ehca: ehca_av.c use constant for max mtu
PATCH 4/4 ehca: ehca_av.c use constant for max mtu
Hello Roland!
This is a patch for ehca, mainly a code change to adhere to kernel coding style. It defines and uses a constant EHCA_MAX_MTU instead hardcoded value.
Thanks!
Nam
Signed-off-by Hoang-Nam Nguyen <[email protected]>
---
ehca_av.c | 5 ++--- hipz_hw.h | 2 ++ 2 files changed, 4 insertions(+), 3 deletions(-)
drivers/infiniband/hw/ehca/ehca_av.c +2 -3
@@ -118,8 +118,7 @@ struct ib_ah *ehca_create_ah(struct ib_p }memcpy (&av->av.grh.word_1, &gid,sizeof (gid)); } -/* for the time being we use a hard coded PMTU of 2048 Bytes */ - av->av.pmtu =4 ; + av->av.pmtu =EHCA_MAX_MTU ;/* dgid comes in grh.word_3 */ memcpy(&av->av.grh.word_3, &ah_attr->grh.dgid,
@@ -193,7 +192,7 @@ int ehca_modify_ah(struct ib_ah *ah, strmemcpy (&new_ehca_av.grh.word_1, &gid,sizeof (gid)); } - new_ehca_av.pmtu =4 ;/* see also comment in create_ah() */ + new_ehca_av.pmtu =EHCA_MAX_MTU ;memcpy (&new_ehca_av.grh.word_3, &ah_attr->grh.dgid,sizeof (ah_attr->grh.dgid));
drivers/infiniband/hw/ehca/hipz_hw.h +2 -0
@@ -45,6 +45,8 @@ #define __HIPZ_HW_H__#include "ehca_tools.h" +#define EHCA_MAX_MTU 4 +/* QP Table Entry Memory Map */ struct hipz_qptemm {u64 qpx_hcr;
drivers/infiniband/hw/ehca/ehca_av.c +2 -3
@@ -118,8 +118,7 @@ struct ib_ah *ehca_create_ah(struct ib_p }memcpy (&av->av.grh.word_1, &gid,sizeof (gid)); } -/* for the time being we use a hard coded PMTU of 2048 Bytes */ - av->av.pmtu =4 ; + av->av.pmtu =EHCA_MAX_MTU ;/* dgid comes in grh.word_3 */ memcpy(&av->av.grh.word_3, &ah_attr->grh.dgid,
@@ -193,7 +192,7 @@ int ehca_modify_ah(struct ib_ah *ah, strmemcpy (&new_ehca_av.grh.word_1, &gid,sizeof (gid)); } - new_ehca_av.pmtu =4 ;/* see also comment in create_ah() */ + new_ehca_av.pmtu =EHCA_MAX_MTU ;memcpy (&new_ehca_av.grh.word_3, &ah_attr->grh.dgid,sizeof (ah_attr->grh.dgid));
drivers/infiniband/hw/ehca/hipz_hw.h +2 -0
@@ -45,6 +45,8 @@ #define __HIPZ_HW_H__#include "ehca_tools.h" +#define EHCA_MAX_MTU 4 +/* QP Table Entry Memory Map */ struct hipz_qptemm {u64 qpx_hcr;
Can you fix your patch sending technique ? A mangled patch inline with a non-mangled one in attachment... that's a bit gross. Just get a proper one inline and be done with it. If your mailer can't be coerced into not damaging patches, then use another one for sending them.
Cheers,
Ben
Thanks, I've applied 1 through 4.