Skip to contentSkip to conversation

RDMA/bnxt_re: Bug fixes

PATCH Applied Selvin Xavier 2 participants 10 messages Aug 19, 2021 – Aug 26, 2021

Subsystems (2)
lore
Share & tools
lore

Selvin Xavier

author maintainer
#0 lore raw

v2 cover RDMA/bnxt_re: Bug fixes

Some simple bug fixes

v1->v2:
Fix a build issue

Naresh Kumar PBS (1):
RDMA/bnxt_re: Add missing spin lock initialization

Selvin Xavier (2):
RDMA/bnxt_re: Disable atomic support on VFs
RDMA/bnxt_re: Fix query SRQ failure

 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 1 +
 drivers/infiniband/hw/bnxt_re/main.c     | 2 +-
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 2 ++
 3 files changed, 4 insertions(+), 1 deletion(-)
Signature · 1 line
2.5.5
author maintainer
#1 lore raw

PATCH v2 1/3 RDMA/bnxt_re: Disable atomic support on VFs

Atomics is not currently supported for VFs. Enabling only for PFs.

Fixes 35f5ace5dea4 ("RDMA/bnxt_re: Enable global atomic ops if platform supports")
---
 drivers/infiniband/hw/bnxt_re/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/infiniband/hw/bnxt_re/main.c +1 -1
index a8688a9..0b870be 100644
@@ -129,7 +129,7 @@ static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev, u8 wqe_mode)
 	rdev->rcfw.res = &rdev->qplib_res;
 
 	bnxt_re_set_drv_mode(rdev, wqe_mode);
-	if (bnxt_qplib_determine_atomics(en_dev->pdev))
+	if (!BNXT_VF(bp) && bnxt_qplib_determine_atomics(en_dev->pdev))
 		ibdev_info(&rdev->ibdev,
 			   "platform doesn't support global atomics.");
 	return 0;
Signature · 1 line
2.5.5
maintainer
#2 lore raw

Re: [PATCH rdma-rc v2 1/3] RDMA/bnxt_re: Disable atomic support on VFs

Quoted 2 lines · Selvin Xavier

Atomics is not currently supported for VFs. Enabling only for PFs.

Again, what is the user experience here? Does the device crash? QP blow up?? Silently not work?

Jason

author maintainer
#3 lore raw

Re: [PATCH rdma-rc v2 1/3] RDMA/bnxt_re: Disable atomic support on VFs

Quoted 8 lines · Jason Gunthorpe
Quoted 2 lines · Selvin Xavier

Atomics is not currently supported for VFs. Enabling only for PFs.

Again, what is the user experience here? Does the device crash? QP blow up?? Silently not work?

This is a host crash when pci_enable_atomic_ops_to_root is called with a VF pci device (pdev->bus->self is NULL). So, I want to avoid calling it for VF now.
crash> bt
PID: 4481 TASK: ffff89c6941b0000 CPU: 53 COMMAND: "bash"
#0 [ffff9a94817136d8] machine_kexec at ffffffffb90601a4
#1 [ffff9a9481713728] __crash_kexec at ffffffffb9190d5d
#2 [ffff9a94817137f0] crash_kexec at ffffffffb9191c4d
#3 [ffff9a9481713808] oops_end at ffffffffb9025cd6
#4 [ffff9a9481713828] page_fault_oops at ffffffffb906e417 #5 [ffff9a9481713888] exc_page_fault at ffffffffb9a0ad14 #6 [ffff9a94817138b0] asm_exc_page_fault at ffffffffb9c00ace

    [exception RIP: pcie_capability_read_dword+28]
    RIP: ffffffffb952fd5c  RSP: ffff9a9481713960  RFLAGS: 00010246
    RAX: 0000000000000001  RBX: ffff89c6b1096000  RCX: 0000000000000000
    RDX: ffff9a9481713990  RSI: 0000000000000024  RDI: 0000000000000000
    RBP: 0000000000000080   R8: 0000000000000008   R9: ffff89c64341a2f8
    R10: 0000000000000002  R11: 0000000000000000  R12: ffff89c648bab000
    R13: 0000000000000000  R14: 0000000000000000  R15: ffff89c648bab0c8
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018

#7 [ffff9a9481713988] pci_enable_atomic_ops_to_root at ffffffffb95359a6 #8 [ffff9a94817139c0] bnxt_qplib_determine_atomics at
ffffffffc08c1a33 [bnxt_re]
#9 [ffff9a94817139d0] bnxt_re_dev_init at ffffffffc08ba2d1 [bnxt_re]
#10 [ffff9a9481713a78] bnxt_re_netdev_event at ffffffffc08bab8f [bnxt_re] #11 [ffff9a9481713aa8] raw_notifier_call_chain at ffffffffb9102cbe #12 [ffff9a9481713ad0] register_netdevice at ffffffffb9803ff3 #13 [ffff9a9481713b08] register_netdev at ffffffffb980410a #14 [ffff9a9481713b18] bnxt_init_one at ffffffffc0349572 [bnxt_en] #15 [ffff9a9481713b70] local_pci_probe at ffffffffb953b92f #16 [ffff9a9481713ba0] pci_device_probe at ffffffffb953cf8f #17 [ffff9a9481713be8] really_probe at ffffffffb9659619
#18 [ffff9a9481713c08] __driver_probe_device at ffffffffb96598fb #19 [ffff9a9481713c28] driver_probe_device at ffffffffb965998f #20 [ffff9a9481713c48] __device_attach_driver at ffffffffb9659cd2 #21 [ffff9a9481713c70] bus_for_each_drv at ffffffffb9657307 #22 [ffff9a9481713ca8] __device_attach at ffffffffb96593e0 #23 [ffff9a9481713ce8] pci_bus_add_device at ffffffffb9530b7a #24 [ffff9a9481713d00] pci_iov_add_virtfn at ffffffffb955b1ca #25 [ffff9a9481713d40] sriov_enable at ffffffffb955b54b
#26 [ffff9a9481713d90] bnxt_sriov_configure at ffffffffc034d913 [bnxt_en] #27 [ffff9a9481713dd8] sriov_numvfs_store at ffffffffb955acb4 #28 [ffff9a9481713e10] kernfs_fop_write_iter at ffffffffb93f09ad #29 [ffff9a9481713e48] new_sync_write at ffffffffb933b82c
#30 [ffff9a9481713ed0] vfs_write at ffffffffb933db64
#31 [ffff9a9481713f00] ksys_write at ffffffffb933dd99
#32 [ffff9a9481713f38] do_syscall_64 at ffffffffb9a07897
#33 [ffff9a9481713f50] entry_SYSCALL_64_after_hwframe at ffffffffb9c0007c

    RIP: 00007f450602f648  RSP: 00007ffe880869e8  RFLAGS: 00000246
    RAX: ffffffffffffffda  RBX: 0000000000000002  RCX: 00007f450602f648
    RDX: 0000000000000002  RSI: 0000555c566c4a60  RDI: 0000000000000001
    RBP: 0000555c566c4a60   R8: 000000000000000a   R9: 00007f45060c2580
    R10: 000000000000000a  R11: 0000000000000246  R12: 00007f45063026e0
    R13: 0000000000000002  R14: 00007f45062fd880  R15: 0000000000000002
    ORIG_RAX: 0000000000000001  CS: 0033  SS: 002b

I will update patch with this stack trace information.

Quoted 1 line

Jason

author maintainer
#4 lore raw

PATCH v2 2/3 RDMA/bnxt_re: Add missing spin lock initialization

In reply to Selvin Xavier

From: Naresh Kumar PBS <[email protected]>

Add the missing initialization of srq lock.

Fixes 37cb11acf1f7 ("RDMA/bnxt_re: Add SRQ support for Broadcom adapters")
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 1 +
 1 file changed, 1 insertion(+)
drivers/infiniband/hw/bnxt_re/ib_verbs.c +1 -0
index 283b6b8..ea0054c 100644
@@ -1681,6 +1681,7 @@ int bnxt_re_create_srq(struct ib_srq *ib_srq,
 	if (nq)
 		nq->budget++;
 	atomic_inc(&rdev->srq_count);
+	spin_lock_init(&srq->lock);
 
 	return 0;
 
Signature · 1 line
2.5.5
author maintainer
#5 lore raw

PATCH v2 3/3 RDMA/bnxt_re: Fix query SRQ failure

In reply to Selvin Xavier

Fill the missing parameters for the FW command while querying SRQ.

Fixes 37cb11acf1f7 ("RDMA/bnxt_re: Add SRQ support for Broadcom adapters")
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/infiniband/hw/bnxt_re/qplib_fp.c +2 -0
index d4d4959..312bf25 100644
@@ -713,6 +713,8 @@ int bnxt_qplib_query_srq(struct bnxt_qplib_res *res,
 	sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
 	if (!sbuf)
 		return -ENOMEM;
+	req.resp_size = sizeof(*sb) / BNXT_QPLIB_CMDQE_UNITS;
+	req.srq_cid = cpu_to_le32(srq->id);
 	sb = sbuf->sb;
 	rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
 					  (void *)sbuf, 0);
Signature · 1 line
2.5.5
maintainer
#6 lore raw

Re: [PATCH rdma-rc v2 3/3] RDMA/bnxt_re: Fix query SRQ failure

Quoted 9 lines · Selvin Xavier

Fill the missing parameters for the FW command while querying SRQ.

Fixes 37cb11acf1f7 ("RDMA/bnxt_re: Add SRQ support for Broadcom adapters")
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 2 ++
 1 file changed, 2 insertions(+)

This commit message is not good enough for -rc, especially with a fixes line so old.

What is the user impact, how did this ever work?

Jason

author maintainer
#7 lore raw

Re: [PATCH rdma-rc v2 3/3] RDMA/bnxt_re: Fix query SRQ failure

Quoted 12 lines · Jason Gunthorpe
Quoted 9 lines · Selvin Xavier

Fill the missing parameters for the FW command while querying SRQ.

Fixes 37cb11acf1f7 ("RDMA/bnxt_re: Add SRQ support for Broadcom adapters")
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 2 ++
 1 file changed, 2 insertions(+)
Quoted 5 lines

This commit message is not good enough for -rc, especially with a fixes line so old.

What is the user impact, how did this ever work?

This never worked. It's an issue with the query_srq only. So was missed in the testing of normal srq traffic.
Will post it for rdma-next along with other patches. Thanks

Quoted 1 line

Jason

maintainer Applied
#8 lore raw
In reply to Selvin Xavier
Quoted 4 lines · Selvin Xavier

Some simple bug fixes

v1->v2:
Fix a build issue

Quoted 8 lines

Naresh Kumar PBS (1):
RDMA/bnxt_re: Add missing spin lock initialization

Selvin Xavier (2):
RDMA/bnxt_re: Disable atomic support on VFs
RDMA/bnxt_re: Fix query SRQ failure

Applied to for-rc, thanks

Jason

maintainer
#9 lore raw
Quoted 7 lines · Jason Gunthorpe
Quoted 6 lines · Selvin Xavier

Some simple bug fixes

v1->v2:
Fix a build issue

Quoted 7 lines
Quoted 5 lines

Naresh Kumar PBS (1):

Selvin Xavier (2):
RDMA/bnxt_re: Disable atomic support on VFs
RDMA/bnxt_re: Fix query SRQ failure

Applied to for-rc, thanks

Woops, I mean only

RDMA/bnxt_re: Add missing spin lock initialization

The others need resend with good commit messages.

Jason