Re: [PATCH] ion_system_heap: support X86 archtecture

[Greg KH]

On Sun, Sep 29, 2019 at 03:28:41PM +0800, jun.zhang@xxxxxxxxx wrote:
> From: zhang jun <jun.zhang@xxxxxxxxx>
> 
> we see tons of warning like:
> [   45.846872] x86/PAT: NDK MediaCodec_:3753 map pfn RAM range req
> write-combining for [mem 0x1e7a80000-0x1e7a87fff], got write-back
> [   45.848827] x86/PAT: .vorbis.decoder:4088 map pfn RAM range req
> write-combining for [mem 0x1e7a58000-0x1e7a58fff], got write-back
> [   45.848875] x86/PAT: NDK MediaCodec_:3753 map pfn RAM range req
> write-combining for [mem 0x1e7a48000-0x1e7a4ffff], got write-back
> [   45.849403] x86/PAT: .vorbis.decoder:4088 map pfn RAM range
> req write-combining for [mem 0x1e7a70000-0x1e7a70fff], got write-back
> 
> check the kernel Documentation/x86/pat.txt, it says:
> A. Exporting pages to users with remap_pfn_range, io_remap_pfn_range,
> vm_insert_pfn
> Drivers wanting to export some pages to userspace do it by using
> mmap interface and a combination of
> 1) pgprot_noncached()
> 2) io_remap_pfn_range() or remap_pfn_range() or vm_insert_pfn()
> With PAT support, a new API pgprot_writecombine is being added.
> So, drivers can continue to use the above sequence, with either
> pgprot_noncached() or pgprot_writecombine() in step 1, followed by step 2.
> 
> In addition, step 2 internally tracks the region as UC or WC in
> memtype list in order to ensure no conflicting mapping.
> 
> Note that this set of APIs only works with IO (non RAM) regions.
> If driver ants to export a RAM region, it has to do set_memory_uc() or
> set_memory_wc() as step 0 above and also track the usage of those pages
> and use set_memory_wb() before the page is freed to free pool.
> 
> the fix follow the pat document, do set_memory_wc() as step 0 and
> use the set_memory_wb() before the page is freed.
> 
> Signed-off-by: he, bo <bo.he@xxxxxxxxx>
> Signed-off-by: zhang jun <jun.zhang@xxxxxxxxx>
> Signed-off-by: Bai, Jie A <jie.a.bai@xxxxxxxxx>
> ---
>  drivers/staging/android/ion/ion_system_heap.c | 28 ++++++++++++++++++-
>  1 file changed, 27 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/staging/android/ion/ion_system_heap.c b/drivers/staging/android/ion/ion_system_heap.c
> index b83a1d16bd89..d298b8194820 100644
> --- a/drivers/staging/android/ion/ion_system_heap.c
> +++ b/drivers/staging/android/ion/ion_system_heap.c
> @@ -13,6 +13,7 @@
>  #include <linux/scatterlist.h>
>  #include <linux/slab.h>
>  #include <linux/vmalloc.h>
> +#include <asm/set_memory.h>
>  
>  #include "ion.h"
>  
> @@ -134,6 +135,13 @@ static int ion_system_heap_allocate(struct ion_heap *heap,
>  	sg = table->sgl;
>  	list_for_each_entry_safe(page, tmp_page, &pages, lru) {
>  		sg_set_page(sg, page, page_size(page), 0);
> +
> +#ifdef CONFIG_X86
> +	if (!(buffer->flags & ION_FLAG_CACHED))
> +		set_memory_wc((unsigned long)page_address(sg_page(sg)),
> +			      PAGE_ALIGN(sg->length) >> PAGE_SHIFT);
> +#endif

There is no way to do this without these #ifdefs?  That feels odd, why
can't you just always test for this?

thanks,

greg k-h