* blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs
@ 2017-01-09 20:48 Nikita Yushchenko
2017-01-09 20:59 ` Christoph Hellwig
2017-01-13 6:07 ` Ming Lei
0 siblings, 2 replies; 4+ messages in thread
From: Nikita Yushchenko @ 2017-01-09 20:48 UTC (permalink / raw)
To: Jens Axboe, linux-block; +Cc: linux-kernel, Artemi Ivanov
Hi
There is a use cases when architecture is 64-bit but hardware supports
only DMA to lower 4G of address space. E.g. NVMe device on RCar PCIe host.
For such cases, it looks proper to call blk_queue_bounce_limit() with
mask set to 0xffffffff - thus making block layer to use bounce buffers
for any addresses beyond 4G. To support that, architecture provides
GFP_DMA zone that covers exactly low 4G on arm64.
However setting this limit does not work:
if (b_pfn < (min_t(u64, 0xffffffffUL, BLK_BOUNCE_HIGH) >> PAGE_SHIFT))
dma = 1;
When mask is 0xffffffff that condition is false
q->limits.bounce_pfn = max(max_low_pfn, b_pfn);
this line is executed and replaces any limit with end of memory (on
64bit arch all memory is low).
Not sure how to fix this properly. Any hints?
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs
2017-01-09 20:48 blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs Nikita Yushchenko
@ 2017-01-09 20:59 ` Christoph Hellwig
2017-01-13 6:07 ` Ming Lei
1 sibling, 0 replies; 4+ messages in thread
From: Christoph Hellwig @ 2017-01-09 20:59 UTC (permalink / raw)
To: Nikita Yushchenko; +Cc: Jens Axboe, linux-block, linux-kernel, Artemi Ivanov
On Mon, Jan 09, 2017 at 11:48:11PM +0300, Nikita Yushchenko wrote:
> Hi
>
> There is a use cases when architecture is 64-bit but hardware supports
> only DMA to lower 4G of address space. E.g. NVMe device on RCar PCIe host.
The solution is to shoot the SOC designer. If that doesn't work use
swiotlb.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs
2017-01-09 20:48 blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs Nikita Yushchenko
2017-01-09 20:59 ` Christoph Hellwig
@ 2017-01-13 6:07 ` Ming Lei
2017-01-13 6:10 ` Nikita Yushchenko
1 sibling, 1 reply; 4+ messages in thread
From: Ming Lei @ 2017-01-13 6:07 UTC (permalink / raw)
To: Nikita Yushchenko; +Cc: Jens Axboe, linux-block, linux-kernel, Artemi Ivanov
Hi,
On Tue, Jan 10, 2017 at 4:48 AM, Nikita Yushchenko
<nikita.yoush@cogentembedded.com> wrote:
> Hi
>
> There is a use cases when architecture is 64-bit but hardware supports
> only DMA to lower 4G of address space. E.g. NVMe device on RCar PCIe host.
>
> For such cases, it looks proper to call blk_queue_bounce_limit() with
> mask set to 0xffffffff - thus making block layer to use bounce buffers
> for any addresses beyond 4G. To support that, architecture provides
> GFP_DMA zone that covers exactly low 4G on arm64.
>
> However setting this limit does not work:
>
> if (b_pfn < (min_t(u64, 0xffffffffUL, BLK_BOUNCE_HIGH) >> PAGE_SHIFT))
> dma = 1;
>
> When mask is 0xffffffff that condition is false
That should have been true in your case, since the b_pfn is smaller than
0xffffffff.
>
> q->limits.bounce_pfn = max(max_low_pfn, b_pfn);
>
> this line is executed and replaces any limit with end of memory (on
> 64bit arch all memory is low).
I don't understand why max() is used? And why not min()?
Looks the above line just disables bounce for 64bit arch, doesn't it?
Thanks,
Ming
>
>
> Not sure how to fix this properly. Any hints?
> --
> To unsubscribe from this list: send the line "unsubscribe linux-block" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Ming Lei
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs
2017-01-13 6:07 ` Ming Lei
@ 2017-01-13 6:10 ` Nikita Yushchenko
0 siblings, 0 replies; 4+ messages in thread
From: Nikita Yushchenko @ 2017-01-13 6:10 UTC (permalink / raw)
To: Ming Lei; +Cc: Jens Axboe, linux-block, linux-kernel, Artemi Ivanov
>> There is a use cases when architecture is 64-bit but hardware supports
>> only DMA to lower 4G of address space. E.g. NVMe device on RCar PCIe host.
>>
>> For such cases, it looks proper to call blk_queue_bounce_limit() with
>> mask set to 0xffffffff - thus making block layer to use bounce buffers
>> for any addresses beyond 4G. To support that, architecture provides
>> GFP_DMA zone that covers exactly low 4G on arm64.
>>
>> However setting this limit does not work:
>>
>> if (b_pfn < (min_t(u64, 0xffffffffUL, BLK_BOUNCE_HIGH) >> PAGE_SHIFT))
>> dma = 1;
>>
>> When mask is 0xffffffff that condition is false
>
> That should have been true in your case, since the b_pfn is smaller than
> 0xffffffff.
b_pfn is exactly 0xffffffffUL >> SHIFT, thus contition is false
>> q->limits.bounce_pfn = max(max_low_pfn, b_pfn);
>>
>> this line is executed and replaces any limit with end of memory (on
>> 64bit arch all memory is low).
>
> I don't understand why max() is used? And why not min()?
>
> Looks the above line just disables bounce for 64bit arch, doesn't it?
Effectively yes. And I don't understand logic behind this code.
Nikita
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2017-01-13 6:10 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-01-09 20:48 blk_queue_bounce_limit() broken for mask=0xffffffff on 64bit archs Nikita Yushchenko
2017-01-09 20:59 ` Christoph Hellwig
2017-01-13 6:07 ` Ming Lei
2017-01-13 6:10 ` Nikita Yushchenko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).