[PATCH v2 0/1] s390: ipl: fix physical-virtual confusion for diag308

[PATCH v2 0/1] s390: ipl: fix physical-virtual confusion for diag308

[Nico Boehr]

v2:
---
* add special handling for addr NULL (thanks Alexander)



Nico Boehr (1):
  s390: ipl: fix physical-virtual confusion for diag308

 arch/s390/kernel/ipl.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

-- 
2.39.1

[PATCH v2 1/1] s390: ipl: fix physical-virtual confusion for diag308

[Nico Boehr]

Diag 308 subcodes expect a physical address as their parameter.

This currently is not a bug, but in the future physical and virtual
addresses might differ.

Fix the confusion by doing a virtual-to-physical conversion in the
exported diag308() and leave the assembly wrapper __diag308() alone.

Note that several callers pass NULL as addr, this is _currently_ fine since
virt_to_phys(0) == 0. But in the future, lowcore addresses might no
longer be special, so make sure we pass 0 to hardware in case addr is
NULL.

Suggested-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
---
 arch/s390/kernel/ipl.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/s390/kernel/ipl.c b/arch/s390/kernel/ipl.c
index 5f0f5c86963a..d8b8a87626f7 100644
--- a/arch/s390/kernel/ipl.c
+++ b/arch/s390/kernel/ipl.c
@@ -176,7 +176,7 @@ static bool reipl_fcp_clear;
 static bool reipl_ccw_clear;
 static bool reipl_eckd_clear;
 
-static inline int __diag308(unsigned long subcode, void *addr)
+static inline int __diag308(unsigned long subcode, u64 addr)
 {
 	union register_pair r1;
 
@@ -195,7 +195,7 @@ static inline int __diag308(unsigned long subcode, void *addr)
 int diag308(unsigned long subcode, void *addr)
 {
 	diag_stat_inc(DIAG_STAT_X308);
-	return __diag308(subcode, addr);
+	return __diag308(subcode, addr ? virt_to_phys(addr) : 0);
 }
 EXPORT_SYMBOL_GPL(diag308);
 
-- 
2.39.1

Re: [PATCH v2 1/1] s390: ipl: fix physical-virtual confusion for diag308

[Alexander Gordeev]

On Fri, Mar 10, 2023 at 02:02:36PM +0100, Nico Boehr wrote:
> Diag 308 subcodes expect a physical address as their parameter.
> 
> This currently is not a bug, but in the future physical and virtual
> addresses might differ.
> 
> Fix the confusion by doing a virtual-to-physical conversion in the
> exported diag308() and leave the assembly wrapper __diag308() alone.

You change the prototype though ;)

> Note that several callers pass NULL as addr, this is _currently_ fine since
> virt_to_phys(0) == 0. But in the future, lowcore addresses might no
> longer be special, so make sure we pass 0 to hardware in case addr is
> NULL.

This paragraph seems to me is extra or not precise at least.
The reason for explicit NULL check is not virt_to_phys(NULL)
might return non-0 in the future, but rather virt_to_phys(NULL)
is not expected returning 0 on any architecture.

> Suggested-by: Marc Hartmayer <mhartmay@linux.ibm.com>
> Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
> ---
>  arch/s390/kernel/ipl.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/s390/kernel/ipl.c b/arch/s390/kernel/ipl.c
> index 5f0f5c86963a..d8b8a87626f7 100644
> --- a/arch/s390/kernel/ipl.c
> +++ b/arch/s390/kernel/ipl.c
> @@ -176,7 +176,7 @@ static bool reipl_fcp_clear;
>  static bool reipl_ccw_clear;
>  static bool reipl_eckd_clear;
>  
> -static inline int __diag308(unsigned long subcode, void *addr)
> +static inline int __diag308(unsigned long subcode, u64 addr)

Physical addresses are unsigned longs. Why u64 then?
Also the cast to unsigned long (not seen here) becomes redundant.

>  {
>  	union register_pair r1;