On Fri, Mar 24, 2017 at 3:31 PM, Johannes Thumshirn [off-list ref] wr=
ote:

On Fri, Mar 24, 2017 at 01:54:04PM +0100, Jinpu Wang wrote:

quoted

quoted

quoted

+
+#define XX(a) case (a): return #a

please no macros with retun in them and XX isn't quite too descriptive=

 as

quoted

quoted

well.

[...]

quoted

+static inline const char *ib_wc_opcode_str(enum ib_wc_opcode opcode)
+{
+     switch (opcode) {
+     XX(IB_WC_SEND);
+     XX(IB_WC_RDMA_WRITE);
+     XX(IB_WC_RDMA_READ);
+     XX(IB_WC_COMP_SWAP);
+     XX(IB_WC_FETCH_ADD);
+     /* recv-side); inbound completion */
+     XX(IB_WC_RECV);
+     XX(IB_WC_RECV_RDMA_WITH_IMM);
+     default: return "IB_WC_OPCODE_UNKNOWN";
+     }
+}

How about:

struct {
        char *name;
        enum ib_wc_opcode opcode;
} ib_wc_opcode_table[] =3D {
        { stringyfy(IB_WC_SEND), IB_WC_SEND },
        { stringyfy(IB_WC_RDMA_WRITE), IB_WC_RDMA_WRITE },
        { stringyfy(IB_WC_RDMA_READ ), IB_WC_RDMA_READ }
        { stringyfy(IB_WC_COMP_SWAP), IB_WC_COMP_SWAP },
        { stringyfy(IB_WC_FETCH_ADD), IB_WC_FETCH_ADD },
        { stringyfy(IB_WC_RECV), IB_WC_RECV },
        { stringyfy(IB_WC_RECV_RDMA_WITH_IMM), IB_WC_RECV_RDMA_WITH_IM=

M },

quoted

quoted

        { NULL, 0 },
};

static inline const char *ib_wc_opcode_str(enum ib_wc_opcode opcode)
{
        int i;

        for (i =3D 0; i < ARRAY_SIZE(ib_wc_opcode_table); i++)
                if (ib_wc_opcode_table[i].opcode =3D=3D opcode)
                        return ib_wc_opcode_table[i].name;

        return "IB_WC_OPCODE_UNKNOWN";
}

Looks nice, might be better to put it into ib_verbs.h?

Probably yes, as are your kvec functions for lib/iov_iter.c

Thanks, will do in next round!

[...]

quoted

quoted

What about resolving the kernel bug instead of making workarounds?

I tried to send a patch upsteam, but was rejected by Sean.
http://www.spinics.net/lists/linux-rdma/msg22381.html

I don't see a NACK in this thread.

From http://www.spinics.net/lists/linux-rdma/msg22410.html:
"The port space (which maps to the service ID) needs to be included as pa=

rt of

the check that determines the format of the private data, and not simply =

the

address family."

After such a state I would have expected to see a v2 of the patch with ab=

ove

comment addressed.

I might busy with other staff at that time, I will check again and
revisit the bug.

Byte,
        Johannes
--
Johannes Thumshirn                                          Storage
jthumshirn@suse.de                                +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 N=C3=BCrnberg
GF: Felix Imend=C3=B6rffer, Jane Smithard, Graham Norton
HRB 21284 (AG N=C3=BCrnberg)
Key fingerprint =3D EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850

Regards,
--=20
Jack Wang
Linux Kernel Developer

ProfitBricks GmbH
Greifswalder Str. 207
D - 10405 Berlin

Tel:       +49 30 577 008  042
Fax:      +49 30 577 008 299
Email:    jinpu.wang@profitbricks.com
URL:      https://www.profitbricks.de

Sitz der Gesellschaft: Berlin
Registergericht: Amtsgericht Charlottenburg, HRB 125506 B
Gesch=C3=A4ftsf=C3=BChrer: Achim Weiss