[PATCH V2 1/2] powerpc/vas: Report proper error code for address translation failure

[PATCH V2 1/2] powerpc/vas: Report proper error code for address translation failure

[Haren Myneni]

P9 DD2 NX workbook (Table 4-36) says DMA controller uses CC=5
internally for translation fault handling. NX reserves CC=250 for
OS to notify user space when NX encounters address translation
failure on the request buffer. Not an issue in earlier releases
as NX does not get faults on kernel addresses.

This patch defines CSB_CC_FAULT_ADDRESS(250) and updates CSB.CC with
this proper error code for user space.

Signed-off-by: Haren Myneni <haren@linux.ibm.com>

Changes in V2:
- Use CSB_CC_FAULT_ADDRESS instead of CSB_CC_ADDRESS_TRANSLATION
  to distinguish from other error codes.
- Add NX workbook reference in the comment.

---
 Documentation/powerpc/vas-api.rst          | 2 +-
 arch/powerpc/include/asm/icswx.h           | 4 ++++
 arch/powerpc/platforms/powernv/vas-fault.c | 2 +-
 3 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/Documentation/powerpc/vas-api.rst b/Documentation/powerpc/vas-api.rst
index 1217c2f..788dc83 100644
--- a/Documentation/powerpc/vas-api.rst
+++ b/Documentation/powerpc/vas-api.rst
@@ -213,7 +213,7 @@ request buffers are not in memory. The operating system handles the fault by
 updating CSB with the following data:
 
 	csb.flags = CSB_V;
-	csb.cc = CSB_CC_TRANSLATION;
+	csb.cc = CSB_CC_FAULT_ADDRESS;
 	csb.ce = CSB_CE_TERMINATION;
 	csb.address = fault_address;
 
diff --git a/arch/powerpc/include/asm/icswx.h b/arch/powerpc/include/asm/icswx.h
index 965b1f3..9bc7c58 100644
--- a/arch/powerpc/include/asm/icswx.h
+++ b/arch/powerpc/include/asm/icswx.h
@@ -77,6 +77,10 @@ struct coprocessor_completion_block {
 #define CSB_CC_CHAIN		(37)
 #define CSB_CC_SEQUENCE		(38)
 #define CSB_CC_HW		(39)
+/*
+ * P9 DD NX Workbook 3.2 (Table 4-36): Address translation fault
+ */
+#define	CSB_CC_FAULT_ADDRESS	(250)
 
 #define CSB_SIZE		(0x10)
 #define CSB_ALIGN		CSB_SIZE
diff --git a/arch/powerpc/platforms/powernv/vas-fault.c b/arch/powerpc/platforms/powernv/vas-fault.c
index 266a6ca..3d21fce 100644
--- a/arch/powerpc/platforms/powernv/vas-fault.c
+++ b/arch/powerpc/platforms/powernv/vas-fault.c
@@ -79,7 +79,7 @@ static void update_csb(struct vas_window *window,
 	csb_addr = (void __user *)be64_to_cpu(crb->csb_addr);
 
 	memset(&csb, 0, sizeof(csb));
-	csb.cc = CSB_CC_TRANSLATION;
+	csb.cc = CSB_CC_FAULT_ADDRESS;
 	csb.ce = CSB_CE_TERMINATION;
 	csb.cs = 0;
 	csb.count = 0;
-- 
1.8.3.1

[PATCH V2 2/2] selftests/powerpc: Use proper error code to check fault address

[Haren Myneni]

ERR_NX_TRANSLATION(CSB.CC=5) is for internal to VAS for fault handling
and should not used by OS. ERR_NX_AT_FAULT(CSB.CC=250) is the proper
error code should be reported by OS when NX encounters address
translation failure.

This patch uses CC=250 to determine the fault address when the request
is not successful.

Signed-off-by: Haren Myneni <haren@linux.ibm.com>
---
 tools/testing/selftests/powerpc/nx-gzip/gunz_test.c  | 4 ++--
 tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c b/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c
index 6ee0fde..7c23d3d 100644
--- a/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c
+++ b/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c
@@ -698,13 +698,13 @@ int decompress_file(int argc, char **argv, void *devhandle)
 
 	switch (cc) {
 
-	case ERR_NX_TRANSLATION:
+	case ERR_NX_AT_FAULT:
 
 		/* We touched the pages ahead of time.  In the most common case
 		 * we shouldn't be here.  But may be some pages were paged out.
 		 * Kernel should have placed the faulting address to fsaddr.
 		 */
-		NXPRT(fprintf(stderr, "ERR_NX_TRANSLATION %p\n",
+		NXPRT(fprintf(stderr, "ERR_NX_AT_FAULT %p\n",
 			      (void *)cmdp->crb.csb.fsaddr));
 
 		if (pgfault_retries == NX_MAX_FAULTS) {
diff --git a/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c b/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c
index 7496a83..02dffb6 100644
--- a/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c
+++ b/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c
@@ -306,13 +306,13 @@ int compress_file(int argc, char **argv, void *handle)
 			lzcounts, cmdp, handle);
 
 		if (cc != ERR_NX_OK && cc != ERR_NX_TPBC_GT_SPBC &&
-		    cc != ERR_NX_TRANSLATION) {
+		    cc != ERR_NX_AT_FAULT) {
 			fprintf(stderr, "nx error: cc= %d\n", cc);
 			exit(-1);
 		}
 
 		/* Page faults are handled by the user code */
-		if (cc == ERR_NX_TRANSLATION) {
+		if (cc == ERR_NX_AT_FAULT) {
 			NXPRT(fprintf(stderr, "page fault: cc= %d, ", cc));
 			NXPRT(fprintf(stderr, "try= %d, fsa= %08llx\n",
 				  fault_tries,
-- 
1.8.3.1

Re: [PATCH V2 1/2] powerpc/vas: Report proper error code for address translation failure

[Michael Ellerman]

Haren Myneni <haren@linux.ibm.com> writes:
> P9 DD2 NX workbook (Table 4-36) says DMA controller uses CC=5
> internally for translation fault handling. NX reserves CC=250 for
> OS to notify user space when NX encounters address translation
> failure on the request buffer. Not an issue in earlier releases
> as NX does not get faults on kernel addresses.
>
> This patch defines CSB_CC_FAULT_ADDRESS(250) and updates CSB.CC with
> this proper error code for user space.

I added:

Fixes: c96c4436aba4 ("powerpc/vas: Update CSB and notify process for fault CRBs")

> Signed-off-by: Haren Myneni <haren@linux.ibm.com>
>
> Changes in V2:
> - Use CSB_CC_FAULT_ADDRESS instead of CSB_CC_ADDRESS_TRANSLATION
>   to distinguish from other error codes.
> - Add NX workbook reference in the comment.

The change log goes after the --- line.

> ---
>  Documentation/powerpc/vas-api.rst          | 2 +-
>  arch/powerpc/include/asm/icswx.h           | 4 ++++
>  arch/powerpc/platforms/powernv/vas-fault.c | 2 +-
>  3 files changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/powerpc/vas-api.rst b/Documentation/powerpc/vas-api.rst
> index 1217c2f..788dc83 100644
> --- a/Documentation/powerpc/vas-api.rst
> +++ b/Documentation/powerpc/vas-api.rst
> @@ -213,7 +213,7 @@ request buffers are not in memory. The operating system handles the fault by
>  updating CSB with the following data:
>  
>  	csb.flags = CSB_V;
> -	csb.cc = CSB_CC_TRANSLATION;
> +	csb.cc = CSB_CC_FAULT_ADDRESS;
>  	csb.ce = CSB_CE_TERMINATION;
>  	csb.address = fault_address;
>  
> diff --git a/arch/powerpc/include/asm/icswx.h b/arch/powerpc/include/asm/icswx.h
> index 965b1f3..9bc7c58 100644
> --- a/arch/powerpc/include/asm/icswx.h
> +++ b/arch/powerpc/include/asm/icswx.h
> @@ -77,6 +77,10 @@ struct coprocessor_completion_block {
>  #define CSB_CC_CHAIN		(37)
>  #define CSB_CC_SEQUENCE		(38)
>  #define CSB_CC_HW		(39)
> +/*
> + * P9 DD NX Workbook 3.2 (Table 4-36): Address translation fault

I changed that to "P9 DD2 NX" which I assume is what you meant, it
matches the change log.

> + */
> +#define	CSB_CC_FAULT_ADDRESS	(250)
>  
>  #define CSB_SIZE		(0x10)
>  #define CSB_ALIGN		CSB_SIZE
> diff --git a/arch/powerpc/platforms/powernv/vas-fault.c b/arch/powerpc/platforms/powernv/vas-fault.c
> index 266a6ca..3d21fce 100644
> --- a/arch/powerpc/platforms/powernv/vas-fault.c
> +++ b/arch/powerpc/platforms/powernv/vas-fault.c
> @@ -79,7 +79,7 @@ static void update_csb(struct vas_window *window,
>  	csb_addr = (void __user *)be64_to_cpu(crb->csb_addr);
>  
>  	memset(&csb, 0, sizeof(csb));
> -	csb.cc = CSB_CC_TRANSLATION;
> +	csb.cc = CSB_CC_FAULT_ADDRESS;
>  	csb.ce = CSB_CE_TERMINATION;
>  	csb.cs = 0;
>  	csb.count = 0;
> -- 
> 1.8.3.1


cheers

Re: [PATCH V2 1/2] powerpc/vas: Report proper error code for address translation failure

[Michael Ellerman]

On Fri, 10 Jul 2020 16:47:19 -0700, Haren Myneni wrote:
> P9 DD2 NX workbook (Table 4-36) says DMA controller uses CC=5
> internally for translation fault handling. NX reserves CC=250 for
> OS to notify user space when NX encounters address translation
> failure on the request buffer. Not an issue in earlier releases
> as NX does not get faults on kernel addresses.
> 
> This patch defines CSB_CC_FAULT_ADDRESS(250) and updates CSB.CC with
> this proper error code for user space.
> 
> [...]

Applied to powerpc/fixes.

[1/2] powerpc/vas: Report proper error code for address translation failure
      https://git.kernel.org/powerpc/c/6068e1a4427e88f5cc62f238d1baf94a8b824ef4
[2/2] selftests/powerpc: Use proper error code to check fault address
      https://git.kernel.org/powerpc/c/f0479c4bcbd92d1a457d4a43bcab79f29d11334a

cheers