the issues it mitigates?

Test Case firefox browse
Test Case firefox media

There is currently a released update for docker which relies on a version of atomic-registries newer than the released version of that. (Like, for example this one.) Assuming the functionality checks out (and it looks like it does from comments), please expedite this update.

karma

Confirmed that the reproducer in bug #1440287 no longer hangs on 100 runs. Also, rest of system seems to work.

BZ#1440287 glibc: Support broken applications which call __tls_get_addr with an unaligned stack (GCC bug workaround)
karma

Updated my system with this a couple of hours ago and have been running existing and new VMs under that. It's an Intel (Skylake) system so I can't test the AMD bug, but general functionality seems... functional.

I have concerns about this. Primarily, the JPEG images are of visibly low quality. The fine detail in the trees here is like a torture-test for JPEG compression. I can see obvious-as-JPEG-artifacts with compression in GIMP at 95% (and these seem to have been saved at 90%). To my eye, at 98% or higher there is still some visible loss but it blends in to the overall texture and isn't discernible as JPEG at least. One possible way to save more space is to use chroma subsampling, which does have a subtle overall effect but is less obvious on inspection without a reference.

Also:

  1. I'm worried about losing the PNG image this late in the cycle, because I'm not sure what all might be referencing it.

  2. The dawn, dusk, and night images aren't just color shifts; the hills behind the trees change in weird ways.

  3. In the previous version, there was no "foreground fog" on the dawn, dusk, and night images, but here it is there in all three with no change in color, which means there's a lot of contrast at night. I'm not opposed to the "fog at night" look, but I think the texture of the brush jumps out too much here.

Tested on Workstation and Cloud Base.

BZ#1466675 updates-testing is still enabled
BZ#1466908 Fedora 26 Final build of fedora-release needed
BZ#1449752 blivet-gui desktop file should not be installed by default in Fedora Workstation