[BUG] 4.17-rcX delays during boot and memory mapping errors

[BUG] 4.17-rcX delays during boot and memory mapping errors

[Bob Tracy]

Apologies for not having more in the way of diagnostic info.

Every one of the 4.17-rcX series has hung during boot at the point where
"acpid" starts up.  Beginning with -rc5, the boot actually proceeds to
completion after an "uncomfortably" long delay.

Attempting to run the X11 server results in the following diagnostic
output in "/var/log/Xorg.0.log":

(...)
[   109.951] (--) Depth 24 pixmap format is 32 bpp
[   109.951] (II) RADEON(0): [DRI2] Setup complete
[   109.951] (II) RADEON(0): [DRI2]   DRI driver: r600
[   109.951] (II) RADEON(0): [DRI2]   VDPAU driver: r600
[   109.951] Failed to map cursor buffer memory
[   109.951] Failed to map cursor buffer memory
[   109.951] Failed to map cursor buffer memory
[   109.952] Failed to map cursor buffer memory
[   109.952] Failed to map cursor buffer memory
[   109.952] Failed to map cursor buffer memory
[   109.952] (II) RADEON(0): Front buffer size: 8160K
[   109.952] (II) RADEON(0): VRAM usage limit set to 218167K
[   109.952] (==) RADEON(0): DRI3 disabled
[   109.952] (==) RADEON(0): Backing store enabled
[   109.952] (II) RADEON(0): Direct rendering enabled
[   109.952] (II) EXA(0): Driver allocated offscreen pixmaps
[   109.952] (II) EXA(0): Driver registered support for the following
operations:
[   109.952] (II)         Solid
[   109.952] (II)         Copy
[   109.952] (II)         Composite (RENDER acceleration)
[   109.952] (II)         UploadToScreen
[   109.952] (II)         DownloadFromScreen
[   109.952] (EE) 
Fatal server error:
[   109.952] (EE) failed to map shader -75
[   109.952] (EE) 
[   109.952] (EE) 
Please consult the The X.Org Foundation support 
         at http://wiki.x.org
 for help. 
[   109.952] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[   109.952] (EE) 
[   109.971] (EE) Server terminated with error (1). Closing log file.

Everything works normally with a 4.16 kernel.

--Bob

Re: [BUG] 4.17-rcX delays during boot and memory mapping errors

[Bob Tracy]

On Fri, May 18, 2018 at 07:32:59PM -0500, Bob Tracy wrote:
> Every one of the 4.17-rcX series has hung during boot at the point where
> "acpid" starts up.  Beginning with -rc5, the boot actually proceeds to
> completion after an "uncomfortably" long delay.
> 
> Attempting to run the X11 server results in the following diagnostic
> output in "/var/log/Xorg.0.log":

v4.17-rc6 took care of the X11 server startup issue.  The
"uncomfortably" long delay issue persists.  An observation as far as
differences between 4.16 and 4.17-rc6 during startup...

In v4.16 and earlier kernels, I wouldn't see the "random: crng init done"
message until moments after the system was done transitioning to multi-
user mode ("login" prompt on the console).  In the v4.17-rcX series, I
don't see a login prompt until all the events represented in the "dmesg"
output below have occurred:

(...)
[    7.314001] NET: Registered protocol family 10
[    7.315781] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[    7.317359] Segment Routing with IPv6
[    8.667444] NET: Registered protocol family 17
[    9.130872] e1000e: eth0 NIC Link is Up 10 Mbps Half Duplex, Flow
Control: None
[    9.131839] e1000e 0000:00:19.0 eth0: 10/100 speed: disabling TSO
[    9.136947] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[   78.567150] random: crng init done
[  102.974609] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[  102.982063] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
[  102.983716] r8152 2-1.5:1.0 eth1: carrier on
(end of dmesg output)

As far as what I see on the console during bootup, I see the
"IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready" message, then
the system "hangs" until the "random: crng init done" message appears.
Then "acpid" startup and other events proceed normally.

Any idea what's going on here?

--Bob