Charles Forsyth | 3 Mar 01:47 2009
Picon

Re: linux distribution

>I didnt know makemk.sh is the one i should use (lost touch, i guess).

you should not need to use that. it's only needed if you remove the
existing copies of mk, iyacc, etc. in Linux/386/bin, which are included
in the svn copy.

>but i noticed some trouble with X11, etc.

ah. you probably have not installed various `developer' packages for X11 for Ubuntu.
not Inferno-related, but Debian packages that anyone needs on Debian/Ubuntu
to compile X11 applications.  ie, packages you select and install through
apt-get  or synaptic.

Alex Efros | 3 Mar 22:01 2009

afd for sys->mount(2)

Hi!

mount(2) says:
 mount: fn(fd: ref FD; afd: ref FD, old: string, flag: int, aname: string): int;
 ...
 If the requested service requires authentication, the file descriptor afd
 must be open on an authentication file for the requested service;
 otherwise it should be nil.

But looks like mount(1) always set afd=nil, even when authorization used.
And I don't found any other app which use afd param.

How it was supposed to be used?
Why it isn't used now?
Maybe man page should be updated to explain current situation with
authorization in mount()?

--

-- 
			WBR, Alex.

Tharaneedharan Vilwanathan | 3 Mar 22:38 2009
Picon

Re: linux distribution

hi charles,

thanks for the response.

actually i sent several messages the other day but they had trouble
getting posted. let me resend them.

thanks
dharani

On Mon, Mar 2, 2009 at 4:47 PM, Charles Forsyth <forsyth@...> wrote:
>>I didnt know makemk.sh is the one i should use (lost touch, i guess).
>
> you should not need to use that. it's only needed if you remove the
> existing copies of mk, iyacc, etc. in Linux/386/bin, which are included
> in the svn copy.
>
>>but i noticed some trouble with X11, etc.
>
> ah. you probably have not installed various `developer' packages for X11 for Ubuntu.
> not Inferno-related, but Debian packages that anyone needs on Debian/Ubuntu
> to compile X11 applications.  ie, packages you select and install through
> apt-get  or synaptic.
>

Tharaneedharan Vilwanathan | 3 Mar 22:40 2009
Picon

Fwd: linux distribution

resending...

---------- Forwarded message ----------
From: Tharaneedharan Vilwanathan <vdharani@...>
Date: Thu, Feb 26, 2009 at 4:34 PM
Subject: Re: [inferno-list] linux distribution
To: inferno-list@...

hi all,

here is the update.

looks like it needed the package libx11-dev. once I installed it,
things were okay.

next, my machine has Core 2 Duo, so I was wondering if i should try
64-bit ubuntu. ubuntu came up fine but inferno didnt build there. i
didnt dig into the details but i thought there may be a 32-bit mode
which will take care of the disparity that inferno needs a 32-bit
architecture.

i have seen an error with graphics mode at least twice. i start emu as
given below and then i click on Start button and then the graphics
screen dies.

----------------------------------------------------------------------------------
dharani <at> dharani:~/inf090226_1130am$ emu
-r/home/dharani/inf090226_1130am -g1500x1100 -pmain=512000000
-pheap=512000000 -pimage=512000000
; ndb/cs
(Continue reading)

Brian L. Stuart | 3 Mar 23:07 2009
Picon

Re: Fwd: linux distribution

> i have seen an error with graphics mode at least twice. i start emu as
> given below and then i click on Start button and then the graphics
> screen dies.
> 
> ; wm/wm
> tk: font not found: /fonts/pelm/unicode.8.font
> emu: xcb_out.c:301: _xcb_out_flush_to: Assertion `((int)
> ((c->out.request_written) - (request)) >= 0)' failed.
> EMU: faults: Segmentation violation

I've seen this sort of thing on FreeBSD as wel, but without
tk complaining about a font.  In my case it's been when I
start acme.  It started after an upgrade to the latest xorg.
So it looks like they changed something that breaks Inferno,
but I haven't had a chance to try to track it down.

BLS

Tharaneedharan Vilwanathan | 3 Mar 22:41 2009
Picon

Fwd: linux distribution

resending...

---------- Forwarded message ----------
From: Tharaneedharan Vilwanathan <vdharani@...>
Date: Thu, Feb 26, 2009 at 6:26 PM
Subject: Re: [inferno-list] linux distribution
To: inferno-list@...

hi,

i copied fonts from vitanuova's inferno release and then it seemed to
be better but i just noticed an issue: this time it is a graphics
screen freeze and i see this message:

dharani <at> dharani:~/inf090226_1130am$ emu -r. -g1500x1100
; ndb/cs
; wm/wm
emu: ../../src/xcb_lock.c:77: _XGetXCBBuffer: Assertion `((int)
((xcb_req) - (dpy->request)) >= 0)' failed.

any clues? appreciate any help.

thanks
dharani

On Thu, Feb 26, 2009 at 6:00 PM, Tharaneedharan Vilwanathan
<vdharani@...> wrote:
> hi,
>
> irrespective of screen size, graphics screen crashes with this message
(Continue reading)

C H Forsyth | 4 Mar 16:41 2009

Re: afd for sys->mount(2)

mount will use the afd if the -9 option is given,
causing its use of Sys->mount via Factotum->mount to
have a non-nil afd, the result of a call to sys->fauth.
From: Alex Efros <powerman@...>
Subject: [inferno-list] afd for sys->mount(2)
Date: 2009-03-03 21:01:02 GMT
Hi!

mount(2) says:
 mount: fn(fd: ref FD; afd: ref FD, old: string, flag: int, aname: string): int;
 ...
 If the requested service requires authentication, the file descriptor afd
 must be open on an authentication file for the requested service;
 otherwise it should be nil.

But looks like mount(1) always set afd=nil, even when authorization used.
And I don't found any other app which use afd param.

How it was supposed to be used?
Why it isn't used now?
Maybe man page should be updated to explain current situation with
authorization in mount()?

--

-- 
			WBR, Alex.
(Continue reading)

Tharaneedharan Vilwanathan | 3 Mar 22:41 2009
Picon

Fwd: linux distribution

resending

---------- Forwarded message ----------
From: Tharaneedharan Vilwanathan <vdharani@...>
Date: Thu, Feb 26, 2009 at 6:00 PM
Subject: Re: [inferno-list] linux distribution
To: inferno-list@...

hi,

irrespective of screen size, graphics screen crashes with this message
either during start or once i start acme or so:

; ndb/cs
;wm/wm
tk: font not found: /fonts/pelm/unicode.8.font
emu: xcb_out.c:301: _xcb_out_flush_to: Assertion `((int)
((c->out.request_written) - (request)) >= 0)' failed.

any clues? currently, inferno is almost unusable.

thanks
dharani

On Thu, Feb 26, 2009 at 4:34 PM, Tharaneedharan Vilwanathan
<vdharani@...> wrote:
> hi all,
>
> here is the update.
>
(Continue reading)

bumbu dorin | 4 Mar 17:44 2009
Picon

Re: Fwd: linux distribution

This happened to me too on Linux: on Debian 5, Slackware 12.1,
OpenSuse 11. It seems that new Linux distributions tries to replace
Xlib with XCB (http://xcb.freedesktop.org/). On Debian 4 Inferno
graphics works fine. I think the Inferno graphics code should be
adapted to XCB.

On Tue, Mar 3, 2009 at 10:07 PM, Brian L. Stuart <blstuart@...> wrote:
>> i have seen an error with graphics mode at least twice. i start emu as
>> given below and then i click on Start button and then the graphics
>> screen dies.
>>
>> ; wm/wm
>> tk: font not found: /fonts/pelm/unicode.8.font
>> emu: xcb_out.c:301: _xcb_out_flush_to: Assertion `((int)
>> ((c->out.request_written) - (request)) >= 0)' failed.
>> EMU: faults: Segmentation violation
>
> I've seen this sort of thing on FreeBSD as wel, but without
> tk complaining about a font.  In my case it's been when I
> start acme.  It started after an upgrade to the latest xorg.
> So it looks like they changed something that breaks Inferno,
> but I haven't had a chance to try to track it down.
>
> BLS
>
>

Russ Cox | 4 Mar 18:05 2009

Re: Fwd: linux distribution

On Wed, Mar 4, 2009 at 8:44 AM, bumbu dorin <bumbudorin@...> wrote:
> This happened to me too on Linux: on Debian 5, Slackware 12.1,
> OpenSuse 11. It seems that new Linux distributions tries to replace
> Xlib with XCB (http://xcb.freedesktop.org/). On Debian 4 Inferno
> graphics works fine. I think the Inferno graphics code should be
> adapted to XCB.

It's not so much a matter of adapting the code to xcb
as just making code using the old interface make sure
that each Display* is only used in its own thread.
I did this for plan9port and 9vx.  It required a little care
but was not hard in the end.  The old, buggy code could
have crashed the old xlib too, it just would have been
a little harder.

There's not much point in moving wholesale to xcb
when the xlib version would still have to be maintained
for non-xcb systems and the xlib interface suffices
for this purpose.

Russ


Gmane