[PATCH V2] ocxl: Fix access to the AFU Descriptor Data

The AFU Information DVSEC capability is a means to extract common,
general information about all of the AFUs associated with a Function
independent of the specific functionality that each AFU provides.

This patch fixes the access to the AFU Descriptor Data indexed by the
AFU Info Index field.

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: stable <redacted>     # 4.16
Signed-off-by: Christophe Lombard <redacted>
---
Changelog[v2]
 - Rebase to latest upstream.
 - Use pci_write_config_byte instead of pci_write_config_word
---
 drivers/misc/ocxl/config.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/misc/ocxl/config.c b/drivers/misc/ocxl/config.c
index 2e30de9..57a6bb1 100644
--- a/drivers/misc/ocxl/config.c
+++ b/drivers/misc/ocxl/config.c

@@ -280,7 +280,9 @@ int ocxl_config_check_afu_index(struct pci_dev *dev,
 	u32 val;
 	int rc, templ_major, templ_minor, len;
 
-	pci_write_config_word(dev, fn->dvsec_afu_info_pos, afu_idx);
+	pci_write_config_byte(dev,
+			fn->dvsec_afu_info_pos + OCXL_DVSEC_AFU_INFO_AFU_IDX,
+			afu_idx);
 	rc = read_afu_info(dev, fn, OCXL_DVSEC_TEMPL_VERSION, &val);
 	if (rc)
 		return rc;

-- 
2.7.4

Le 13/08/2018 à 16:09, Christophe Lombard a écrit :

The AFU Information DVSEC capability is a means to extract common,
general information about all of the AFUs associated with a Function
independent of the specific functionality that each AFU provides.

This patch fixes the access to the AFU Descriptor Data indexed by the
AFU Info Index field.

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: stable <redacted>     # 4.16
Signed-off-by: Christophe Lombard <redacted>
---

Thanks!
Acked-by: Frederic Barrat <redacted>

On 14/08/18 00:09, Christophe Lombard wrote:

The AFU Information DVSEC capability is a means to extract common,
general information about all of the AFUs associated with a Function
independent of the specific functionality that each AFU provides.

This patch fixes the access to the AFU Descriptor Data indexed by the
AFU Info Index field.

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: stable <redacted>     # 4.16
Signed-off-by: Christophe Lombard <redacted>

Thanks

Acked-by: Andrew Donnellan <redacted>

-- 
Andrew Donnellan              OzLabs, ADL Canberra
andrew.donnellan@au1.ibm.com  IBM Australia Limited

Hi Christophe,

The patch looks fine, just a nit about the change log:

Christophe Lombard [off-list ref] writes:

The AFU Information DVSEC capability is a means to extract common,
general information about all of the AFUs associated with a Function
independent of the specific functionality that each AFU provides.

This patch fixes the access to the AFU Descriptor Data indexed by the
AFU Info Index field.

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: stable <redacted>     # 4.16
Signed-off-by: Christophe Lombard <redacted>

When fixing a bug it's always good to describe how the bug manifests.
ie. in this case we are clearly writing to the wrong location in config
space, but what is the consequence of that? Does it kill the device, or
just fails to initialise something correctly? How could I tell if I'm
hitting this bug currently? How would I tell if the fix is applied
correctly?

cheers

Le 14/08/2018 à 05:26, Michael Ellerman a écrit :

Hi Christophe,

The patch looks fine, just a nit about the change log:

Christophe Lombard [off-list ref] writes:

quoted

The AFU Information DVSEC capability is a means to extract common,
general information about all of the AFUs associated with a Function
independent of the specific functionality that each AFU provides.

This patch fixes the access to the AFU Descriptor Data indexed by the
AFU Info Index field.

quoted

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: stable <redacted>     # 4.16
Signed-off-by: Christophe Lombard <redacted>

When fixing a bug it's always good to describe how the bug manifests.
ie. in this case we are clearly writing to the wrong location in config
space, but what is the consequence of that? Does it kill the device, or
just fails to initialise something correctly? How could I tell if I'm
hitting this bug currently? How would I tell if the fix is applied
correctly?

You are right, let me send a new version.

Thanks

cheers

quoted

---
Changelog[v2]
  - Rebase to latest upstream.
  - Use pci_write_config_byte instead of pci_write_config_word
---
  drivers/misc/ocxl/config.c | 4 +++-
  1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/misc/ocxl/config.c b/drivers/misc/ocxl/config.c
index 2e30de9..57a6bb1 100644
--- a/drivers/misc/ocxl/config.c
+++ b/drivers/misc/ocxl/config.c

@@ -280,7 +280,9 @@ int ocxl_config_check_afu_index(struct pci_dev *dev,
  	u32 val;
  	int rc, templ_major, templ_minor, len;
  
-	pci_write_config_word(dev, fn->dvsec_afu_info_pos, afu_idx);
+	pci_write_config_byte(dev,
+			fn->dvsec_afu_info_pos + OCXL_DVSEC_AFU_INFO_AFU_IDX,
+			afu_idx);
  	rc = read_afu_info(dev, fn, OCXL_DVSEC_TEMPL_VERSION, &val);
  	if (rc)
  		return rc;

-- 
2.7.4