nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
PATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
This fix adds the "host_traddr" to the contructed string that is passed to the kernel.
--- fabrics.c | 5 +++++ 1 file changed, 5 insertions(+)
fabrics.c +5 -0
@@ -595,6 +595,11 @@ static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e)return -EINVAL ; p += len; + len =sprintf (p,",host_traddr=%s" , cfg.host_traddr); +if (len <0 ) +return -EINVAL ; + p+= len; + len =sprintf (p,",traddr=%s" , e->traddr);if (len <0 )return -EINVAL ;
Signature · 1 line
1.7.10.GIT
PATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Hi Darren,
Can you send an example for usage ?
I want to try this out.
thanks,
Max.
Quoted 25 lines · Duane Grigsby
This fix adds the "
host_traddr" to the contructed string that is passed to the kernel.Signed-off-by Darren Trapp <darren.trapp at cavium.com>--- fabrics.c | 5 +++++ 1 file changed, 5 insertions(+)fabrics.c +5
-0@@ -595,6 +595,11 @@ static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e)return -EINVAL ; p += len; + len =sprintf (p,",host_traddr=%s" , cfg.host_traddr); +if (len <0 ) +return -EINVAL ; + p+= len; + len =sprintf (p,",traddr=%s" , e->traddr);if (len <0 )return -EINVAL ;
Sagi Grimberg
maintainerPATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
In reply to Duane Grigsbyfabrics.c static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e) · line 602
Quoted 10 lines
fabrics.c +5
-0@@ -595,6 +595,11 @@ static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e)return -EINVAL ; p += len; + len =sprintf (p,",host_traddr=%s" , cfg.host_traddr); +if (len <0 ) +return -EINVAL ; + p+= len; +
I think this needs to be conditional on if provided at all.
PATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Quoted 6 lines · Sagi Grimberg
Quoted 6 lines
fabrics.c +0
-0@@ -595,6 +595,11 @@ static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e)return -EINVAL ; p += len;
Quoted 8 lines
fabrics.c static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e) · line 602
Quoted 5 lines
fabrics.c +5
-0+ len =sprintf (p,",host_traddr=%s" , cfg.host_traddr); +if (len <0 ) +return -EINVAL ; + p+= len; +I think this needs to be conditional on if provided at all.
Also, currently this is implemented for FC but I'll want to add an option for rdma too.
so we'll be able to use it like that:
nvme connect -t rdma -n testsubsystem -a <target_ip> -w <host_ip> -s 1023
in case we want to choose the interface for rdma_resolve_addr. I'll put it on my plate.
Max.
Quoted 4 lines
_______________________________________________
Linux-nvme mailing list
Linux-nvme at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
Sagi Grimberg
maintainerPATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Quoted 9 lines
Also, currently this is implemented for FC but I'll want to add an option for rdma too.
so we'll be able to use it like that:nvme connect -t rdma -n testsubsystem -a <
target_ip> -w <host_ip> -s 1023in case we want to choose the interface for
rdma_resolve_addr. I'll put it on my plate.
That would be nice.
PATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
In reply to Sagi GrimbergFC requires the host_traddr to make the connection. If it wasn?t provided, the fc transport wouldn?t have been able to connect to the discovery controller and request the disc page. Hence, we wouldn?t be here parsing the response.
Darren Trapp
Senior Director - Engineering
darren.trapp at cavium.com <mailto:darren.trapp at caviumnetworks.com> T 949-389-6273
M 949-212-8576
26650 Aliso Viejo Pkwy, Aliso Viejo, CA 92656 www.qlogic.com <http://www.qlogic.com/>
<http://www.qlogic.com/>
fabrics.c static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e) · line 602
Quoted 9 lines · Linux-nvme on behalf of Sagi Grimberg
fabrics.c +5
-0@@ -595,6 +595,11 @@ static int connect_ctrl(struct nvmf_disc_rsp_page_entry *e)return -EINVAL ; p += len; + len =sprintf (p,",host_traddr=%s" , cfg.host_traddr); +if (len <0 ) +return -EINVAL ; + p+= len; +
I think this needs to be conditional on if provided at all.
_______________________________________________
Linux-nvme mailing list
Linux-nvme at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
Sagi Grimberg
maintainerPATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Quoted 2 lines
FC requires the
host_traddrto make the connection. If it wasn?t provided, the fc transport wouldn?t have been able to connect to the discovery controller and request the disc page. Hence, we wouldn?t be here parsing the response.
Did you test this patch with nvme-loop?
PATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
I have not tried the fc target loopback. But since the discovery response is processed above the fc transport level it should return the same information for the host to process. (ie transport type FC etc).
The connection to the target, from the host perspective, would be the same.
Darren Trapp
Senior Director - Engineering
darren.trapp at cavium.com <mailto:darren.trapp at caviumnetworks.com> T 949-389-6273
M 949-212-8576
26650 Aliso Viejo Pkwy, Aliso Viejo, CA 92656 www.qlogic.com <http://www.qlogic.com/>
<http://www.qlogic.com/>
Quoted 1 line · Sagi Grimberg
FC requires the
host_traddrto make the connection. If it wasn?t provided, the fc transport wouldn?t have been able to connect to the discovery controller and request the disc page. Hence, we wouldn?t be here parsing the response.
Did you test this patch with nvme-loop?
Sagi Grimberg
maintainerPATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Quoted 2 lines
I have not tried the fc target loopback.
After looking at the code again I now recall that the connection string construction is not shared by transports. My concern was that other transports (rdma) won't blow up...
So this looks good,
Keith Busch
AppliedPATCH 1/1 nvme-cli: fabrics: Fix to pass "host_traddr" to the kernel driver.
Quoted 11 lines · Sagi Grimberg
Quoted 2 lines
I have not tried the fc target loopback.
After looking at the code again I now recall that the connection string construction is not shared by transports. My concern was that other transports (rdma) won't blow up...
So this looks good,
Reviewed-by Sagi Grimberg <sagi at grimberg.me>
Thanks, applied the patch with the review.