Re: [Xen-devel] [PATCH 2/2] ioemu: Enable guest OS to program D0-D3hot states of an assigned device

[Yuji Shimada <shimada-yxb@xxxxxxxxxxxxxxx>]

On Fri, 6 Feb 2009 11:45:29 +0900
Isaku Yamahata <yamahata@xxxxxxxxxxxxx> wrote:

> On Thu, Feb 05, 2009 at 07:23:11PM +0900, Yuji Shimada wrote:
> > We can clean up the existing code in pass-through.c in the same
> > manner. I will submit the patch to clean up the code.
> 
> Thank you for clean up.
> There are another logic duplications with mask operations.
> I suppose helper functions like the following will help
> to reduce the code more.
> 
> thanks,
> 
> static void __pt_read_{byte, word, long}(uint{8, 16, 32}_t *value,
>                                          uint{8, 16, 32}_t dev_value,
>                                          uint{8, 16, 32}_t emu_mask,
>                                          uint{8, 16, 32}_t valid_mask)
> {
>     uint{8, 16, 32}_t valid_emu_mask = emu_mask & valid_mask;
>     *value = ((*value & ~valid_emu_mask) | (dev_data & valid_emu_mask));
> }
> 
> static void __pt_write_cfg_{byte, word, long}(struct pt_reg_tbl *cfg_entry,
>                                               uint{8, 16, 32}_t value
>                                             uint{8, 16, 32}_t emu_mask,
>                                             uint{8, 16, 32}_t ro_mask,
>                                             uint{8, 16, 32}_t valid_mask)
> {
>     uint{8, 16, 32}_t writable_mask = emu_mask & ~ro_mask & valid_mask;
>     cfg_entry->data = ((value & writable_mask) |
>                        (cfg_entry->data & ~writable_mask));
> }
> 
> static void __pt_write_{byte, word, long}(uint{8, 16, 32}_t *value,
>                                         uint{8, 16, 32}_t dev_value,
>                                         uint{8, 16, 32}_t emu_mask,
>                                         uint{8, 16, 32}_t valid_mask)
> {
>     uint{8, 16, 32}_t throughable_mask = ~emu_mask & valid_mask;
>     *value = ((*value & throughable_mask) | (dev_value & ~throughable_mask));
> }

I realize the duplications with mask operations.
I have already introduced PT_MERGE_VALUE macro with the previous
patch. I'd like to create the new patch which cleanups them using
PT_MERGE_VALUE macro.
Is this acceptable?

These are examples.

Returned value to guest software on reading:
    *value = PT_MERGE_VALUE(*value, dev_value, emu_mask & valid_mask);

Emulated value on writing.
    cfg_entry->data = PT_MERGE_VALUE(value, cfg_entry->data,
        emu_mask & ~ro_mask & valid_mask)

Written value to actual device on writing:
*value = PT_MERGE_VALUE(*value, dev_value, ~emu_mask & valid_mask);

Thanks,
--
Yuji Shimada


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel