Le 23/10/2024 à 07:36, Christoph Hellwig a écrit :
page_to_pfn is usually implemented by pointer arithmetics on the memory
map, which means that bogus input can lead to even more bogus output.
Powerpc had a pfn_valid check on the intermediate pfn in the page_to_phys
implementation when CONFIG_DEBUG_VIRTUAL is defined, which seems
generally useful, so add that to the generic version.
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
---
include/asm-generic/memory_model.h | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/include/asm-generic/memory_model.h b/include/asm-generic/memory_model.h
index a73a140cbecd..6d1fb6162ac1 100644
--- a/include/asm-generic/memory_model.h
+++ b/include/asm-generic/memory_model.h
@@ -64,7 +64,17 @@ static inline int pfn_valid(unsigned long pfn)
#define page_to_pfn __page_to_pfn
#define pfn_to_page __pfn_to_page
+#ifdef CONFIG_DEBUG_VIRTUAL
+#define page_to_phys(page) \
+({ \
+ unsigned long __pfn = page_to_pfn(page); \
+ \
+ WARN_ON_ONCE(!pfn_valid(__pfn)); \
On powerpc I think it was a WARN_ON().
Will a WARN_ON_ONCE() be enough ?
+ PFN_PHYS(__pfn); \
+})
+#else
#define page_to_phys(page) PFN_PHYS(page_to_pfn(page))
+#endif /* CONFIG_DEBUG_VIRTUAL */
#define phys_to_page(phys) pfn_to_page(PHYS_PFN(phys))
#endif /* __ASSEMBLY__ */
