[haiku-development] Re: any idea about gcc4 build firewire's instability(ticket 1727)
- From: "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Tue, 27 May 2008 13:11:42 +0200 CEST
"JiSheng Zhang" <jszhang3@xxxxxxxxx> wrote:
> http://dev.haiku-os.org/ticket/1727
> I can reproduce this bug with one via chip 1394 card. I checked again
> and again the firewire code, but couldn't find which cause the
> instability. Someone mentioned there's no such problem with gcc2
> build. IIRC, gcc4 build last year is also ok. Any idea about this
> bug.
> This bug reminds me haiku's rtl8139 driver's instability, IMHO, there
> are common issues between them.
I don't know how old your hardware is, but we seem to have some
problems on more recent hardware which I would attribute to general
interrupt problems. One thing that is obviously wrong is what
interrupts are edge vs. level triggered - at least what the old PIC
method returns on current hardware. The outcome is that we might miss
or not correctly acknowledge an interrupt.
When I'm done with the device manager, I wanted to have a look at
implementing an APIC backend for the interrupts, but if someone wants
to have a look at that before, I definitely wouldn't mind :-)
Bye,
Axel.
Other related posts:
