Re: [Qemu-devel] [PATCH qemu] fdc: Exit if ISA controller does not support DMA

[Philippe Mathieu-Daudé]

On 03/05/2018 07:29 AM, Philippe Mathieu-Daudé wrote:
> On 10/26/2017 03:46 AM, Alexey Kardashevskiy wrote:
>> A "powernv" machine type defines an ISA bus but it does not add any
>> DMA
>> controller to it so it is possible to hit assert(fdctrl->dma) by
>> adding "-machine powernv -device isa-fdc".
> 
> The same happens with the Alpha machine.
> 
> On 12/13/2017 03:19 AM, Hervé Poussineau wrote:
>> Le 08/12/2017 à 22:29, John Snow a écrit :
>> [...]
>>>
>>> It looks like Herve was working on decoupling floppies from i8257, but
>>> perhaps didn't get all the way through -- I'm not actually clear on what
>>> work remains to be done here, maybe he can chime in if he's still
>>> interested in the project?
>>>
>>
>> Indeed, I worked on decoupling floppies (and other ISA devices) from
>> i8257, for three reasons:
>> 1) having a working floppy on MIPS Magnum machine. fdc is PC-compatible,
>> but DMA controller is not i8257-compatible.
>> 2) reimplement another ISA bus with a somewhat i8257-compatible DMA
>> 3) for fun, support multiple ISA buses on a same machine but on
>> different PCI bridges.
>>
>> I failed on point 1), mostly due to lack of documentation.
>> I succeeded on point 2), by having locally some patches for a ISAPNP
>> bus, where IO/IRQ/DMA addresses of ISA devices can change dynamically,
>> and devices can be discovered by operating system.
>> About point 3, most of the patches are ready, but some details are still
>> to be fixed.
> 
> I'm having hard time refactoring all these PC chipsets, I'll make some
> room to send my current work as RFC before soft freeze.

Sorry I misunderstood the problem; my series only modify the Super I/O
devices, being affected by the same problem Alexey has with the ISA DMA :(