swiotlb: Export swiotlb_max_segment to users

So they can figure out what is the optimal number of pages
that can be contingously stitched together without fear of
bounce buffer.

We also expose an mechanism for sub-users of SWIOTLB API, such
as Xen-SWIOTLB to set the max segment value. And lastly
if swiotlb=force is set (which mandates we bounce buffer everything)
we set max_segment so at least we can bounce buffer one 4K page
instead of a giant 512KB one for which we may not have space.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reported-and-Tested-by: Juergen Gross <jgross@suse.com>

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index d0dcaf3..115fa39 100644 (file)
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2290,15 +2290,6 @@ unlock:
        mutex_unlock(&obj->mm.lock);
 }
 
-static unsigned int swiotlb_max_size(void)
-{
-#if IS_ENABLED(CONFIG_SWIOTLB)
-       return rounddown(swiotlb_nr_tbl() << IO_TLB_SHIFT, PAGE_SIZE);
-#else
-       return 0;
-#endif
-}
-
 static void i915_sg_trim(struct sg_table *orig_st)
 {
        struct sg_table new_st;
@@ -2345,7 +2336,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
        GEM_BUG_ON(obj->base.read_domains & I915_GEM_GPU_DOMAINS);
        GEM_BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS);
 
-       max_segment = swiotlb_max_size();
+       max_segment = swiotlb_max_segment();
        if (!max_segment)
                max_segment = rounddown(UINT_MAX, PAGE_SIZE);
 

diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index aba1200..f905d6e 100644 (file)
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -275,6 +275,10 @@ retry:
                rc = 0;
        } else
                rc = swiotlb_late_init_with_tbl(xen_io_tlb_start, xen_io_tlb_nslabs);
+
+       if (!rc)
+               swiotlb_set_max_segment(PAGE_SIZE);
+
        return rc;
 error:
        if (repeat--) {

diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h
index d9c84a9..4ee479f 100644 (file)
--- a/include/linux/swiotlb.h
+++ b/include/linux/swiotlb.h
@@ -114,11 +114,14 @@ swiotlb_dma_supported(struct device *hwdev, u64 mask);
 
 #ifdef CONFIG_SWIOTLB
 extern void __init swiotlb_free(void);
+unsigned int swiotlb_max_segment(void);
 #else
 static inline void swiotlb_free(void) { }
+static inline unsigned int swiotlb_max_segment(void) { return 0; }
 #endif
 
 extern void swiotlb_print_info(void);
 extern int is_swiotlb_buffer(phys_addr_t paddr);
+extern void swiotlb_set_max_segment(unsigned int);
 
 #endif /* __LINUX_SWIOTLB_H */

diff --git a/lib/swiotlb.c b/lib/swiotlb.c
index 9def738..975b8fc 100644 (file)
--- a/lib/swiotlb.c
+++ b/lib/swiotlb.c
@@ -83,6 +83,12 @@ static unsigned int *io_tlb_list;
 static unsigned int io_tlb_index;
 
 /*
+ * Max segment that we can provide which (if pages are contingous) will
+ * not be bounced (unless SWIOTLB_FORCE is set).
+ */
+unsigned int max_segment;
+
+/*
  * We need to save away the original address corresponding to a mapped entry
  * for the sync operations.
  */
@@ -124,6 +130,20 @@ unsigned long swiotlb_nr_tbl(void)
 }
 EXPORT_SYMBOL_GPL(swiotlb_nr_tbl);
 
+unsigned int swiotlb_max_segment(void)
+{
+       return max_segment;
+}
+EXPORT_SYMBOL_GPL(swiotlb_max_segment);
+
+void swiotlb_set_max_segment(unsigned int val)
+{
+       if (swiotlb_force == SWIOTLB_FORCE)
+               max_segment = 1;
+       else
+               max_segment = rounddown(val, PAGE_SIZE);
+}
+
 /* default to 64MB */
 #define IO_TLB_DEFAULT_SIZE (64UL<<20)
 unsigned long swiotlb_size_or_default(void)
@@ -205,6 +225,7 @@ int __init swiotlb_init_with_tbl(char *tlb, unsigned long nslabs, int verbose)
        if (verbose)
                swiotlb_print_info();
 
+       swiotlb_set_max_segment(io_tlb_nslabs << IO_TLB_SHIFT);
        return 0;
 }
 
@@ -283,6 +304,7 @@ swiotlb_late_init_with_default_size(size_t default_size)
        rc = swiotlb_late_init_with_tbl(vstart, io_tlb_nslabs);
        if (rc)
                free_pages((unsigned long)vstart, order);
+
        return rc;
 }
 
@@ -337,6 +359,8 @@ swiotlb_late_init_with_tbl(char *tlb, unsigned long nslabs)
 
        late_alloc = 1;
 
+       swiotlb_set_max_segment(io_tlb_nslabs << IO_TLB_SHIFT);
+
        return 0;
 
 cleanup4:
@@ -351,6 +375,7 @@ cleanup2:
        io_tlb_end = 0;
        io_tlb_start = 0;
        io_tlb_nslabs = 0;
+       max_segment = 0;
        return -ENOMEM;
 }
 
@@ -379,6 +404,7 @@ void __init swiotlb_free(void)
                                   PAGE_ALIGN(io_tlb_nslabs << IO_TLB_SHIFT));
        }
        io_tlb_nslabs = 0;
+       max_segment = 0;
 }
 
 int is_swiotlb_buffer(phys_addr_t paddr)