diff mbox series

[1/2] xen/arm64: do not set SWIOTLB_NO_FORCE when swiotlb is required

Message ID 20210511174142.12742-1-sstabellini@kernel.org (mailing list archive)
State New
Headers show
Series [1/2] xen/arm64: do not set SWIOTLB_NO_FORCE when swiotlb is required | expand

Commit Message

Stefano Stabellini May 11, 2021, 5:41 p.m. UTC
From: Stefano Stabellini <stefano.stabellini@xilinx.com>

Although SWIOTLB_NO_FORCE is meant to allow later calls to swiotlb_init,
today dma_direct_map_page returns error if SWIOTLB_NO_FORCE.

For now, without a larger overhaul of SWIOTLB_NO_FORCE, the best we can
do is to avoid setting SWIOTLB_NO_FORCE in mem_init when we know that it
is going to be required later (e.g. Xen requires it).

To make xen_swiotlb_detect available to !CONFIG_XEN builds, move it to a
static inline function.

CC: boris.ostrovsky@oracle.com
CC: jgross@suse.com
CC: catalin.marinas@arm.com
CC: will@kernel.org
CC: linux-arm-kernel@lists.infradead.org
Fixes: 2726bf3ff252 ("swiotlb: Make SWIOTLB_NO_FORCE perform no allocation")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Stefano Stabellini <stefano.stabellini@xilinx.com>
---
 arch/arm/xen/mm.c             | 12 ------------
 arch/arm64/mm/init.c          |  3 ++-
 include/xen/arm/swiotlb-xen.h | 15 ++++++++++++++-
 3 files changed, 16 insertions(+), 14 deletions(-)

Comments

Christoph Hellwig May 12, 2021, 7:26 a.m. UTC | #1
> -int xen_swiotlb_detect(void)
> -{
> -	if (!xen_domain())
> -		return 0;
> -	if (xen_feature(XENFEAT_direct_mapped))
> -		return 1;
> -	/* legacy case */
> -	if (!xen_feature(XENFEAT_not_direct_mapped) && xen_initial_domain())
> -		return 1;
> -	return 0;
> -}

I think this move should be a separate prep patch.
Stefano Stabellini May 12, 2021, 7:55 p.m. UTC | #2
On Wed, 12 May 2021, Christoph Hellwig wrote:
> > -int xen_swiotlb_detect(void)
> > -{
> > -	if (!xen_domain())
> > -		return 0;
> > -	if (xen_feature(XENFEAT_direct_mapped))
> > -		return 1;
> > -	/* legacy case */
> > -	if (!xen_feature(XENFEAT_not_direct_mapped) && xen_initial_domain())
> > -		return 1;
> > -	return 0;
> > -}
> 
> I think this move should be a separate prep patch.

Sure, I can do that
diff mbox series

Patch

diff --git a/arch/arm/xen/mm.c b/arch/arm/xen/mm.c
index f8f07469d259..223b1151fd7d 100644
--- a/arch/arm/xen/mm.c
+++ b/arch/arm/xen/mm.c
@@ -135,18 +135,6 @@  void xen_destroy_contiguous_region(phys_addr_t pstart, unsigned int order)
 	return;
 }
 
-int xen_swiotlb_detect(void)
-{
-	if (!xen_domain())
-		return 0;
-	if (xen_feature(XENFEAT_direct_mapped))
-		return 1;
-	/* legacy case */
-	if (!xen_feature(XENFEAT_not_direct_mapped) && xen_initial_domain())
-		return 1;
-	return 0;
-}
-
 static int __init xen_mm_init(void)
 {
 	struct gnttab_cache_flush cflush;
diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index 16a2b2b1c54d..e55409caaee3 100644
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@@ -43,6 +43,7 @@ 
 #include <linux/sizes.h>
 #include <asm/tlb.h>
 #include <asm/alternative.h>
+#include <asm/xen/swiotlb-xen.h>
 
 /*
  * We need to be able to catch inadvertent references to memstart_addr
@@ -482,7 +483,7 @@  void __init mem_init(void)
 	if (swiotlb_force == SWIOTLB_FORCE ||
 	    max_pfn > PFN_DOWN(arm64_dma_phys_limit))
 		swiotlb_init(1);
-	else
+	else if (!xen_swiotlb_detect())
 		swiotlb_force = SWIOTLB_NO_FORCE;
 
 	set_max_mapnr(max_pfn - PHYS_PFN_OFFSET);
diff --git a/include/xen/arm/swiotlb-xen.h b/include/xen/arm/swiotlb-xen.h
index 2994fe6031a0..33336ab58afc 100644
--- a/include/xen/arm/swiotlb-xen.h
+++ b/include/xen/arm/swiotlb-xen.h
@@ -2,6 +2,19 @@ 
 #ifndef _ASM_ARM_SWIOTLB_XEN_H
 #define _ASM_ARM_SWIOTLB_XEN_H
 
-extern int xen_swiotlb_detect(void);
+#include <xen/features.h>
+#include <xen/xen.h>
+
+static inline int xen_swiotlb_detect(void)
+{
+	if (!xen_domain())
+		return 0;
+	if (xen_feature(XENFEAT_direct_mapped))
+		return 1;
+	/* legacy case */
+	if (!xen_feature(XENFEAT_not_direct_mapped) && xen_initial_domain())
+		return 1;
+	return 0;
+}
 
 #endif /* _ASM_ARM_SWIOTLB_XEN_H */