Messages in this thread | | | Date | Sat, 11 Mar 2000 08:37:58 -0800 (PST) | From | Linus Torvalds <> | Subject | Re: 2.3.51 cardbus |
| |
On Sat, 11 Mar 2000, Wakko Warner wrote: > > The last stable 2.3 kernel I could use was around 2.3.3x, since around > 2.3.39 I guess is when 2.3 became quite unstable on this nec versa sx > laptop. I have attached the oops the I get when rmmod'ing pci_socket.
Please check if it works fine for you if you compile everything into the kernel - what I'd like to know is just whether the _basic_ functionality is fine, and that it is only the fact that the new cardbus code really doesn't work as a module.
I don't use modules. As a result, I haven't been all that interested in making them work - at worst I will just not allow the "m"odule answer in configure as nobody seems to be interested in fixing this.
> By the way, the 2 irq's yenta gets for the 2 sockets are 5 and 10. LSPCI > reports irq 5 being used by the sound controller (which causes problems when > playing sound and using a cardbus device on that irq), and 10 is reported as > being used by the vga controller. What was wrong with the way pcmcia was > done in 2.2 (with David's pcmcia modules) by allocating IRQs as they were > needed?
PCCARD really has two differentkinds of interrupts: the PCI interrupts and the legacy ISA interrupts. The legacy ISA interrupts are dynamic, the PCI interrupts are wired.
The new driver will use both. What you see as the "yenta irq" is the PCI interrupt routing information. It was always there, but the original PCMCIA driver didn't use it by default (aside: it DOES use it by default in newer versions, you are comparing a new kernel against an old PCMCIA package ;)
CardBus does need PCI interrupts. That's how it is spec'ed, and that's how it needs to be done at least for true cardbus cards (as opposed to the old 16-bit cards). Using the legacy ISA interrupts is a fallback that does not work on a number of machines. My Sony VAIO, for example, never could handle a regular tulip cardbus card with the old setup - which is kind of scary as that's probably the most common new cardbus network card out there.
There seems to be many bad (or just one, but common) sound drivers that are unhappy with sharing an interrupt. That is a sounddriver bug, nothing less, nothing more. PCI interrupts are fundamentally shared, and they get shared depending on things like which slot the cards are in etc etc.
Is this the same sound driver (ESS Maestro) or a new one?
Linus
- 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.tux.org/lkml/
| |