On 02/06/14 02:14, tyler knosis wrote:
> Public bug reported:
>
> Trying to install NeXTSTEP 3.3 onto a 2GB file with QEMU 1.7.0.
> In the terminal that started QEMU, there are a lot of
> dma: command 10 not supported
> and
> dma: command 14 not supported
> messages.
These correspond to
CMD_CYCLIC_PRIORITY
and
CMD_CYCLIC_PRIORITY | CMD_BLOCK_CONTROLLER
in write_cont() [hw/dma/i8257.c]. They are not supported (see
CMD_NOT_SUPPORTED in the same file).
> When the installation of NeXTSTEP gets to
> 'preparing disk for nextstep installation', there are a lot
> of messages that ATA command c5 failed and other info.
> The result is a failed installation.
0xC5 is WIN_MULTWRITE ("write sectors using multiple mode"), and it
seems to hook into DMA.
> Is this a bug in QEMU?
Probably not, it's more likely "incomplete" DMA emulation ("lack of
support").
> Is there a workaround, e.g. by
> disabling DMA altogether?
It's worth a try I guess, if you can figure out a way how to do that.
FWIW, ide_identify() appears to report unconditional support for:
- single word dma0-2
- mdma0-2
- udma5
(I have no clue about IDE or DMA btw.)
Laszlo
