[PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[Ivan Vecera]

Commit 5b64c66f58d ("ethtool: Add netlink handler for
getmodule\r (-m)") provided a netlink variant for getmodule
but also introduced a regression as netlink output is different
from ioctl output that provides information from A2h page
via sff8472_show_all().

To fix this the netlink path should check a presence of A2h page
by value of bit 6 in byte 92 of page A0h and if it is set then
get A2h page and call sff8472_show_all().

Fixes: 5b64c66f58d ("ethtool: Add netlink handler for getmodule\r (-m)")
Tested-by: Daniel Juarez <djuarezg@cern.ch>
Co-authored-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
---
 sfpid.c | 54 ++++++++++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 46 insertions(+), 8 deletions(-)

diff --git a/sfpid.c b/sfpid.c
index 621d1e86c278..1bc45c183770 100644
--- a/sfpid.c
+++ b/sfpid.c
@@ -13,8 +13,9 @@
 #include "sff-common.h"
 #include "netlink/extapi.h"
 
-#define SFF8079_PAGE_SIZE	0x80
-#define SFF8079_I2C_ADDRESS_LOW	0x50
+#define SFF8079_PAGE_SIZE		0x80
+#define SFF8079_I2C_ADDRESS_LOW		0x50
+#define SFF8079_I2C_ADDRESS_HIGH	0x51
 
 static void sff8079_show_identifier(const __u8 *id)
 {
@@ -450,18 +451,55 @@ void sff8079_show_all_ioctl(const __u8 *id)
 	sff8079_show_all_common(id);
 }
 
-int sff8079_show_all_nl(struct cmd_context *ctx)
+static int sff8079_get_eeprom_page(struct cmd_context *ctx, u8 i2c_address,
+				   __u8 *buf)
 {
 	struct ethtool_module_eeprom request = {
 		.length = SFF8079_PAGE_SIZE,
-		.i2c_address = SFF8079_I2C_ADDRESS_LOW,
+		.i2c_address = i2c_address,
 	};
 	int ret;
 
 	ret = nl_get_eeprom_page(ctx, &request);
-	if (ret < 0)
-		return ret;
-	sff8079_show_all_common(request.data);
+	if (!ret)
+		memcpy(buf, request.data, SFF8079_PAGE_SIZE);
+
+	return ret;
+}
+
+int sff8079_show_all_nl(struct cmd_context *ctx)
+{
+	u8 *buf;
+	int ret;
+
+	/* The SFF-8472 parser expects a single buffer that contains the
+	 * concatenation of the first 256 bytes from addresses A0h and A2h,
+	 * respectively.
+	 */
+	buf = calloc(1, ETH_MODULE_SFF_8472_LEN);
+	if (!buf)
+		return -ENOMEM;
+
+	/* Read A0h page */
+	ret = sff8079_get_eeprom_page(ctx, SFF8079_I2C_ADDRESS_LOW, buf);
+	if (ret)
+		goto out;
+
+	sff8079_show_all_common(buf);
+
+	/* Finish if A2h page is not present */
+	if (!(buf[92] & (1 << 6)))
+		goto out;
+
+	/* Read A2h page */
+	ret = sff8079_get_eeprom_page(ctx, SFF8079_I2C_ADDRESS_HIGH,
+				      buf + ETH_MODULE_SFF_8079_LEN);
+	if (ret)
+		goto out;
+
+	sff8472_show_all(buf);
+out:
+	free(buf);
 
-	return 0;
+	return ret;
 }
-- 
2.35.1

Re: [PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[Michal Kubecek]

[-- Attachment #1: Type: text/plain, Size: 799 bytes --]

On Thu, Jun 16, 2022 at 05:50:09PM +0200, Ivan Vecera wrote:
> Commit 5b64c66f58d ("ethtool: Add netlink handler for
> getmodule\r (-m)") provided a netlink variant for getmodule
> but also introduced a regression as netlink output is different
> from ioctl output that provides information from A2h page
> via sff8472_show_all().
> 
> To fix this the netlink path should check a presence of A2h page
> by value of bit 6 in byte 92 of page A0h and if it is set then
> get A2h page and call sff8472_show_all().
> 
> Fixes: 5b64c66f58d ("ethtool: Add netlink handler for getmodule\r (-m)")

Looks like the leading "2" in commit id got lost and "^M" got into the
subject somehow. AFAICS this should be

  Fixes: 25b64c66f58d ("ethtool: Add netlink handler for getmodule (-m)")

Michal

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[Ido Schimmel]

On Thu, Jun 16, 2022 at 05:50:09PM +0200, Ivan Vecera wrote:
> Commit 5b64c66f58d ("ethtool: Add netlink handler for
> getmodule\r (-m)") provided a netlink variant for getmodule
> but also introduced a regression as netlink output is different
> from ioctl output that provides information from A2h page
> via sff8472_show_all().
> 
> To fix this the netlink path should check a presence of A2h page
> by value of bit 6 in byte 92 of page A0h and if it is set then
> get A2h page and call sff8472_show_all().
> 
> Fixes: 5b64c66f58d ("ethtool: Add netlink handler for getmodule\r (-m)")
> Tested-by: Daniel Juarez <djuarezg@cern.ch>
> Co-authored-by: Ido Schimmel <idosch@nvidia.com>
> Signed-off-by: Ivan Vecera <ivecera@redhat.com>

With Michal's comment:

Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Tested-by: Ido Schimmel <idosch@nvidia.com>

Tested with both netlink and ioctl, but only with SFP modules that lack
diagnostic information, as I don't have these at hand.

Michal, note that for ethtool-next we plan to get rid of sfpdiag.c and
fold it into sfpid.c, so that the latter will be able to handle both
SFF-8079 and SFF-8472 using the same memory map. We felt that it's too
big of a change for the main branch.

Re: [PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[Ivan Vecera]

On Thu, 16 Jun 2022 18:19:45 +0200
Michal Kubecek <mkubecek@suse.cz> wrote:

> On Thu, Jun 16, 2022 at 05:50:09PM +0200, Ivan Vecera wrote:
> > Commit 5b64c66f58d ("ethtool: Add netlink handler for
> > getmodule\r (-m)") provided a netlink variant for getmodule
> > but also introduced a regression as netlink output is different
> > from ioctl output that provides information from A2h page
> > via sff8472_show_all().
> > 
> > To fix this the netlink path should check a presence of A2h page
> > by value of bit 6 in byte 92 of page A0h and if it is set then
> > get A2h page and call sff8472_show_all().
> > 
> > Fixes: 5b64c66f58d ("ethtool: Add netlink handler for getmodule\r (-m)")  
> 
> Looks like the leading "2" in commit id got lost and "^M" got into the
> subject somehow. AFAICS this should be
> 
>   Fixes: 25b64c66f58d ("ethtool: Add netlink handler for getmodule (-m)")
> 
> Michal

Will send v2

Ivan

Re: [PATCH ethtool] sff-8079/8472: Fix missing sff-8472 output in netlink path

[Ivan Vecera]

On Thu, 16 Jun 2022 20:34:15 +0300
Ido Schimmel <idosch@nvidia.com> wrote:

> On Thu, Jun 16, 2022 at 05:50:09PM +0200, Ivan Vecera wrote:
> > Commit 5b64c66f58d ("ethtool: Add netlink handler for
> > getmodule  
>  (-m)") provided a netlink variant for getmodule
> > but also introduced a regression as netlink output is different
> > from ioctl output that provides information from A2h page
> > via sff8472_show_all().
> > 
> > To fix this the netlink path should check a presence of A2h page
> > by value of bit 6 in byte 92 of page A0h and if it is set then
> > get A2h page and call sff8472_show_all().
> > 
> > Fixes: 5b64c66f58d ("ethtool: Add netlink handler for getmodule  
>  (-m)")
> > Tested-by: Daniel Juarez <djuarezg@cern.ch>
> > Co-authored-by: Ido Schimmel <idosch@nvidia.com>
> > Signed-off-by: Ivan Vecera <ivecera@redhat.com>  
> 
> With Michal's comment:
> 
> Reviewed-by: Ido Schimmel <idosch@nvidia.com>
> Tested-by: Ido Schimmel <idosch@nvidia.com>
> 
> Tested with both netlink and ioctl, but only with SFP modules that lack
> diagnostic information, as I don't have these at hand.

I have tested this with SFP modules with diag info...
(together with kernel fix: https://patchwork.kernel.org/project/netdevbpf/patch/20220616160856.3623273-1-ivecera@redhat.com/)

> Michal, note that for ethtool-next we plan to get rid of sfpdiag.c and
> fold it into sfpid.c, so that the latter will be able to handle both
> SFF-8079 and SFF-8472 using the same memory map. We felt that it's too
> big of a change for the main branch.
+1

Ivan