Fix reading capability bit from device capabilities for hcall locking.
Signed-off-by: Alexander Schmidt <[email protected]>
---
Hi Roland, this applies against 2.6.35-rc2 and we would like to get
it into your -next tree, thanks!
drivers/infiniband/hw/ehca/ehca_main.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
--- linux-2.6.35-rc2.orig/drivers/infiniband/hw/ehca/ehca_main.c
+++ linux-2.6.35-rc2/drivers/infiniband/hw/ehca/ehca_main.c
@@ -360,7 +360,8 @@ static int ehca_sense_attributes(struct
* a firmware property, so it's valid across all adapters
*/
if (ehca_lock_hcalls == -1)
- ehca_lock_hcalls = !(shca->hca_cap & HCA_CAP_H_ALLOC_RES_SYNC);
+ ehca_lock_hcalls = !EHCA_BMASK_GET(HCA_CAP_H_ALLOC_RES_SYNC,
+ shca->hca_cap);
/* translate supported MR page sizes; always support 4K */
shca->hca_cap_mr_pgsize = EHCA_PAGESIZE;
What is the impact of this problem? ie is this something we need to fix
for 2.6.35? I suppose not, since the current code has been there a long time.
--
Roland Dreier <[email protected]> || For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/index.html
On Fri, 11 Jun 2010 09:02:43 -0700
Roland Dreier <[email protected]> wrote:
> What is the impact of this problem? ie is this something we need to fix
> for 2.6.35? I suppose not, since the current code has been there a long time.
Right, we are okay with integrating this into 2.6.36, thanks!