Re: Linux-1.3.71..
Aaron Ucko (UCKO@vax1.rockhurst.edu)
Tue, 12 Mar 1996 21:11:08 -0600 (CST)
>>Sorry Linus, but I am still getting console bleeds in 71. I was using tty2
>>for telnet over a slow ppp link and tty3 for an ftp over the same link.
>>tty4 and later on tty6, had some of the text from tty3 shifted to the left.
>>tty4 was not being used except by agetty and tty6 was idle. After that I
>>tried switching VCs around a lot, and tty2 and 3 got corrupted with text from
>>other VCs and the text wasn't shifted to the left.
>>
>>I have a VLB Cirrus 5426 based card, and the text resolution is 132x60.
>
>I also have a VLB 5426, and an odd text resolution (100x48). I just
>suffered my first crash (1.3.68) involving console corruption since,
>ooh, sometime in the 1.3.20s. I also just reinstalled SVGATextMode,
>which I haven't used since then.
>
>Coincidence? Or is there perhaps something in console.c and friends
>which objects to screens not being the same size as they were at
>bootup? (it can't practically be an 80x25 assumption, as I spent the
>intervening time using the BIOS 80x50 mode anyway). I used to be able
>to make some 1.3.x (10<=x<=30) kernels crash quite quickly using
>VT_RESIZE ioctls, I'll see if I can repeat this.
>
>1.3.68 was up for 18 days or so on this machine. It had one screen
>corruption of the `random coloured block' form the other day, and the
>crash involved what looked like repeated panic messages scrolling up
>th screen on some consoles and large colourful random things on
>others. No EIPs for you to look at (I can't read _that_ fast ...),
>and it rebooted spontaneously (I don't have the watchdog installed)
>about twenty seconds later. It was under reasonable load (one
>compilation, one autoconf) but I wasn't actually using the keyboard at
>the time.
>
>Of course I don't expect anyone to find a bug on the strength of this,
>but if other people are having problems too, maybe we can look for
>similarities. It'd be nice to think it wasn't a hardware problem.
I have a VLB Cirrus Logic GL5428, and have also been having problems,
so I suspect that there is something about this type of card that Linux
does not like. I've experienced several categories: random text, all
1's (flashing white umlauted y's on a light gray background), incorrect
screen offsets (sometimes only about the bottom half shows up, improperly
aligned, with garbage below it, and complete lockups. (I use jsrd in
an attempt to handle lockups, but it doesn't do anything in these cases;
I invariably have to hit reset.) I have observed specific "dangerous"
behaviors, including using SVGATextMode, switching too quickly between
graphics VCs and text VCs, and using zgv in 24-bit mode. SVGATextMode
is likeliest to cause Oopses and the like; too-fast switching results
in garbage; 24-bit zgv causes lockups. [I sometimes get garbage in
graphics mode too, most often in the form of a mangled palette.]
Oh, and I have also sometimes been in situations where everything's
normal except for the fact that the screen is totally blank in text mode
(X works fine, but exiting it again yields blankness, and SVGATextMode
doesn't help. I have to soft reboot and watch text come back on when
the BIOS reinitializes.)
All of this behavior occurs in both old and new kernels.
-- Aaron Ucko (ucko@vax1.rockhurst.edu; finger for PGP public key) | httyp!
"That's right," he said. "We're philosophers. We think, therefore we am."
-- Terry Pratchett, _Small Gods_ | Geek Code 3.1 [for explanation, finger
hayden@mankato.msus.edu]: GCS/M/S/C d- s: a18 C++(+++)>++++ UL++>++++ P++
L++>+++++ E- W(-) N++(+) o+ K- w--- O M@ V-(--) PS++(+++) PE- Y(+) PGP(+) t(+)
!5 X-- R(-) tv-@ b++(+++) DI+ !D-- G++(+++) e->+++++(*) h!>+ r-(--)>+++ y?