svn commit: r346687 - head/sys/compat/linuxkpi/common/src

[Tycho Nightingale]

> On Apr 25, 2019, at 2:50 PM, John Baldwin <jhb at FreeBSD.org> wrote:
> 
> On 4/25/19 11:43 AM, Konstantin Belousov wrote:
>> On Thu, Apr 25, 2019 at 06:13:56PM +0000, Tycho Nightingale wrote:
>>> Author: tychon
>>> Date: Thu Apr 25 18:13:55 2019
>>> New Revision: 346687
>>> URL: https://svnweb.freebsd.org/changeset/base/346687
>>> 
>>> Log:
>>>  LinuxKPI buildfix for ppc64 after r346645.
>>> 
>>>  Proposed by:	hselasky
>>>  Sponsored by:	Dell EMC Isilon
>>> 
>>> Modified:
>>>  head/sys/compat/linuxkpi/common/src/linux_pci.c
>>> 
>>> Modified: head/sys/compat/linuxkpi/common/src/linux_pci.c
>>> ==============================================================================
>>> --- head/sys/compat/linuxkpi/common/src/linux_pci.c	Thu Apr 25 17:28:36 2019	(r346686)
>>> +++ head/sys/compat/linuxkpi/common/src/linux_pci.c	Thu Apr 25 18:13:55 2019	(r346687)
>>> @@ -137,9 +137,9 @@ linux_dma_tag_init(struct device *dev, u64 dma_mask)
>>> 	    dma_mask,			/* lowaddr */
>>> 	    BUS_SPACE_MAXADDR,		/* highaddr */
>>> 	    NULL, NULL,			/* filtfunc, filtfuncarg */
>>> -	    BUS_SPACE_MAXADDR,		/* maxsize */
>>> +	    BUS_SPACE_MAXSIZE,		/* maxsize */
>>> 	    1,				/* nsegments */
>>> -	    BUS_SPACE_MAXADDR,		/* maxsegsz */
>>> +	    BUS_SPACE_MAXSIZE,		/* maxsegsz */
>>> 	    0,				/* flags */
>>> 	    NULL, NULL,			/* lockfunc, lockfuncarg */
>>> 	    &priv->dmat);
>> 
>> It seems that amd64 BUS_SPACE_MAXSIZE is 4G, I do not know why.
>> Either we should fix that, or the following fix is more appropriate.
>> i386 and ppc both have PAE-like configs where maxaddr is 64bit but
>> maxsize is 32bit.
> 
> I think we should fix amd64's maxsize.

That sounds reasonable.  Perhaps this was a conservative choice along with the 4GB DMA boundary which I removed recently.

After some testing I can make this change.

Tycho