On Fri, Jul 06, 2012 at 03:23:54PM +0300, Dan Carpenter wrote:
> On Fri, Jul 06, 2012 at 11:58:14AM +0000, Marciniszyn, Mike wrote:
> > I'm not seeing this on either of my build test systems.
> >
> > Can you provide your .config file and gcc version?
> >
>
> This isn't a GCC warning, it's a sparse warning. It's saying that
> your Sparse rcu anotations aren't correct, probably.
Yeah these warnings can be reproduced with
1) CONFIG_SPARSE_RCU_POINTER=y
2) make C=1 drivers/infiniband/hw/qib/qib_keys.o
Paul has a nice article on the details
The RCU API, 2010 Edition
https://lwn.net/Articles/418853/
For the entitled warning, the fix should be
diff --git a/drivers/infiniband/hw/qib/qib_keys.c b/drivers/infiniband/hw/qib/qib_keys.c
index 970165b..3c8c27d 100644
--- a/drivers/infiniband/hw/qib/qib_keys.c
+++ b/drivers/infiniband/hw/qib/qib_keys.c
@@ -349,7 +349,7 @@ int qib_fast_reg_mr(struct qib_qp *qp, struct ib_send_wr *wr)
if (pd->user || rkey == 0)
goto bail;
- mr = rkt->table[(rkey >> (32 - ib_qib_lkey_table_size))];
+ mr = rcu_dereference(rkt->table[(rkey >> (32 - ib_qib_lkey_table_size))]);
if (unlikely(mr == NULL || qp->ibqp.pd != mr->pd))
goto bail;
diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h
index 85751fd..1293133 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.h
+++ b/drivers/infiniband/hw/qib/qib_verbs.h
@@ -650,7 +650,7 @@ struct qib_lkey_table {
u32 next; /* next unused index (speeds search) */
u32 gen; /* generation count */
u32 max; /* size of the table */
- struct qib_mregion **table;
+ struct qib_mregion __rcu **table;
};
struct qib_opcode_stats {
@@ -732,7 +732,7 @@ struct qib_ibdev {
struct list_head pending_mmaps;
spinlock_t mmap_offset_lock; /* protect mmap_offset */
u32 mmap_offset;
- struct qib_mregion *dma_mr;
+ struct qib_mregion __rcu *dma_mr;
/* QP numbers are shared by all IB ports */
struct qib_qpn_table qpn_table;
> I've added Fengguang's other email address in case wfg isn't fixed
> yet.
Thank you! The wfg address seems to work fine except for the previous
abnormal case :)
Thanks,
Fengguang
--
To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
