Re: Update Linux 2.4 Status/TODO list

From: David Ford (david@kalifornia.com)
Date: Wed Sep 13 2000 - 18:23:25 EST


"Theodore Y. Ts'o" wrote:

> Date: Tue, 12 Sep 2000 23:55:55 -0700
> From: David Ford <david@kalifornia.com>
>
> Please add 'APM resume returns the machine to the first tty, crashes
> X' This appeared w/ test8. If this is intended, I'd be very happy to
> know if so and I can write in to xfree86 about it. If not
> intended..fix needed.
>
> Can you send more information? Hardware, etc. Do we know whether this
> is a hardware specific problem, or a more general problem?

(To several people)

Yes, I will. Until later this evening I need my laptop for work. This evening
I'll grab the serial cable and start making kernels and modules.

AFAIK, this is a kernel issue, going backwards to a kernel that doesn't switch
back to the first tty on resume, X remains in control of the screen and is quite
happy. The first kernel that resumes and jumps back to the first tty, X crashes.
Doing a gdb on it yields a writev() that results in a segfault. I'll do these two
problems individually.

> > 10. To Do But Non Showstopper
> > * PCMCIA/Cardbus hangs (Basically unusable - Hinds pcmcia code is
> > reliable)
> > + PCMCIA crashes on unloading pci_socket
>
> With test8, pcmcia either with kernel code, mixed, or dhinds code;
> nothing is usable.
>
> Can you send more details? I don't have a scratch laptop to try 2.4.0,
> and I'm too chicken to put it there. (My laptop has to work; it's a
> production machine; failure is not an option. Among other things, the
> master 2.4.0 bug list lives there. :-)
>
> > * cdrecord doesn't work (produces CD-ROM coasters) w/o any errors
> > reported, works under 2.2 (Damon LoCascio)
>
> Hmm, upgrade cdrecord perhaps? I'm using the second to last release and I'm
> not burning coasters. cdrecord is undergoing a scsi subsystem rewrite.
>
> According to Damon LoCascio, he's using the latest cdrecord.
>
> Kernel : 2.4.0-test5
> SCSI : Advansys APB940U Quad scsi controller
> CDRW : Yamaha 8824CDRW
> CPU : AMD K6-III @450 MHz
> Mem : 256Mb
> MB : FIC 503+
> CDrecord : 1.6, 1.81, 1.9, 1.10
>
> He says that it works just fine under 2.2.16, and he's seeing subtle
> data corruption problems under 2.4 (sometimes only 100 bytes between
> binary files). It could be a hardware problem where the 2.4 kernel is
> stressing things more than the 2.2 kernel; or it could be a SCSI
> controller specific problem. It's not clear.
>
> > Fixed
> > * Keyboard/mouse problems (should be fixed?)
>
> Related to pcmcia, no, not fixed. Certain mixes of drivers and hardware can
> stick the interrupt until they're removed and a different card inserted.
>
> Can you give more information on this one? What specifically is going
> on? There's no name associated with it, which means it was entered
> while Alan was maintaining the list, so I have no history associated
> with it. (For that matter, it may be referring to different problem
> from the one you're seeing now.)
>
> Thanks for the update, and your comments. (The rest of your comments
> will be on the linux24.sourceforge.net web page shortly.)
>
> - Ted

Certainly. I amonst others reported the key/mouse issue many kernels back and it
was a guess game for a while then Linus said the yenta was a red herring. I don't
have that email anymore but there's a lot of detail to the key/mouse/pcmcia story
and I'll dig it all up and write it out again.

Basically I believe the kernel pcmcia code is off by one in the socket numbering
as dhinds socket numbering works and w/ the kernel, one socket hangs the machine
and the other socket appears to be numbered as #0 where w/ dhinds it was #1. It
seems to wedge something physically because my laptop will forever hang at maestro
init after trying to use the top socket w/ the kernel pcmcia. I have to remove
all power then it will boot up past the maestro init. Note, the machine isn't
hard hung at maestro, the magic key works. It just won't proceed through the
boot.

More to come this evening.

-d

--
"The difference between 'involvement' and 'commitment' is like an
eggs-and-ham breakfast: the chicken was 'involved' - the pig was
'committed'."


- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Fri Sep 15 2000 - 21:00:22 EST