[9fans] Parallels mysteriously loosing /dev/draw

[9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

"You are at Witt's End ..."

Well, I am.  Parallels 8.0.18608.  Mac OS 10.8.5.  Plan 9 from roughly August 30.  9pcf kernel running flawlessly in a 2560x1440x32 full screen window, for weeks.

Two days ago, I boot the VM and it (9pcf) decides there is no longer a frame buffer.  It worked fine the day before.  I haven't touched any of the Plan9 binaries or kernels, nor have I (knowingly) updated Parallels.  I did do the Mac OS 10.8.4 -> 10.8.5 update recently, but I am positive this worked after the OS upgrade.

I have had this happen before with Parallels, and have never been able to discover the underlying cause.  And it is driving me completely fscking insane.  Someone please tell me I am not the only one being inflicted with this.  Or better yet, come exorcise the bad bloody hardware karma I am suffering!  I don't deserve this ...

--possessed

Re: [9fans] Parallels mysteriously loosing /dev/draw

[erik quanstrom]

On Mon Oct 14 20:58:07 EDT 2013, lyndon@orthanc.ca wrote:
> "You are at Witt's End ..."
>
> Well, I am.  Parallels 8.0.18608.  Mac OS 10.8.5.  Plan 9 from roughly
> August 30.  9pcf kernel running flawlessly in a 2560x1440x32 full
> screen window, for weeks.
>
> Two days ago, I boot the VM and it (9pcf) decides there is no longer a
> frame buffer.  It worked fine the day before.  I haven't touched any
> of the Plan9 binaries or kernels, nor have I (knowingly) updated
> Parallels.  I did do the Mac OS 10.8.4 -> 10.8.5 update recently, but
> I am positive this worked after the OS upgrade.
>
> I have had this happen before with Parallels, and have never been able
> to discover the underlying cause.  And it is driving me completely
> fscking insane.  Someone please tell me I am not the only one being
> inflicted with this.  Or better yet, come exorcise the bad bloody
> hardware karma I am suffering!  I don't deserve this ...

does pci output show a vga device?

- erik

Re: [9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

> does pci output show a vga device?

Yes, and aux/vga shows all the VGA BIOS entries I expect to see.

Re: [9fans] Parallels mysteriously loosing /dev/draw

[erik quanstrom]

On Mon Oct 14 22:36:25 EDT 2013, lyndon@orthanc.ca wrote:
> > does pci output show a vga device?
>
> Yes, and aux/vga shows all the VGA BIOS entries I expect to see.

does "aux/vga -l $vgasize" do what you expect?

- erik

Re: [9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

On 2013-10-14, at 7:38 PM, erik quanstrom <quanstro@quanstro.net> wrote:

> does "aux/vga -l $vgasize" do what you expect?

No, it actually kicks the display into graphics framebuffer node.  (That's *not* what I expected.)  But this turns up a new problem: no data from the mouse.  /dev/mouse is there, but it returns no data.  I can start rio at this point, but without the mouse it doesn't do a whole lot ...

I will continue to tinker ...

Re: [9fans] Parallels mysteriously loosing /dev/draw

[erik quanstrom]

> > does "aux/vga -l $vgasize" do what you expect?
> >
> No, it actually kicks the display into graphics framebuffer node.
> (That's *not* what I expected.) But this turns up a new problem: no
> data from the mouse.  /dev/mouse is there, but it returns no data.  I
> can start rio at this point, but without the mouse it doesn't do a
> whole lot ...

due to the second law of thermohorification, fixing one thing means that at
least one other thing goes broken.  perhaps your mouse:parallels connection
kept the universe's hork in balance.

- erik

Re: [9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

On 2013-10-15, at 1:36 PM, erik quanstrom <quanstro@labs.coraid.com> wrote:

> due to the second law of thermohorification, fixing one thing means that at
> least one other thing goes broken.  perhaps your mouse:parallels connection
> kept the universe's hork in balance.

Whatever it is, it's very bizarre.

The aux/vga in termrc isn't running.  The one I added to $home/lib/profile is.  (And why isn't 'flag +x' added to /rc/bin/termrc barfing out the expected trace data? The 'echo kill -roy' immediately below it fires off.)

A second look at the pci output shows that nothing behind the USB bridges is being detected.  This explains the absence of a mouse.  But changing the plan9.ini mouse setting from intellimouse to ps2 solves that for now.

If we were a couple of weeks closer to Hallow'een, I would buy in to this.  As it is, it's just spooooky ...

Re: [9fans] Parallels mysteriously loosing /dev/draw

[erik quanstrom]

> The aux/vga in termrc isn't running.  The one I added to
> $home/lib/profile is.  (And why isn't 'flag +x' added to
> /rc/bin/termrc barfing out the expected trace data?  The 'echo kill
> -roy' immediately below it fires off.)

at least on my machine, the simplified relevant bit looks like

	if(test -f /dev/mousectl && ~ $mouseport ps2 ps2intellimouse 0 1 2 usb){
		if(! ~ $"monitor '' && ! ~ `{cat /dev/user} none)@{
			run vga from in here
		}

so if yours is similar, which one of the follwing is false?

1.	test -f /dev/mousectl
2	~ $mouseport ps2 ps2intellimouse 0 1 2 usb
3.	! ~ $"monitor ''
4.	! ~ `{cat /dev/user} none


could /dev/user be none, while $user = your user?

- erik

Re: [9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

On 2013-10-15, at 6:33 PM, erik quanstrom <quanstro@quanstro.net> wrote:

> 1.	test -f /dev/mousectl
> 2	~ $mouseport ps2 ps2intellimouse 0 1 2 usb
> 3.	! ~ $"monitor ''
> 4.	! ~ `{cat /dev/user} none
> 
> 
> could /dev/user be none, while $user = your user?

If the 'flag +x' worked as documented, it would be easy to tell :-p

I will printf my way through it tomorrow.  Tonight I am going ignore it and burn a steak ...

Re: [9fans] Parallels mysteriously loosing /dev/draw

[erik quanstrom]

> If the 'flag +x' worked as documented, it would be easy to tell :-p

the syntax is "flag x +".

- erik

Re: [9fans] Parallels mysteriously loosing /dev/draw

[Lyndon Nerenberg]

On 2013-10-15, at 6:39 PM, erik quanstrom <quanstro@quanstro.net> wrote:

> the syntax is "flag x +".

Oh gawd I have been subsumed by the Bourne Supremacy ...