| Message ID | 1580745415-19744-1-git-send-email-devesh.sharma@broadcom.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v4] libibverbs: display gid type in ibv_devinfo | expand |
On Mon, Feb 03, 2020 at 10:56:55AM -0500, Devesh Sharma wrote: > It becomes difficult to make out from the output of ibv_devinfo > if a particular gid index is RoCE v2 or not. > > Adding a string to the output of ibv_devinfo -v to display the > gid type at the end of gid. > > The output would look something like below: > $ ibv_devinfo -v -d bnxt_re2 > hca_id: bnxt_re2 > transport: InfiniBand (0) > fw_ver: 216.0.220.0 > node_guid: b226:28ff:fed3:b0f0 > sys_image_guid: b226:28ff:fed3:b0f0 > . > . > . > . > phys_state: LINK_UP (5) > GID[ 0]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, IB/RoCE v1 > GID[ 1]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, RoCE v2 > GID[ 2]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, IB/RoCE v1 > GID[ 3]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, RoCE v2 I think you should display the RoCEv2 GID in IPv6 notation, since it isn't really a GID anyhmore. The IPv6 notation should automatically show the IPv4 dotted quad Jason
On Mon, Feb 3, 2020 at 9:38 PM Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Mon, Feb 03, 2020 at 10:56:55AM -0500, Devesh Sharma wrote: > > It becomes difficult to make out from the output of ibv_devinfo > > if a particular gid index is RoCE v2 or not. > > > > Adding a string to the output of ibv_devinfo -v to display the > > gid type at the end of gid. > > > > The output would look something like below: > > $ ibv_devinfo -v -d bnxt_re2 > > hca_id: bnxt_re2 > > transport: InfiniBand (0) > > fw_ver: 216.0.220.0 > > node_guid: b226:28ff:fed3:b0f0 > > sys_image_guid: b226:28ff:fed3:b0f0 > > . > > . > > . > > . > > phys_state: LINK_UP (5) > > GID[ 0]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, IB/RoCE v1 > > GID[ 1]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, RoCE v2 > > GID[ 2]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, IB/RoCE v1 > > GID[ 3]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, RoCE v2 > > I think you should display the RoCEv2 GID in IPv6 notation, since it > isn't really a GID anyhmore. The IPv6 notation should automatically > show the IPv4 dotted quad There are many format specifiers, which one are you indicating? are those supported in printf()? > > Jason
On Mon, Feb 03, 2020 at 09:57:39PM +0530, Devesh Sharma wrote: > On Mon, Feb 3, 2020 at 9:38 PM Jason Gunthorpe <jgg@ziepe.ca> wrote: > > > > On Mon, Feb 03, 2020 at 10:56:55AM -0500, Devesh Sharma wrote: > > > It becomes difficult to make out from the output of ibv_devinfo > > > if a particular gid index is RoCE v2 or not. > > > > > > Adding a string to the output of ibv_devinfo -v to display the > > > gid type at the end of gid. > > > > > > The output would look something like below: > > > $ ibv_devinfo -v -d bnxt_re2 > > > hca_id: bnxt_re2 > > > transport: InfiniBand (0) > > > fw_ver: 216.0.220.0 > > > node_guid: b226:28ff:fed3:b0f0 > > > sys_image_guid: b226:28ff:fed3:b0f0 > > > . > > > . > > > . > > > . > > > phys_state: LINK_UP (5) > > > GID[ 0]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, IB/RoCE v1 > > > GID[ 1]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, RoCE v2 > > > GID[ 2]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, IB/RoCE v1 > > > GID[ 3]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, RoCE v2 > > > > I think you should display the RoCEv2 GID in IPv6 notation, since it > > isn't really a GID anyhmore. The IPv6 notation should automatically > > show the IPv4 dotted quad > > There are many format specifiers, which one are you indicating? are > those supported in printf()? inet_ntop(AF_INET6) Jason
On Mon, Feb 3, 2020 at 10:20 PM Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Mon, Feb 03, 2020 at 09:57:39PM +0530, Devesh Sharma wrote: > > On Mon, Feb 3, 2020 at 9:38 PM Jason Gunthorpe <jgg@ziepe.ca> wrote: > > > > > > On Mon, Feb 03, 2020 at 10:56:55AM -0500, Devesh Sharma wrote: > > > > It becomes difficult to make out from the output of ibv_devinfo > > > > if a particular gid index is RoCE v2 or not. > > > > > > > > Adding a string to the output of ibv_devinfo -v to display the > > > > gid type at the end of gid. > > > > > > > > The output would look something like below: > > > > $ ibv_devinfo -v -d bnxt_re2 > > > > hca_id: bnxt_re2 > > > > transport: InfiniBand (0) > > > > fw_ver: 216.0.220.0 > > > > node_guid: b226:28ff:fed3:b0f0 > > > > sys_image_guid: b226:28ff:fed3:b0f0 > > > > . > > > > . > > > > . > > > > . > > > > phys_state: LINK_UP (5) > > > > GID[ 0]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, IB/RoCE v1 > > > > GID[ 1]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, RoCE v2 > > > > GID[ 2]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, IB/RoCE v1 > > > > GID[ 3]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, RoCE v2 > > > > > > I think you should display the RoCEv2 GID in IPv6 notation, since it > > > isn't really a GID anyhmore. The IPv6 notation should automatically > > > show the IPv4 dotted quad > > > > There are many format specifiers, which one are you indicating? are > > those supported in printf()? > > inet_ntop(AF_INET6) Done! > > Jason
On 03/02/2020 17:56, Devesh Sharma wrote: > diff --git a/libibverbs/driver.h b/libibverbs/driver.h > index a0e6f89..fc0699d 100644 > --- a/libibverbs/driver.h > +++ b/libibverbs/driver.h > @@ -84,6 +84,7 @@ enum verbs_qp_mask { > enum ibv_gid_type { > IBV_GID_TYPE_IB_ROCE_V1, > IBV_GID_TYPE_ROCE_V2, > + IBV_GID_TYPE_INVALID > }; I don't think that's right. You're adding a new enum value to libibverbs, but it's not really used/implemented there. If devinfo needs an invalid GID value, make it local to that program.
On Mon, Feb 3, 2020 at 11:48 PM Gal Pressman <galpress@amazon.com> wrote: > > On 03/02/2020 17:56, Devesh Sharma wrote: > > diff --git a/libibverbs/driver.h b/libibverbs/driver.h > > index a0e6f89..fc0699d 100644 > > --- a/libibverbs/driver.h > > +++ b/libibverbs/driver.h > > @@ -84,6 +84,7 @@ enum verbs_qp_mask { > > enum ibv_gid_type { > > IBV_GID_TYPE_IB_ROCE_V1, > > IBV_GID_TYPE_ROCE_V2, > > + IBV_GID_TYPE_INVALID > > }; > > I don't think that's right. > You're adding a new enum value to libibverbs, but it's not really > used/implemented there. > If devinfo needs an invalid GID value, make it local to that program. the enum can be used by other applications too, if those are yet to be coded in future. I thought its a good practice to put things at one place once for all.
On Mon, Feb 03, 2020 at 11:53:47PM +0530, Devesh Sharma wrote: > On Mon, Feb 3, 2020 at 11:48 PM Gal Pressman <galpress@amazon.com> wrote: > > > > On 03/02/2020 17:56, Devesh Sharma wrote: > > > diff --git a/libibverbs/driver.h b/libibverbs/driver.h > > > index a0e6f89..fc0699d 100644 > > > --- a/libibverbs/driver.h > > > +++ b/libibverbs/driver.h > > > @@ -84,6 +84,7 @@ enum verbs_qp_mask { > > > enum ibv_gid_type { > > > IBV_GID_TYPE_IB_ROCE_V1, > > > IBV_GID_TYPE_ROCE_V2, > > > + IBV_GID_TYPE_INVALID > > > }; > > > > I don't think that's right. > > You're adding a new enum value to libibverbs, but it's not really > > used/implemented there. > > If devinfo needs an invalid GID value, make it local to that program. > > the enum can be used by other applications too, if those are yet to be > coded in future. I thought its a good practice to put things at one > place once for all. Yes, as long as IBV_GID_TYPE_INVALID can be returned by ibv_query_gid_type(), but it doesn't. Thanks
diff --git a/libibverbs/driver.h b/libibverbs/driver.h index a0e6f89..fc0699d 100644 --- a/libibverbs/driver.h +++ b/libibverbs/driver.h @@ -84,6 +84,7 @@ enum verbs_qp_mask { enum ibv_gid_type { IBV_GID_TYPE_IB_ROCE_V1, IBV_GID_TYPE_ROCE_V2, + IBV_GID_TYPE_INVALID }; enum ibv_mr_type { diff --git a/libibverbs/examples/devinfo.c b/libibverbs/examples/devinfo.c index bf53eac..5bf1a39 100644 --- a/libibverbs/examples/devinfo.c +++ b/libibverbs/examples/devinfo.c @@ -162,8 +162,18 @@ static const char *vl_str(uint8_t vl_num) } } +static const char *gid_type_str(enum ibv_gid_type type) +{ + switch (type) { + case IBV_GID_TYPE_IB_ROCE_V1: return "IB/RoCE v1"; + case IBV_GID_TYPE_ROCE_V2: return "RoCE v2"; + default: return "Invalid gid type"; + } +} + static int print_all_port_gids(struct ibv_context *ctx, uint8_t port_num, int tbl_len) { + enum ibv_gid_type type; union ibv_gid gid; int rc = 0; int i; @@ -175,8 +185,15 @@ static int print_all_port_gids(struct ibv_context *ctx, uint8_t port_num, int tb port_num, i); return rc; } + + rc = ibv_query_gid_type(ctx, port_num, i, &type); + if (rc) { + rc = 0; + type = IBV_GID_TYPE_INVALID; + } + if (!null_gid(&gid)) - printf("\t\t\tGID[%3d]:\t\t%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x\n", + printf("\t\t\tGID[%3d]:\t\t%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x:%02x%02x, %s\n", i, gid.raw[ 0], gid.raw[ 1], gid.raw[ 2], gid.raw[ 3], @@ -185,7 +202,8 @@ static int print_all_port_gids(struct ibv_context *ctx, uint8_t port_num, int tb gid.raw[ 8], gid.raw[ 9], gid.raw[10], gid.raw[11], gid.raw[12], gid.raw[13], - gid.raw[14], gid.raw[15]); + gid.raw[14], gid.raw[15], + gid_type_str(type)); } return rc; }
It becomes difficult to make out from the output of ibv_devinfo if a particular gid index is RoCE v2 or not. Adding a string to the output of ibv_devinfo -v to display the gid type at the end of gid. The output would look something like below: $ ibv_devinfo -v -d bnxt_re2 hca_id: bnxt_re2 transport: InfiniBand (0) fw_ver: 216.0.220.0 node_guid: b226:28ff:fed3:b0f0 sys_image_guid: b226:28ff:fed3:b0f0 . . . . phys_state: LINK_UP (5) GID[ 0]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, IB/RoCE v1 GID[ 1]: fe80:0000:0000:0000:b226:28ff:fed3:b0f0, RoCE v2 GID[ 2]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, IB/RoCE v1 GID[ 3]: 0000:0000:0000:0000:0000:ffff:c0aa:0165, RoCE v2 $ $ Reviewed-by: Leon Romanovsky <leon@kernel.org> Reviewed-by: Gal Pressman <galpress@amazon.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com> --- libibverbs/driver.h | 1 + libibverbs/examples/devinfo.c | 22 ++++++++++++++++++++-- 2 files changed, 21 insertions(+), 2 deletions(-)