The following commit has been merged into the x86/mm branch of tip:

Commit-ID:     8e5647a723c49d73b9f108a8bb38e8c29d3948ea
Gitweb:        https://git.kernel.org/tip/8e5647a723c49d73b9f108a8bb38e8c29d3948ea
Author:        Michael Roth <michael.roth@amd.com>
AuthorDate:    Wed, 22 Nov 2023 10:37:00 -06:00
Committer:     Ingo Molnar <mingo@kernel.org>
CommitterDate: Thu, 23 Nov 2023 11:13:21 +01:00

x86/mm: Ensure input to pfn_to_kaddr() is treated as a 64-bit type

On 64-bit platforms, the pfn_to_kaddr() macro requires that the input
value is 64 bits in order to ensure that valid address bits don't get
lost when shifting that input by PAGE_SHIFT to calculate the physical
address to provide a virtual address for.

One such example is in pvalidate_pages() (used by SEV-SNP guests), where
the GFN in the struct used for page-state change requests is a 40-bit
bit-field, so attempts to pass this GFN field directly into
pfn_to_kaddr() ends up causing guest crashes when dealing with addresses
above the 1TB range due to the above.

Fix this issue with SEV-SNP guests, as well as any similar cases that
might cause issues in current/future code, by using an inline function,
instead of a macro, so that the input is implicitly cast to the
expected 64-bit input type prior to performing the shift operation.

While it might be argued that the issue is on the caller side, other
archs/macros have taken similar approaches to deal with instances like
this, such as ARM explicitly casting the input to phys_addr_t:

  e48866647b48 ("ARM: 8396/1: use phys_addr_t in pfn_to_kaddr()")

A C inline function is even better though.

[ mingo: Refined the changelog some more & added __always_inline. ]

Fixes: 6c3211796326 ("x86/sev: Add SNP-specific unaccepted memory support")
Suggested-by: Dave Hansen <dave.hansen@intel.com>
Suggested-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20231122163700.400507-1-michael.roth@amd.com
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
---
 arch/x86/include/asm/page.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/page.h b/arch/x86/include/asm/page.h
index d18e5c3..1b93ff8 100644
--- a/arch/x86/include/asm/page.h
+++ b/arch/x86/include/asm/page.h
@@ -66,10 +66,14 @@ static inline void copy_user_page(void *to, void *from, unsigned long vaddr,
  * virt_addr_valid(kaddr) returns true.
  */
 #define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
-#define pfn_to_kaddr(pfn)      __va((pfn) << PAGE_SHIFT)
 extern bool __virt_addr_valid(unsigned long kaddr);
 #define virt_addr_valid(kaddr)	__virt_addr_valid((unsigned long) (kaddr))
 
+static __always_inline void *pfn_to_kaddr(unsigned long pfn)
+{
+	return __va(pfn << PAGE_SHIFT);
+}
+
 static __always_inline u64 __canonical_address(u64 vaddr, u8 vaddr_bits)
 {
 	return ((s64)vaddr << (64 - vaddr_bits)) >> (64 - vaddr_bits);

On Thu, 23 Nov 2023 at 02:31, tip-bot2 for Michael Roth
<tip-bot2@linutronix.de> wrote:
>
> On 64-bit platforms, the pfn_to_kaddr() macro requires that the input
> value is 64 bits in order to ensure that valid address bits don't get
> lost when shifting that input by PAGE_SHIFT to calculate the physical
> address to provide a virtual address for.

Bah. The commit is obviously fine, but can we please just get rid of
that broken pfn_to_kaddr() thing entirely?

It's a bogus mis-spelling of pfn_to_virt(), and I don't know why that
horrid thing exists. In *all* other situations we talk about "virt"
for kernel virtual addresses, I don't know where that horrid "kaddr"
came from (ie "virt_to_page()" and friends).

Most notably, we have "virt_to_pfn()" being quite commonly used. We
don't even have that kaddr_to_pfn(), which just shows *how* bogus this
whole "pfn_to_kaddr()" crud is.

The good news is that there aren't a ton of users. Anybody willing to
just do a search-and-replace and get rid of this pointless and wrong
thing?

Using "pfn_to_virt()" has the added advantage that we have a generic
implementation of it that isn't duplicated pointlessly for N
architectures, and that didn't have this bug:

  static inline void *pfn_to_virt(unsigned long pfn)
  {
        return __va(pfn) << PAGE_SHIFT;
  }
  #define pfn_to_virt pfn_to_virt

Hmm?

Amusingly (or sadly), we have s390 holding up the flag of sanity, and having

    #define pfn_to_kaddr(pfn)  pfn_to_virt(pfn)

and then we'd only need to fix the hexagon version of that macro
(since Hexagon made its own version, with the old bug - but I guess
Hexagon is 32-bit only and hopefully never grows 64-bit (??) so maybe
nobody cares).

           Linus