Date: Sat, 19 Apr 1997 14:28:16 -0700 From: Amancio Hasty <hasty@rah.star-gate.com> To: Randall Hopper <rhh@ct.picker.com> Cc: multimedia@freebsd.org, David Dawes <dawes@rf900.physics.usyd.edu.au> Subject: Re: bt848/fxtv: More info on system freezes Message-ID: <199704192128.OAA03885@rah.star-gate.com> In-Reply-To: Your message of "Sat, 19 Apr 1997 14:13:14 EDT." <19970419141314.20369@ct.picker.com>
next in thread | previous in thread | raw e-mail | index | archive | help
We may have to file a bug report with the X server group to see if we can
get them to help us out. I will be honest with you every day I use
fxtv and I don't get a crash -- once I did get the font corruption and
the system didn't craash however I have not been able to reproduce it.
My take is that is probably more to do with the PCI chipset.
The "mistery-bit" on the Bt848 is supposed to be a work around for
the Natoma chipset. I still don't know what is supposed to do other
than what I stated. All I am saying is that is difficult to troubleshoot
on my PPRo . Guess that I will have to switch testing to my P100 which
does have a Natoma chipset.
You got a high praise on fxtv . Just showed the program to a friend of mine
(switch between full screen and normal, use the keypad to switch channel, etc..)
and my friend said : "Wow" .
Regards,
Amancio
P.S.: To David, we are doing high speed data transfer (PCI to PCI) from the
bt848 (video capture chip) to the S3's frame buffer.
>From The Desk Of Randall Hopper :
> I beat on the system freeze problem this morning, and whittled it down
> to a very simple procedure that always quickly reproduces the problem on my
> system.
>
> First, how to reproduce it, followed by my system configuration:
>
> HOW TO REPRODUCE
>
> 1) Start up X with no clients except a single stock xterm.
> 2) Start up fxtv 0.4 such that it doesn't overlap the xterm.
> fxtv is continuous displaying to the screen w/ direct video.
> 3) Now, start up a full-screen app (a text-mode app; runs in the xterm)
> 4) Lean on the key that causes the app to redraw its display
>
> After about 10-30 seconds, the xterm font gets corrupted. Shortly
> thereafter (milliseconds or seconds later), the X server locks up
> hard. Display dead, keyboard dead, mouse dead.
>
> Sometimes if I'm quick about it, I can get up off the "refresh" key in
> time for the queued refreshes to quit "after" the font is first
> corrupted but "before" the X server has locked. I can then
> Ctrl-Alt-Bksp to kill the server, and start it up again, giving me a
> sensible display with no traces of the corruption.
>
> Seems to be evidence that the X server state is being corrupted
> somehow, and not the system in general.
>
> Further evidence that this is typically localized to the X server and
> its state is that, if I have a kernel rebuild going on in an xterm and
> the the lock-up occurs, the kernel build will continue for 30-60 sec,
> presumably until the ptty device queue fills up because there's no
> process on the other end slirping off of it.
>
> Also of note is that after the font corruption starts, it continues to
> mutate some with each refresh. Line corruption through characters is
> the most common and stays fairly constant across refreshes, but there
> is some "white noise" (stray pixels) that appear and disappear in the
> font characters. Also, the corruption is in the colors that are used
> on the xterm at the time (purple, white, and blue). Seems to indicate
> that the font data is being corrupted before it's being blasted on the
> screen rather than after the fact by some corruption directly applied
> to the frame buffer.
>
> SYSTEM CONFIGURATION
>
> Motherboard: ASUS P55TP4XE (Triton 1), P100, 32Meg mem
> Video Card : STB Velocity 3D (S3988 "Virge", 4Meg VRAM)
>
> FreeBSD Ver: 2.2-GAMMA
> XServer Cfg: XFree 3.2A (S3V server)
> 1024x768 555 16bpp w/ 1152x900 desktop
>
> XTerm Cfg : stock xterm; 80x47 geometry w/ 9x15 font
>
> TV Software: Driver - bt848-970401 w/ Amancio's "missing fields" patch
> Client - FXTV 0.4
>
> BIOS Cfg : I did try Steve's suggestion of:
>
> PCI Streaming = OFF; Video BIOS Cache = OFF;
> VGA Palette Snoop = OFF
>
> All testing done/problems encountered this morning were
> with these settings.
>
>
> Randall
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704192128.OAA03885>