Re: Linux-2.1.98..

F Harvell (spam@fts.net)
Fri, 24 Apr 1998 17:05:48 -0300


On Fri, 24 Apr 1998 12:20:11 PDT, Bill Broadhurst wrote:
> On Fri, Apr 24, 1998 at 10:34:20AM -0700, Linus Torvalds wrote:
> > Can you check what happens with:
> > (a) no floppy driver compiled in at all
>
> Boots. Runs. But hung hard during the "make dep" portion of the build.
> Not far into it either. Had to stop and clean drives.

I can verify that it booted this way on my system as well. I
immediately rebooted to check the second (b) case.

> > (b) the floppy driver compiled in, but with the calls to
> > "fd_enable_irq()" and "fd_disable_irq()" commented out.
>
> Boots and runs too. I'll let this one run 'till it dies.

This version also ran on my system.

Some further info: compiling the floppy driver as a module lets
the system boot, but, if there is a floppy access, the system will
hard lock. Commenting out the "fd_enable_irq()" and
"fd_disable_irq()" and compiling as a module boots and allows floppy
access.

> > The irq enable/disable code has some problems with the IO-APIC, and I'd
> > like to pinpoint whether this is the reason for your problems or not. We
> > should fix the IO-APIC problems too, of course, but that will take some
> > more doing.
>
> I'm all for that. What can I do to help?

I'm also ready and willing to help support this. I really want to
get these problems solved. I'm hoping that, when we solve this, the
problems with processes getting stuck in the "D" state will
disappear. Lately, I have begun to see update (bdflush) get stuck.
This effectively crashes my system.

>
> One more thing. I lost some of my earlier kernels due to a disk crash
> last night. In replacing them from source I discovered that .81, .82,
> .83, and .84 all die with "Aiee, interrupt in swapper task..."
> There was a minor work-around for this back then. Do you remember what
> it was. I'd like to go back through those kernels and verify that the
> load problem I saw started at .85. One of my production machines has
> locked twice in the middle of a model resolution. It was running .83
> at the time. I've dropped it back to the 2.0 series kernel since it's
> extra HW isn't in use right now. At any rate, I'd just like to run
> what ever the most stable of that pre-85 series. Anyone remember which
> was most solid of the .8x kernels?

I was using 2.1.84 as my "stable" kernel. The IO-APIC stuff turned
on in 2.1.85-94 really triggered the "D" state problems. The changes
in 2.1.95, though, allowed me to begin using that one as my "stable"
kernel. (Of course, as I type this, I am running on the 2.1.98
kernel. :)

-- 
Mr. F Harvell                          Phone:407 696-4340
FTS International, Systems Division    Phone:407 399-0342 (cell)
3498 Buffam Place                        Fax:407 696-4244
Casselberry, FL 32707                 mailto:fharvell@fts.net

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu