Am 25.02.2016 um 11:10 hat Alex Pyrgiotis geschrieben:
> > I think the first step is understanding what to do about the weird "&
> > ~BDRV_SECTOR_MASK" case, then.
>
> We can discuss about this case in the "dma-helpers: Do not truncate
> small qiovs" thread. I'm all for the removal of this check,
On 25/02/2016 12:19, Alex Pyrgiotis wrote:
>> The patch could first introduce address_space_map_direct that never uses
>> the bounce buffer. dma-helpers.c can call address_space_map_direct and,
>> if it fails, proceed to allocate (and fill if writing to the device) a
>> bounce buffer.
>
> You m
Hi Paolo,
On 02/25/2016 12:22 PM, Paolo Bonzini wrote:
>
>
> On 25/02/2016 11:10, Alex Pyrgiotis wrote:
>> 8<... snip ...>8
>>
>> Sure, you're right, there's no sensible way to break an ioctl()
>> operation in many. However, I'd argue that we shouldn't need to, as it
>> would be much better if t
On 25/02/2016 11:10, Alex Pyrgiotis wrote:
> All normal regions in a QEMUSGList point to an address range in the
> guest's RAM. The MMIO regions of QEMU's virtual devices, however, do not
> correspond to such an address range, so QEMU must create a bounce buffer
> to represent them. This bounce b
Hi Paolo,
Thanks a lot for your clarifications. See my comments inline:
(tl;dr: I suggest we reconsider Fam Zheng's attempt to remove the global
bounce buffer, which would make dma-helpers simpler and unblock this patch)
On 02/22/2016 12:43 PM, Paolo Bonzini wrote:
>
>
> On 19/02/2016 12:50, A
On 19/02/2016 12:50, Alex Pyrgiotis wrote:
> QEMU/Hardware space:
> 5. The SCSI controller code will create a QEMUSGList that points to
>the memory regions of the SCSI request. This QEMUSGList will also
>include the MMIO regions.
> 6. The QEMU device implementation, e.g. scsi-block, choos
Hi Paolo,
Sorry for the late reply and thanks for the review. See my comments inline:
On 02/11/2016 01:17 PM, Paolo Bonzini wrote:
>
>
> On 16/12/2015 17:55, Alex Pyrgiotis wrote:
>> +/*
>> + * Create a QEMUIOVector from a scatter-gather list.
>> + *
>> + * This function does not copy the data
On 16/12/2015 17:55, Alex Pyrgiotis wrote:
> +/*
> + * Create a QEMUIOVector from a scatter-gather list.
> + *
> + * This function does not copy the data of the scatter-gather list. Instead,
> it
> + * uses the dma_memory_map() function to map physical memory regions of the
> + * virtual device
The mapping of scatter-gather lists from physical addresses (as
perceived by the guest kernel) to the virtual address space of the QEMU
process is a vital step for a DMA operation. This step is currently
implemented, amongst other things, in dma_blk_cb(), making it impossible
to be used by anyone e
