Re: keyboard loss in DF

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

    A bunch of keyboard work was occuring through those dates.

    I think you've got me well on the way to finding the problem.

    I am also seeing keyboard loss on my workstation.. I don't lose it 
    entirely, but keystrokes are being dropped.  I think a critical section
    is being held accidently somewhere.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>


:   I had just about enough time before running out the door to work this
:morning to try a couple of things. Here is what I've done so far.
:
:I had the dfly-stable-20041009.iso.gz file on hand so I went ahead and
:installed it instead of the dfly-20041107-stable.iso.gz. Unable to repeat
:the keyboard loss (non-debug kernel).
:
:I had downloaded the two following kernels:
:
: 2CSNAP-20041001-2330-GCC2-GENERIC-kernel.debug.gz
: 2CSNAP-20041020-2330-GCC2-GENERIC-kernel.debug.gz
:
:When I boot from either of these, the problem is easily reproduced.
:
:20041001:
:DragonFly workstation.zip 1.1-CURRENT DragonFly 1.1-CURRENT #0: Sat Oct  2
:00:19:37 GMT 2004    
:root@xxxxxxxxxxxxxxxxxxxxxx:/usr/obj/usr/src/sys/GENERIC  i386
:
:20041009 (non-debug):
:DragonFly workstation.zip 1.1-CURRENT DragonFly 1.1-CURRENT #0: Sat Oct  9
:01:47:25 PDT 2004    
:dillon@xxxxxxxxxxxxxxxxxxxxx:/usr/obj/usr/src/sys/GENERIC  i386
:
:20041020:
:DragonFly workstation.zip 1.1-CURRENT DragonFly 1.1-CURRENT #0: Thu Oct 21
:00:17:13 GMT 2004    
:root@xxxxxxxxxxxxxxxxxxxxxx:/usr/obj/usr/src/sys/GENERIC  i386
:
:The two 2CSNAP kernels are debug kernels and the 20041009 is not. Maybe the
:problem manifests with a debug kernel only. At this stage I can either put
:the 20041107 iso build back on, cvsup and try building a non-debug kernel
:or look for a repository of non-debug kernels. This morning my time was
:limited so I only looked at:
:
:ftp://ftp.theshell.com/pub/DragonFly/snapshots/i386/kernels/
:
:If there is a non-debug kernel repository where I could get a non-debug
:version of the 20041107 kernel it would be quicker than downloading all the
:source from cvs, as at present I currently have no source code and I'd have
:to do so by controlling the box from remote (telnet/ssh). 
:
:If I am unable to reproduce with a newer non-debug it might focus attention
:on what is different in the debug kernels that is causing the problem.
:
:-Mike