[PATCH net-next 0/2] nfp: fix rtsym and XDP register handling in debug dump

[PATCH net-next 0/2] nfp: fix rtsym and XDP register handling in debug dump

[Simon Horman]

Hi,

this series resolves two problems in the recently added debug dump facility.

* Correctly handle reading absolute rtysms
* Correctly handle special-case PB register reads

These fixes are for code only present in net-next.

Carl Heymann (2):
  nfp: fix absolute rtsym handling in debug dump
  nfp: fix XPB register reads in debug dump

 .../net/ethernet/netronome/nfp/nfp_net_debugdump.c | 57 +++++++++++++++-------
 1 file changed, 39 insertions(+), 18 deletions(-)

-- 
2.11.0

[PATCH net-next 1/2] nfp: fix absolute rtsym handling in debug dump

[Simon Horman]

From: Carl Heymann <carl.heymann@netronome.com>

In TLV-based ethtool debug dumps, don't do a CPP read for absolute
rtsyms, use the addr field in the symbol table directly as the value.

Without this fix rtsym gro_release_ring_0 is 4 bytes of zeros.
With this fix the correct value, 0x0000004a 0x00000000 is reported.

The values may be read using ethtool debug level 2.
 # ethtool -W <netdev> 2
 # ethtool -w <netdev> data dump.dat

Fixes: e1e798e3fd93 ("nfp: dump rtsyms")
Signed-off-by: Carl Heymann <carl.heymann@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
---
 .../net/ethernet/netronome/nfp/nfp_net_debugdump.c | 42 +++++++++++++---------
 1 file changed, 26 insertions(+), 16 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c b/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
index cbff0adad235..315bd22819ff 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
@@ -265,6 +265,7 @@ nfp_calc_rtsym_dump_sz(struct nfp_pf *pf, struct nfp_dump_tl *spec)
 	struct nfp_dumpspec_rtsym *spec_rtsym;
 	const struct nfp_rtsym *sym;
 	u32 tl_len, key_len;
+	u32 size;
 
 	spec_rtsym = (struct nfp_dumpspec_rtsym *)spec;
 	tl_len = be32_to_cpu(spec->length);
@@ -276,8 +277,13 @@ nfp_calc_rtsym_dump_sz(struct nfp_pf *pf, struct nfp_dump_tl *spec)
 	if (!sym)
 		return nfp_dump_error_tlv_size(spec);
 
+	if (sym->type == NFP_RTSYM_TYPE_ABS)
+		size = sizeof(sym->addr);
+	else
+		size = sym->size;
+
 	return ALIGN8(offsetof(struct nfp_dump_rtsym, rtsym) + key_len + 1) +
-	       ALIGN8(sym->size);
+	       ALIGN8(size);
 }
 
 static int
@@ -654,21 +660,25 @@ nfp_dump_single_rtsym(struct nfp_pf *pf, struct nfp_dumpspec_rtsym *spec,
 		header_size - offsetof(struct nfp_dump_rtsym, rtsym);
 	memcpy(dump_header->rtsym, spec->rtsym, key_len + 1);
 
-	cpp_params.target = sym->target;
-	cpp_params.action = NFP_CPP_ACTION_RW;
-	cpp_params.token  = 0;
-	cpp_params.island = sym->domain;
-	cpp_id = nfp_get_numeric_cpp_id(&cpp_params);
-
-	dump_header->cpp.cpp_id = cpp_params;
-	dump_header->cpp.offset = cpu_to_be32(sym->addr);
-	dump_header->cpp.dump_length = cpu_to_be32(sym->size);
-
-	bytes_read = nfp_cpp_read(pf->cpp, cpp_id, sym->addr, dest, sym->size);
-	if (bytes_read != sym->size) {
-		if (bytes_read >= 0)
-			bytes_read = -EIO;
-		dump_header->error = cpu_to_be32(bytes_read);
+	if (sym->type == NFP_RTSYM_TYPE_ABS) {
+		dump_header->cpp.dump_length = cpu_to_be32(sizeof(sym->addr));
+		*(u64 *)dest = sym->addr;
+	} else {
+		cpp_params.target = sym->target;
+		cpp_params.action = NFP_CPP_ACTION_RW;
+		cpp_params.token  = 0;
+		cpp_params.island = sym->domain;
+		cpp_id = nfp_get_numeric_cpp_id(&cpp_params);
+		dump_header->cpp.cpp_id = cpp_params;
+		dump_header->cpp.offset = cpu_to_be32(sym->addr);
+		dump_header->cpp.dump_length = cpu_to_be32(sym->size);
+		bytes_read = nfp_cpp_read(pf->cpp, cpp_id, sym->addr, dest,
+					  sym->size);
+		if (bytes_read != sym->size) {
+			if (bytes_read >= 0)
+				bytes_read = -EIO;
+			dump_header->error = cpu_to_be32(bytes_read);
+		}
 	}
 
 	return 0;
-- 
2.11.0

[PATCH net-next 2/2] nfp: fix XPB register reads in debug dump

[Simon Horman]

From: Carl Heymann <carl.heymann@netronome.com>

For XPB registers reads, some island IDs require special handling (e.g.
ARM island), which is already taken care of in nfp_xpb_readl(), so use
that instead of a straight CPP read.

Without this fix all "xpbm:ArmIsldXpbmMap.*" registers are reported as
0xffffffff. It has also been observed to cause a system reboot.

With this fix correct values are reported, none of which are 0xffffffff.

The values may be read using ethtool debug level 2.
 # ethtool -W <netdev> 2
 # ethtool -w <netdev> data dump.dat

Fixes: 0e6c4955e149 ("nfp: dump CPP, XPB and direct ME CSRs")
Signed-off-by: Carl Heymann <carl.heymann@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
---
 drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c b/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
index 315bd22819ff..173646e17e94 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_debugdump.c
@@ -38,6 +38,7 @@
 #include "nfp_main.h"
 #include "nfpcore/nfp.h"
 #include "nfpcore/nfp_nffw.h"
+#include "nfpcore/nfp6000/nfp6000.h"
 
 #define NFP_DUMP_SPEC_RTSYM	"_abi_dump_spec"
 
@@ -478,6 +479,12 @@ static int nfp_dump_hwinfo_field(struct nfp_pf *pf, struct nfp_dump_tl *spec,
 	return 0;
 }
 
+static bool is_xpb_read(struct nfp_dumpspec_cpp_isl_id *cpp_id)
+{
+	return cpp_id->target == NFP_CPP_TARGET_ISLAND_XPB &&
+	       cpp_id->action == 0 && cpp_id->token == 0;
+}
+
 static int
 nfp_dump_csr_range(struct nfp_pf *pf, struct nfp_dumpspec_csr *spec_csr,
 		   struct nfp_dump_state *dump)
@@ -511,8 +518,12 @@ nfp_dump_csr_range(struct nfp_pf *pf, struct nfp_dumpspec_csr *spec_csr,
 	max_rd_addr = cpp_rd_addr + be32_to_cpu(spec_csr->cpp.dump_length);
 
 	while (cpp_rd_addr < max_rd_addr) {
-		bytes_read = nfp_cpp_read(pf->cpp, cpp_id, cpp_rd_addr, dest,
-					  reg_sz);
+		if (is_xpb_read(&spec_csr->cpp.cpp_id))
+			bytes_read = nfp_xpb_readl(pf->cpp, cpp_rd_addr,
+						   (u32 *)dest);
+		else
+			bytes_read = nfp_cpp_read(pf->cpp, cpp_id, cpp_rd_addr,
+						  dest, reg_sz);
 		if (bytes_read != reg_sz) {
 			if (bytes_read >= 0)
 				bytes_read = -EIO;
-- 
2.11.0

Re: [PATCH net-next 0/2] nfp: fix rtsym and XDP register handling in debug dump

[Simon Horman]

On Thu, Dec 14, 2017 at 10:50:24AM +0100, Simon Horman wrote:
> Hi,
> 
> this series resolves two problems in the recently added debug dump facility.
> 
> * Correctly handle reading absolute rtysms
> * Correctly handle special-case PB register reads
> 
> These fixes are for code only present in net-next.

Hi Dave,

It seems that I made a thinko in the title of the cover letter.  s/XDP/XPB/
These changes do not relate to XDP.

> Carl Heymann (2):
>   nfp: fix absolute rtsym handling in debug dump
>   nfp: fix XPB register reads in debug dump
> 
>  .../net/ethernet/netronome/nfp/nfp_net_debugdump.c | 57 +++++++++++++++-------
>  1 file changed, 39 insertions(+), 18 deletions(-)
> 
> -- 
> 2.11.0
> 

Re: [PATCH net-next 0/2] nfp: fix rtsym and XDP register handling in debug dump

[David Miller]

From: Simon Horman <simon.horman@netronome.com>
Date: Fri, 15 Dec 2017 15:18:29 +0100

> On Thu, Dec 14, 2017 at 10:50:24AM +0100, Simon Horman wrote:
>> Hi,
>> 
>> this series resolves two problems in the recently added debug dump facility.
>> 
>> * Correctly handle reading absolute rtysms
>> * Correctly handle special-case PB register reads
>> 
>> These fixes are for code only present in net-next.
> 
> Hi Dave,
> 
> It seems that I made a thinko in the title of the cover letter.  s/XDP/XPB/
> These changes do not relate to XDP.

Series applied with this fixed, thanks.