On Sun, Jul 08, 2018 at 04:59:51PM +0300, Yuval Bason wrote:
> diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
> index b2a7ec4..a065a6a 100644
> +++ b/drivers/infiniband/hw/qedr/verbs.c
> @@ -65,6 +65,14 @@ static inline int qedr_ib_copy_to_udata(struct ib_udata *udata, void *src,
> return ib_copy_to_udata(udata, src, min_len);
> }
>
> +static inline int qedr_ib_copy_from_udata(void *dest, struct ib_udata *udata,
> + size_t len)
> +{
> + size_t min_len = min_t(size_t, len, udata->inlen);
> +
> + return ib_copy_from_udata(dest, udata, min_len);
> +}
Can we please not have these sorts of common things inlined in
drivers? Same for copy_to
This should also zero fill the trailing data.
> @@ -1471,7 +1566,10 @@ int qedr_destroy_srq(struct ib_srq *ibsrq)
> in_params.srq_id = srq->srq_id;
> dev->ops->rdma_destroy_srq(dev->rdma_ctx, &in_params);
>
> - qedr_free_srq_kernel_params(srq);
> + if (ibsrq->pd->uobject && ibsrq->pd->uobject->context)
What is the test for uobject->context doing? If uboject exists then
uobject->context is always valid.
> +struct qedr_create_srq_ureq {
> + /* user space virtual address of producer pair */
> + __u64 prod_pair_addr;
> +
> + /* user space virtual address of SRQ buffer */
> + __u64 srq_addr;
> +
> + /* length of SRQ buffer */
> + __u64 srq_len;
> +};
> +
> +struct qedr_create_srq_uresp {
> + __u16 srq_id;
> +};
Surprising this compiles in rdma-core..
Oh look, the rdma-core patch changes it:
+struct qedr_create_srq_uresp {
+ __u16 srq_id;
+ __u16 reserved0;
+ __u32 reserved1;
+};
Words fail me. :(
You must use the kernel-headers/update script in rdma-core to make any
changes to the kernel-headers directory.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
