On 05/18/2017 01:59 PM, Ferruh Yigit wrote:
On 5/17/2017 1:25 PM, Andrew Rybchenko wrote:
Required to be able to use logging in the secondary process
where Ethernet device pointer stored in sfc_adapter is invalid.
Signed-off-by: Andrew Rybchenko <arybche...@solarflare.com>
Reviewed-by: Andy Moreton <amore...@solarflare.com>
<...>
diff --git a/drivers/net/sfc/sfc_log.h b/drivers/net/sfc/sfc_log.h
index d0f8921..6c43925 100644
--- a/drivers/net/sfc/sfc_log.h
+++ b/drivers/net/sfc/sfc_log.h
@@ -35,18 +35,16 @@
/* Log PMD message, automatically add prefix and \n */
#define SFC_LOG(sa, level, ...) \
do { \
- const struct rte_eth_dev *_dev = (sa)->eth_dev; \
- const struct rte_pci_device *_pci_dev = \
- RTE_ETH_DEV_TO_PCI(_dev); \
+ const struct sfc_adapter *_sa = (sa); \
Getting following build error with clang and icc [1]. I guess that is
because "_sa" declared on both sfc_log_init() and in the macro that
function uses (SFC_LOG).
Thanks, I'll fix it in v2.
[1]
.../drivers/net/sfc/sfc_filter.c:121:2:
error: variable '_sa' is uninitialized when used within its own
initialization [-Werror,-Wuninitialized]
sfc_log_init(sa, "failed %d", rc);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.../drivers/net/sfc/sfc_log.h:68:12: note: expanded from macro
'sfc_log_init'
SFC_LOG(_sa, INFO, \
~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.../drivers/net/sfc/sfc_log.h:38:36: note: expanded from macro 'SFC_LOG'
const struct sfc_adapter *_sa = (sa); \
~~~ ^~
