Re: 2.1.111: IDE DMA disabled...BLAH...BLAH...

Geert Uytterhoeven (Geert.Uytterhoeven@cs.kuleuven.ac.be)
Thu, 30 Jul 1998 10:40:24 +0200 (CEST)


Let's tell a short story...

I had mysterious IDE problems in my Amiga 4000 for years, on heavy disk
activity (busy, IDE reset, busy, ..., system lock up). Nothing seemed to help,
no one had a clue. Some kernels seemed to fix it, but some weeks later the
problem reappeared.

Important things to mention:

- 1 have 2 IDE disks on one interface. The slave is an old Conner (4.5 years
old), the master is a newer Quantum Fireball.
- Amigas use shared interrupts. E.g. my ethernet board hangs on the same
interrupt line.
- My '040 is a slow CPU, measured to current norms.

I assume the old Conner has some bugs.

But then, one small fix in 2.1[89]x (I think it was the moving of a cli or sti)
seem to have fixed my problem. Since then I haven't seen a single crash
related to IDE (in fact I haven't seen a single crash on that machine anymore).

Now the zillion dollar question: do I have bad hardware (my disk)? I don't
know. And I'll probably never will.

One of your remarks was that interrupts are serviced faster on SMP. That's
surely a valid argument. It's also true with shared interrupts: imagine an
ethernet packet arriving just before the IDE disk interrupt. When traveling
the interrupt service chain, the IDE driver may already see that the IDE disk
interrupted, while it would have seen that later if there had arrived no
ethernet packet.

Note: I'm not suggesting the IDE driver is to blame. Lot's of modern hardware
have more or less serious design flaws. Saying `Windows works fine on it' is no
excuse (cfr. the 3Com Boomerang PIO vs busmastering). Linus, you really
disappointed me when you used Win95 as a reference in one of your claims.

Greetings,

Geert (the m68k IDE guy)

--
Geert Uytterhoeven                     Geert.Uytterhoeven@cs.kuleuven.ac.be
Wavelets, Linux/{m68k~Amiga,PPC~CHRP}  http://www.cs.kuleuven.ac.be/~geert/
Department of Computer Science -- Katholieke Universiteit Leuven -- Belgium

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