* [PATCH for-next 1/8] IB/hfi1: Harden state transition to Armed and Active
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 2/8] IB/hfi1: Do not enable disabled port on cable insert Dennis Dalessandro
` (7 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Mike Marciniszyn, Alex Estrin
From: Alex Estrin <alex.estrin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
There is a window that allows other threads to read state of
'host_link_state' as a new, before the hardware actual state is set.
This patch closes the window by indicating a new state only after
hardware transition is complete.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Alex Estrin <alex.estrin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 9 +++------
1 files changed, 3 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 8f7ce74..eca31dd 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -10602,16 +10602,15 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state)
break;
}
- ppd->host_link_state = HLS_UP_ARMED;
set_logical_state(dd, LSTATE_ARMED);
ret = wait_logical_linkstate(ppd, IB_PORT_ARMED, 1000);
if (ret) {
- /* logical state didn't change, stay at init */
- ppd->host_link_state = HLS_UP_INIT;
dd_dev_err(dd,
"%s: logical state did not change to ARMED\n",
__func__);
+ break;
}
+ ppd->host_link_state = HLS_UP_ARMED;
/*
* The simulator does not currently implement SMA messages,
* so neighbor_normal is not set. Set it here when we first
@@ -10624,18 +10623,16 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state)
if (ppd->host_link_state != HLS_UP_ARMED)
goto unexpected;
- ppd->host_link_state = HLS_UP_ACTIVE;
set_logical_state(dd, LSTATE_ACTIVE);
ret = wait_logical_linkstate(ppd, IB_PORT_ACTIVE, 1000);
if (ret) {
- /* logical state didn't change, stay at armed */
- ppd->host_link_state = HLS_UP_ARMED;
dd_dev_err(dd,
"%s: logical state did not change to ACTIVE\n",
__func__);
} else {
/* tell all engines to go running */
sdma_all_running(dd);
+ ppd->host_link_state = HLS_UP_ACTIVE;
/* Signal the IB layer that the port has went active */
event.device = &dd->verbs_dev.rdi.ibdev;
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 2/8] IB/hfi1: Do not enable disabled port on cable insert
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
2017-07-29 15:43 ` [PATCH for-next 1/8] IB/hfi1: Harden state transition to Armed and Active Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 3/8] IB/hfi1: Only set fd pointer when base context is completely initialized Dennis Dalessandro
` (6 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Michael J. Ruhl,
Jan Sokolowski, Jakub Byczkowski
From: Jan Sokolowski <jan.sokolowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Fix issue where a disabled port can be enabled by
inserting a cable. The port should be explicitly
enabled instead.
Reviewed-by: Michael J. Ruhl <michael.j.ruhl-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Jakub Byczkowski <jakub.byczkowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jan Sokolowski <jan.sokolowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 13 +++++++------
1 files changed, 7 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index eca31dd..bd594e7 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -9306,12 +9306,6 @@ int start_link(struct hfi1_pportdata *ppd)
*/
tune_serdes(ppd);
- if (!ppd->link_enabled) {
- dd_dev_info(ppd->dd,
- "%s: stopping link start because link is disabled\n",
- __func__);
- return 0;
- }
if (!ppd->driver_link_ready) {
dd_dev_info(ppd->dd,
"%s: stopping link start because driver is not ready\n",
@@ -9529,6 +9523,13 @@ void qsfp_event(struct work_struct *work)
if (!qsfp_mod_present(ppd))
return;
+ if (ppd->host_link_state == HLS_DN_DISABLE) {
+ dd_dev_info(ppd->dd,
+ "%s: stopping link start because link is disabled\n",
+ __func__);
+ return;
+ }
+
/*
* Turn DC back on after cable has been re-inserted. Up until
* now, the DC has been in reset to save power.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 3/8] IB/hfi1: Only set fd pointer when base context is completely initialized
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
2017-07-29 15:43 ` [PATCH for-next 1/8] IB/hfi1: Harden state transition to Armed and Active Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 2/8] IB/hfi1: Do not enable disabled port on cable insert Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 4/8] IB/hfi1: Disambiguate corruption and uninitialized error cases Dennis Dalessandro
` (5 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Michael J. Ruhl,
Sebastian Sanchez
From: Michael J. Ruhl <michael.j.ruhl-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
The allocate_ctxt() function adds the context to the fd data structure.
Since the context is not completely initialized, this can cause confusion
as to whether the context is valid or not.
Move the fd reference from allocate_ctxt() to setup_base_ctxt().
Update the necessary functions to be aware of this move.
Reviewed-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/file_ops.c | 55 +++++++++++++++++------------
drivers/infiniband/hw/hfi1/user_exp_rcv.c | 4 +-
drivers/infiniband/hw/hfi1/user_exp_rcv.h | 3 +-
drivers/infiniband/hw/hfi1/user_sdma.c | 4 +-
drivers/infiniband/hw/hfi1/user_sdma.h | 3 +-
5 files changed, 40 insertions(+), 29 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c
index 650c1e5..a0c13fa 100644
--- a/drivers/infiniband/hw/hfi1/file_ops.c
+++ b/drivers/infiniband/hw/hfi1/file_ops.c
@@ -81,19 +81,22 @@
static int assign_ctxt(struct hfi1_filedata *fd, struct hfi1_user_info *uinfo);
static int init_subctxts(struct hfi1_ctxtdata *uctxt,
const struct hfi1_user_info *uinfo);
-static int init_user_ctxt(struct hfi1_filedata *fd);
+static int init_user_ctxt(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt);
static void user_init(struct hfi1_ctxtdata *uctxt);
static int get_ctxt_info(struct hfi1_filedata *fd, void __user *ubase,
__u32 len);
static int get_base_info(struct hfi1_filedata *fd, void __user *ubase,
__u32 len);
-static int setup_base_ctxt(struct hfi1_filedata *fd);
+static int setup_base_ctxt(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt);
static int setup_subctxt(struct hfi1_ctxtdata *uctxt);
static int find_sub_ctxt(struct hfi1_filedata *fd,
const struct hfi1_user_info *uinfo);
static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
- struct hfi1_user_info *uinfo);
+ struct hfi1_user_info *uinfo,
+ struct hfi1_ctxtdata **cd);
static void deallocate_ctxt(struct hfi1_ctxtdata *uctxt);
static unsigned int poll_urgent(struct file *fp, struct poll_table_struct *pt);
static unsigned int poll_next(struct file *fp, struct poll_table_struct *pt);
@@ -759,7 +762,7 @@ static int hfi1_file_close(struct inode *inode, struct file *fp)
flush_wc();
/* drain user sdma queue */
- hfi1_user_sdma_free_queues(fdata);
+ hfi1_user_sdma_free_queues(fdata, uctxt);
/* release the cpu */
hfi1_put_proc_affinity(fdata->rec_cpu_num);
@@ -845,6 +848,7 @@ static int assign_ctxt(struct hfi1_filedata *fd, struct hfi1_user_info *uinfo)
{
int ret;
unsigned int swmajor, swminor;
+ struct hfi1_ctxtdata *uctxt = NULL;
swmajor = uinfo->userversion >> 16;
if (swmajor != HFI1_USER_SWMAJOR)
@@ -870,7 +874,7 @@ static int assign_ctxt(struct hfi1_filedata *fd, struct hfi1_user_info *uinfo)
* couldn't find a sub context.
*/
if (!ret)
- ret = allocate_ctxt(fd, fd->dd, uinfo);
+ ret = allocate_ctxt(fd, fd->dd, uinfo, &uctxt);
mutex_unlock(&hfi1_mutex);
@@ -888,28 +892,27 @@ static int assign_ctxt(struct hfi1_filedata *fd, struct hfi1_user_info *uinfo)
/* The only thing a sub context needs is the user_xxx stuff */
if (!ret)
- ret = init_user_ctxt(fd);
+ ret = init_user_ctxt(fd, fd->uctxt);
if (ret)
clear_bit(fd->subctxt, fd->uctxt->in_use_ctxts);
} else if (!ret) {
- ret = setup_base_ctxt(fd);
- if (fd->uctxt->subctxt_cnt) {
+ ret = setup_base_ctxt(fd, uctxt);
+ if (uctxt->subctxt_cnt) {
/* If there is an error, set the failed bit. */
if (ret)
set_bit(HFI1_CTXT_BASE_FAILED,
- &fd->uctxt->event_flags);
+ &uctxt->event_flags);
/*
* Base context is done, notify anybody using a
* sub-context that is waiting for this completion
*/
- clear_bit(HFI1_CTXT_BASE_UNINIT,
- &fd->uctxt->event_flags);
- wake_up(&fd->uctxt->wait);
+ clear_bit(HFI1_CTXT_BASE_UNINIT, &uctxt->event_flags);
+ wake_up(&uctxt->wait);
}
if (ret)
- deallocate_ctxt(fd->uctxt);
+ deallocate_ctxt(uctxt);
}
/* If an error occurred, clear the reference */
@@ -976,7 +979,8 @@ static int find_sub_ctxt(struct hfi1_filedata *fd,
}
static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
- struct hfi1_user_info *uinfo)
+ struct hfi1_user_info *uinfo,
+ struct hfi1_ctxtdata **cd)
{
struct hfi1_ctxtdata *uctxt;
u16 ctxt;
@@ -1071,14 +1075,13 @@ static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
*/
if (dd->freectxts-- == dd->num_user_contexts)
aspm_disable_all(dd);
- fd->uctxt = uctxt;
- /* Count the reference for the fd */
- hfi1_rcd_get(uctxt);
+ *cd = uctxt;
return 0;
ctxdata_free:
+ *cd = NULL;
dd->rcd[ctxt] = NULL;
hfi1_rcd_put(uctxt);
return ret;
@@ -1243,23 +1246,25 @@ static int get_ctxt_info(struct hfi1_filedata *fd, void __user *ubase,
return ret;
}
-static int init_user_ctxt(struct hfi1_filedata *fd)
+static int init_user_ctxt(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt)
{
- struct hfi1_ctxtdata *uctxt = fd->uctxt;
int ret;
ret = hfi1_user_sdma_alloc_queues(uctxt, fd);
if (ret)
return ret;
- ret = hfi1_user_exp_rcv_init(fd);
+ ret = hfi1_user_exp_rcv_init(fd, uctxt);
+ if (ret)
+ hfi1_user_sdma_free_queues(fd, uctxt);
return ret;
}
-static int setup_base_ctxt(struct hfi1_filedata *fd)
+static int setup_base_ctxt(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt)
{
- struct hfi1_ctxtdata *uctxt = fd->uctxt;
struct hfi1_devdata *dd = uctxt->dd;
int ret = 0;
@@ -1284,12 +1289,16 @@ static int setup_base_ctxt(struct hfi1_filedata *fd)
if (ret)
goto setup_failed;
- ret = init_user_ctxt(fd);
+ ret = init_user_ctxt(fd, uctxt);
if (ret)
goto setup_failed;
user_init(uctxt);
+ /* Now that the context is set up, the fd can get a reference. */
+ fd->uctxt = uctxt;
+ hfi1_rcd_get(uctxt);
+
return 0;
setup_failed:
diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
index 3baf719..d9036ba 100644
--- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c
+++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
@@ -104,9 +104,9 @@ static int unprogram_rcvarray(struct hfi1_filedata *fd, u32 tidinfo,
* receive caching. This needs to be done after the context has
* been configured with the eager/expected RcvEntry counts.
*/
-int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd)
+int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt)
{
- struct hfi1_ctxtdata *uctxt = fd->uctxt;
struct hfi1_devdata *dd = uctxt->dd;
int ret = 0;
diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.h b/drivers/infiniband/hw/hfi1/user_exp_rcv.h
index 1bdc61b..6cbaa4c 100644
--- a/drivers/infiniband/hw/hfi1/user_exp_rcv.h
+++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.h
@@ -51,7 +51,8 @@
#include "exp_rcv.h"
-int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd);
+int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt);
void hfi1_user_exp_rcv_free(struct hfi1_filedata *fd);
int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
struct hfi1_tid_info *tinfo);
diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c
index 64d31eb..8f02707 100644
--- a/drivers/infiniband/hw/hfi1/user_sdma.c
+++ b/drivers/infiniband/hw/hfi1/user_sdma.c
@@ -437,9 +437,9 @@ int hfi1_user_sdma_alloc_queues(struct hfi1_ctxtdata *uctxt,
return ret;
}
-int hfi1_user_sdma_free_queues(struct hfi1_filedata *fd)
+int hfi1_user_sdma_free_queues(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt)
{
- struct hfi1_ctxtdata *uctxt = fd->uctxt;
struct hfi1_user_sdma_pkt_q *pq;
hfi1_cdbg(SDMA, "[%u:%u:%u] Freeing user SDMA queues", uctxt->dd->unit,
diff --git a/drivers/infiniband/hw/hfi1/user_sdma.h b/drivers/infiniband/hw/hfi1/user_sdma.h
index 73c2455..84c199d 100644
--- a/drivers/infiniband/hw/hfi1/user_sdma.h
+++ b/drivers/infiniband/hw/hfi1/user_sdma.h
@@ -81,7 +81,8 @@ struct hfi1_user_sdma_comp_q {
int hfi1_user_sdma_alloc_queues(struct hfi1_ctxtdata *uctxt,
struct hfi1_filedata *fd);
-int hfi1_user_sdma_free_queues(struct hfi1_filedata *fd);
+int hfi1_user_sdma_free_queues(struct hfi1_filedata *fd,
+ struct hfi1_ctxtdata *uctxt);
int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
struct iovec *iovec, unsigned long dim,
unsigned long *count);
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 4/8] IB/hfi1: Disambiguate corruption and uninitialized error cases
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (2 preceding siblings ...)
2017-07-29 15:43 ` [PATCH for-next 3/8] IB/hfi1: Only set fd pointer when base context is completely initialized Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 5/8] IB/{rdmavt, hfi1, qib}: Fix panic with post receive and SGE compression Dennis Dalessandro
` (4 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Easwar Hariharan,
Jan Sokolowski, Jakub Byczkowski
From: Jan Sokolowski <jan.sokolowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
The error messages when checksum validation of the platform
configuration fields populated into the ASIC scratch registers fails are
ambiguous. Disambiguate them.
Reviewed-by: Jakub Byczkowski <jakub.byczkowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Easwar Hariharan <easwar.hariharan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jan Sokolowski <jan.sokolowski-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/platform.c | 14 ++++++++------
1 files changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/platform.c b/drivers/infiniband/hw/hfi1/platform.c
index 41307e4..5c38a24 100644
--- a/drivers/infiniband/hw/hfi1/platform.c
+++ b/drivers/infiniband/hw/hfi1/platform.c
@@ -58,8 +58,13 @@ static int validate_scratch_checksum(struct hfi1_devdata *dd)
version = (temp_scratch & BITMAP_VERSION_SMASK) >> BITMAP_VERSION_SHIFT;
/* Prevent power on default of all zeroes from passing checksum */
- if (!version)
+ if (!version) {
+ dd_dev_err(dd, "%s: Config bitmap uninitialized\n", __func__);
+ dd_dev_err(dd,
+ "%s: Please update your BIOS to support active channels\n",
+ __func__);
return 0;
+ }
/*
* ASIC scratch 0 only contains the checksum and bitmap version as
@@ -84,6 +89,8 @@ static int validate_scratch_checksum(struct hfi1_devdata *dd)
if (checksum + temp_scratch == 0xFFFF)
return 1;
+
+ dd_dev_err(dd, "%s: Configuration bitmap corrupted\n", __func__);
return 0;
}
@@ -144,11 +151,6 @@ void get_platform_config(struct hfi1_devdata *dd)
save_platform_config_fields(dd);
return;
}
- dd_dev_err(dd, "%s: Config bitmap corrupted/uninitialized\n",
- __func__);
- dd_dev_err(dd,
- "%s: Please update your BIOS to support active channels\n",
- __func__);
} else {
ret = eprom_read_platform_config(dd,
(void **)&temp_platform_config,
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 5/8] IB/{rdmavt, hfi1, qib}: Fix panic with post receive and SGE compression
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (3 preceding siblings ...)
2017-07-29 15:43 ` [PATCH for-next 4/8] IB/hfi1: Disambiguate corruption and uninitialized error cases Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:43 ` [PATCH for-next 6/8] IB/hfi1: Create workqueue for link events Dennis Dalessandro
` (3 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Mike Marciniszyn
From: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
The server side of qperf panics as follows:
[242446.336860] IP: report_bug+0x64/0x10
[242446.341031] PGD 1c0c067
[242446.341032] P4D 1c0c067
[242446.343951] PUD 1c0d063
[242446.346870] PMD 8587ea067
[242446.349788] PTE 800000083e14016
[242446.352901]
[242446.358352] Oops: 0003 [#1] SM
[242446.437919] CPU: 1 PID: 7442 Comm: irq/92-hfi1_0 k Not tainted 4.12.0-mam-asm #1
[242446.446365] Hardware name: Intel Corporation S2600WT2/S2600WT2, BIOS SE5C610.86B.01.01.0018.C4.072020161249 07/20/201
[242446.458397] task: ffff8808392d2b80 task.stack: ffffc9000664000
[242446.465097] RIP: 0010:report_bug+0x64/0x10
[242446.469859] RSP: 0018:ffffc900066439c0 EFLAGS: 0001000
[242446.475784] RAX: ffffffffa06647e4 RBX: ffffffffa06461e1 RCX: 000000000000000
[242446.483840] RDX: 0000000000000907 RSI: ffffffffa0675040 RDI: ffffffffffff740
[242446.491897] RBP: ffffc900066439e0 R08: 0000000000000001 R09: 000000000000025
[242446.499953] R10: ffffffff81a253df R11: 0000000000000133 R12: ffffc90006643b3
[242446.508010] R13: ffffffffa065bbf0 R14: 00000000000001e5 R15: 000000000000000
[242446.516067] FS: 0000000000000000(0000) GS:ffff88085f640000(0000) knlGS:000000000000000
[242446.525191] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003
[242446.531698] CR2: ffffffffa06647ee CR3: 0000000001c09000 CR4: 00000000001406e
[242446.539756] Call Trace
[242446.542582] fixup_bug+0x2c/0x5
[242446.546277] do_trap+0x12b/0x18
[242446.549972] do_error_trap+0x89/0x11
[242446.554171] ? hfi1_copy_sge+0x271/0x2b0 [hfi1
[242446.559324] ? ttwu_do_wakeup+0x1e/0x14
[242446.563795] ? ttwu_do_activate+0x77/0x8
[242446.568363] do_invalid_op+0x20/0x3
[242446.572448] invalid_op+0x1e/0x3
[242446.576247] RIP: 0010:hfi1_copy_sge+0x271/0x2b0 [hfi1
[242446.582075] RSP: 0018:ffffc90006643be8 EFLAGS: 0001004
[242446.587999] RAX: 0000000000000000 RBX: ffff88083e0fa240 RCX: 000000000000000
[242446.596058] RDX: 0000000000000000 RSI: ffff880842508000 RDI: ffff88083e0fa24
[242446.604116] RBP: ffffc90006643c28 R08: 0000000000000000 R09: 000000000000000
[242446.612172] R10: ffffc90009473640 R11: 0000000000000133 R12: 000000000000000
[242446.620228] R13: 0000000000000000 R14: 0000000000002000 R15: ffff88084250800
[242446.628293] ? hfi1_copy_sge+0x1a1/0x2b0 [hfi1
[242446.633449] hfi1_rc_rcv+0x3da/0x1270 [hfi1
[242446.638312] ? sc_buffer_alloc+0x113/0x150 [hfi1
[242446.643662] hfi1_ib_rcv+0x1c9/0x2e0 [hfi1
[242446.648428] process_receive_ib+0x19a/0x270 [hfi1
[242446.653866] ? process_rcv_qp_work+0xd2/0x160 [hfi1
[242446.659505] handle_receive_interrupt_nodma_rtail+0x184/0x2e0 [hfi1
[242446.666693] ? irq_finalize_oneshot+0x100/0x10
[242446.671846] receive_context_thread+0x1b/0x140 [hfi1
[242446.677576] irq_thread_fn+0x1e/0x4
[242446.681659] irq_thread+0x13c/0x1b
[242446.685646] ? irq_forced_thread_fn+0x60/0x6
[242446.690604] kthread+0x112/0x15
[242446.694298] ? irq_thread_check_affinity+0xe0/0xe
[242446.699738] ? kthread_park+0x60/0x6
[242446.703919] ? do_syscall_64+0x67/0x15
[242446.708292] ret_from_fork+0x25/0x3
[242446.712374] Code: 63 78 04 44 0f b7 70 08 41 89 d0 4c 8d 2c 38 41 83 e0 01 f6 c2 02 74 17 66 45 85 c0 74 11 f6 c2 04 b9 01 00 00 00 75 bb 83 ca 04 <66> 89 50 0a 66 45 85 c0 74 52 0f b6 48 0b 41 0f b7 f6 4d 89 e0
[242446.733527] RIP: report_bug+0x64/0x100 RSP: ffffc900066439c
[242446.739935] CR2: ffffffffa06647e
[242446.743763] ---[ end trace 0e90a20d0aa494f7 ]--
The root cause is that the qib/hfi1 post receive call to rvt_lkey_ok()
doesn't interpret the new return value from rvt_lkey_ok() properly
leading to an mr reference count underrun.
Additionally, remove an unused argument in rvt_sge_adjacent()
aw well as an unneeded incr local in rvt_post_one_wr().
Fixes: Commit 14fe13fcd3af ("IB/rdmavt: Compress adjacent SGEs in rvt_lkey_ok()")
Signed-off-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/ruc.c | 6 ++++--
drivers/infiniband/hw/qib/qib_ruc.c | 6 ++++--
drivers/infiniband/sw/rdmavt/mr.c | 8 +++-----
drivers/infiniband/sw/rdmavt/qp.c | 16 +++++++---------
4 files changed, 18 insertions(+), 18 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/ruc.c b/drivers/infiniband/hw/hfi1/ruc.c
index f13ddb2..4afa00f 100644
--- a/drivers/infiniband/hw/hfi1/ruc.c
+++ b/drivers/infiniband/hw/hfi1/ruc.c
@@ -74,8 +74,10 @@ static int init_sge(struct rvt_qp *qp, struct rvt_rwqe *wqe)
if (wqe->sg_list[i].length == 0)
continue;
/* Check LKEY */
- if (!rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
- NULL, &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
+ ret = rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
+ NULL, &wqe->sg_list[i],
+ IB_ACCESS_LOCAL_WRITE);
+ if (unlikely(ret <= 0))
goto bad_lkey;
qp->r_len += wqe->sg_list[i].length;
j++;
diff --git a/drivers/infiniband/hw/qib/qib_ruc.c b/drivers/infiniband/hw/qib/qib_ruc.c
index 2852845..e6a42a8 100644
--- a/drivers/infiniband/hw/qib/qib_ruc.c
+++ b/drivers/infiniband/hw/qib/qib_ruc.c
@@ -58,8 +58,10 @@ static int qib_init_sge(struct rvt_qp *qp, struct rvt_rwqe *wqe)
if (wqe->sg_list[i].length == 0)
continue;
/* Check LKEY */
- if (!rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
- NULL, &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
+ ret = rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
+ NULL, &wqe->sg_list[i],
+ IB_ACCESS_LOCAL_WRITE);
+ if (unlikely(ret <= 0))
goto bad_lkey;
qp->r_len += wqe->sg_list[i].length;
j++;
diff --git a/drivers/infiniband/sw/rdmavt/mr.c b/drivers/infiniband/sw/rdmavt/mr.c
index ea95672..1b3801f 100644
--- a/drivers/infiniband/sw/rdmavt/mr.c
+++ b/drivers/infiniband/sw/rdmavt/mr.c
@@ -779,7 +779,6 @@ int rvt_dealloc_fmr(struct ib_fmr *ibfmr)
/**
* rvt_sge_adjacent - is isge compressible
- * @isge: outgoing internal SGE
* @last_sge: last outgoing SGE written
* @sge: SGE to check
*
@@ -787,8 +786,7 @@ int rvt_dealloc_fmr(struct ib_fmr *ibfmr)
*
* Return: true if isge is adjacent to last sge
*/
-static inline bool rvt_sge_adjacent(struct rvt_sge *isge,
- struct rvt_sge *last_sge,
+static inline bool rvt_sge_adjacent(struct rvt_sge *last_sge,
struct ib_sge *sge)
{
if (last_sge && sge->lkey == last_sge->mr->lkey &&
@@ -840,7 +838,7 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
if (pd->user)
return -EINVAL;
- if (rvt_sge_adjacent(isge, last_sge, sge))
+ if (rvt_sge_adjacent(last_sge, sge))
return 0;
rcu_read_lock();
mr = rcu_dereference(dev->dma_mr);
@@ -857,7 +855,7 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
isge->n = 0;
goto ok;
}
- if (rvt_sge_adjacent(isge, last_sge, sge))
+ if (rvt_sge_adjacent(last_sge, sge))
return 0;
rcu_read_lock();
mr = rcu_dereference(rkt->table[sge->lkey >> rkt->shift]);
diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c
index 740611e..1878a97 100644
--- a/drivers/infiniband/sw/rdmavt/qp.c
+++ b/drivers/infiniband/sw/rdmavt/qp.c
@@ -1613,7 +1613,7 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
struct rvt_pd *pd;
struct rvt_dev_info *rdi = ib_to_rvt(qp->ibqp.device);
u8 log_pmtu;
- int ret, incr;
+ int ret;
size_t cplen;
bool reserved_op;
int local_ops_delayed = 0;
@@ -1695,14 +1695,14 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
if (length == 0)
continue;
- incr = rvt_lkey_ok(rkt, pd, &wqe->sg_list[j], last_sge,
- &wr->sg_list[i], acc);
- if (unlikely(incr < 0))
- goto bail_lkey_error;
+ ret = rvt_lkey_ok(rkt, pd, &wqe->sg_list[j], last_sge,
+ &wr->sg_list[i], acc);
+ if (unlikely(ret < 0))
+ goto bail_inval_free;
wqe->length += length;
- if (incr)
+ if (ret)
last_sge = &wqe->sg_list[j];
- j += incr;
+ j += ret;
}
wqe->wr.num_sge = j;
}
@@ -1755,8 +1755,6 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
return 0;
-bail_lkey_error:
- ret = incr;
bail_inval_free:
/* release mr holds */
while (j) {
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 6/8] IB/hfi1: Create workqueue for link events
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (4 preceding siblings ...)
2017-07-29 15:43 ` [PATCH for-next 5/8] IB/{rdmavt, hfi1, qib}: Fix panic with post receive and SGE compression Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
[not found] ` <20170729154348.21487.66943.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
2017-07-29 15:43 ` [PATCH for-next 7/8] IB/hfi1: Prevent link down request double queuing Dennis Dalessandro
` (2 subsequent siblings)
8 siblings, 1 reply; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Dean Luick, Sebastian Sanchez
From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Currently, link down interrupts queue link entries
on a workqueue intended for sending events only.
Create a workqueue for queuing link events.
Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 20 ++++++++++----------
drivers/infiniband/hw/hfi1/driver.c | 2 +-
drivers/infiniband/hw/hfi1/hfi.h | 1 +
drivers/infiniband/hw/hfi1/init.c | 26 ++++++++++++++++++++++++++
drivers/infiniband/hw/hfi1/pio.c | 2 +-
drivers/infiniband/hw/hfi1/sdma.c | 2 +-
6 files changed, 40 insertions(+), 13 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index bd594e7..8443d41 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -5545,7 +5545,7 @@ static void update_rcverr_timer(unsigned long opaque)
set_link_down_reason(
ppd, OPA_LINKDOWN_REASON_EXCESSIVE_BUFFER_OVERRUN, 0,
OPA_LINKDOWN_REASON_EXCESSIVE_BUFFER_OVERRUN);
- queue_work(ppd->hfi1_wq, &ppd->link_bounce_work);
+ queue_work(ppd->link_wq, &ppd->link_bounce_work);
}
dd->rcv_ovfl_cnt = (u32)cur_ovfl_cnt;
@@ -6100,7 +6100,7 @@ static void handle_qsfp_int(struct hfi1_devdata *dd, u32 src_ctx, u64 reg)
* will not happen. We have to do it here
* before turning the DC off.
*/
- queue_work(ppd->hfi1_wq, &ppd->link_down_work);
+ queue_work(ppd->link_wq, &ppd->link_down_work);
}
} else {
dd_dev_info(dd, "%s: QSFP module inserted\n",
@@ -6135,7 +6135,7 @@ static void handle_qsfp_int(struct hfi1_devdata *dd, u32 src_ctx, u64 reg)
/* Schedule the QSFP work only if there is a cable attached. */
if (qsfp_mod_present(ppd))
- queue_work(ppd->hfi1_wq, &ppd->qsfp_info.qsfp_work);
+ queue_work(ppd->link_wq, &ppd->qsfp_info.qsfp_work);
}
static int request_host_lcb_access(struct hfi1_devdata *dd)
@@ -7738,12 +7738,12 @@ static void handle_8051_interrupt(struct hfi1_devdata *dd, u32 unused, u64 reg)
host_msg &= ~(u64)HOST_REQ_DONE;
}
if (host_msg & BC_SMA_MSG) {
- queue_work(ppd->hfi1_wq, &ppd->sma_message_work);
+ queue_work(ppd->link_wq, &ppd->sma_message_work);
host_msg &= ~(u64)BC_SMA_MSG;
}
if (host_msg & LINKUP_ACHIEVED) {
dd_dev_info(dd, "8051: Link up\n");
- queue_work(ppd->hfi1_wq, &ppd->link_up_work);
+ queue_work(ppd->link_wq, &ppd->link_up_work);
host_msg &= ~(u64)LINKUP_ACHIEVED;
}
if (host_msg & EXT_DEVICE_CFG_REQ) {
@@ -7751,7 +7751,7 @@ static void handle_8051_interrupt(struct hfi1_devdata *dd, u32 unused, u64 reg)
host_msg &= ~(u64)EXT_DEVICE_CFG_REQ;
}
if (host_msg & VERIFY_CAP_FRAME) {
- queue_work(ppd->hfi1_wq, &ppd->link_vc_work);
+ queue_work(ppd->link_wq, &ppd->link_vc_work);
host_msg &= ~(u64)VERIFY_CAP_FRAME;
}
if (host_msg & LINK_GOING_DOWN) {
@@ -7766,7 +7766,7 @@ static void handle_8051_interrupt(struct hfi1_devdata *dd, u32 unused, u64 reg)
host_msg &= ~(u64)LINK_GOING_DOWN;
}
if (host_msg & LINK_WIDTH_DOWNGRADED) {
- queue_work(ppd->hfi1_wq, &ppd->link_downgrade_work);
+ queue_work(ppd->link_wq, &ppd->link_downgrade_work);
host_msg &= ~(u64)LINK_WIDTH_DOWNGRADED;
}
if (host_msg) {
@@ -7809,7 +7809,7 @@ static void handle_8051_interrupt(struct hfi1_devdata *dd, u32 unused, u64 reg)
dd_dev_info(dd, "%s: not queuing link down\n",
__func__);
} else {
- queue_work(ppd->hfi1_wq, &ppd->link_down_work);
+ queue_work(ppd->link_wq, &ppd->link_down_work);
}
}
}
@@ -8017,7 +8017,7 @@ static void handle_dcc_err(struct hfi1_devdata *dd, u32 unused, u64 reg)
dd_dev_info_ratelimited(dd, "%s: PortErrorAction bounce\n",
__func__);
set_link_down_reason(ppd, lcl_reason, 0, lcl_reason);
- queue_work(ppd->hfi1_wq, &ppd->link_bounce_work);
+ queue_work(ppd->link_wq, &ppd->link_bounce_work);
}
}
@@ -9685,7 +9685,7 @@ static void try_start_link(struct hfi1_pportdata *ppd)
"QSFP not responding, waiting and retrying %d\n",
(int)ppd->qsfp_retry_count);
ppd->qsfp_retry_count++;
- queue_delayed_work(ppd->hfi1_wq, &ppd->start_link_work,
+ queue_delayed_work(ppd->link_wq, &ppd->start_link_work,
msecs_to_jiffies(QSFP_RETRY_WAIT));
return;
}
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 01235d4..0b7ca0e 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -914,7 +914,7 @@ static inline int set_armed_to_active(struct hfi1_ctxtdata *rcd,
return 0;
}
- queue_work(rcd->ppd->hfi1_wq, lsaw);
+ queue_work(rcd->ppd->link_wq, lsaw);
return 1;
}
return 0;
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 2d32c5c..ee6c389 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -576,6 +576,7 @@ struct hfi1_pportdata {
/* SendDMA related entries */
struct workqueue_struct *hfi1_wq;
+ struct workqueue_struct *link_wq;
/* move out of interrupt context */
struct work_struct link_vc_work;
diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index 7f5e4c7..be027c9 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -660,6 +660,20 @@ static int create_workqueues(struct hfi1_devdata *dd)
if (!ppd->hfi1_wq)
goto wq_error;
}
+ if (!ppd->link_wq) {
+ /*
+ * Make the link workqueue single-threaded to enforce
+ * serialization.
+ */
+ ppd->link_wq =
+ alloc_workqueue(
+ "hfi_link_%d_%d",
+ WQ_SYSFS | WQ_MEM_RECLAIM | WQ_UNBOUND,
+ 1, /* max_active */
+ dd->unit, pidx);
+ if (!ppd->link_wq)
+ goto wq_error;
+ }
}
return 0;
wq_error:
@@ -670,6 +684,10 @@ static int create_workqueues(struct hfi1_devdata *dd)
destroy_workqueue(ppd->hfi1_wq);
ppd->hfi1_wq = NULL;
}
+ if (ppd->link_wq) {
+ destroy_workqueue(ppd->link_wq);
+ ppd->link_wq = NULL;
+ }
}
return -ENOMEM;
}
@@ -954,6 +972,10 @@ static void shutdown_device(struct hfi1_devdata *dd)
destroy_workqueue(ppd->hfi1_wq);
ppd->hfi1_wq = NULL;
}
+ if (ppd->link_wq) {
+ destroy_workqueue(ppd->link_wq);
+ ppd->link_wq = NULL;
+ }
}
sdma_exit(dd);
}
@@ -1575,6 +1597,10 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
destroy_workqueue(ppd->hfi1_wq);
ppd->hfi1_wq = NULL;
}
+ if (ppd->link_wq) {
+ destroy_workqueue(ppd->link_wq);
+ ppd->link_wq = NULL;
+ }
}
if (!j)
hfi1_device_remove(dd);
diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c
index adb6a4d..7108a4b 100644
--- a/drivers/infiniband/hw/hfi1/pio.c
+++ b/drivers/infiniband/hw/hfi1/pio.c
@@ -1012,7 +1012,7 @@ static void sc_wait_for_packet_egress(struct send_context *sc, int pause)
"%s: context %u(%u) timeout waiting for packets to egress, remaining count %u, bouncing link\n",
__func__, sc->sw_index,
sc->hw_context, (u32)reg);
- queue_work(dd->pport->hfi1_wq,
+ queue_work(dd->pport->link_wq,
&dd->pport->link_bounce_work);
break;
}
diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c
index 71b4258..6781bcd 100644
--- a/drivers/infiniband/hw/hfi1/sdma.c
+++ b/drivers/infiniband/hw/hfi1/sdma.c
@@ -325,7 +325,7 @@ static void sdma_wait_for_packet_egress(struct sdma_engine *sde,
/* timed out - bounce the link */
dd_dev_err(dd, "%s: engine %u timeout waiting for packets to egress, remaining count %u, bouncing link\n",
__func__, sde->this_idx, (u32)reg);
- queue_work(dd->pport->hfi1_wq,
+ queue_work(dd->pport->link_wq,
&dd->pport->link_bounce_work);
break;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 7/8] IB/hfi1: Prevent link down request double queuing
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (5 preceding siblings ...)
2017-07-29 15:43 ` [PATCH for-next 6/8] IB/hfi1: Create workqueue for link events Dennis Dalessandro
@ 2017-07-29 15:43 ` Dennis Dalessandro
2017-07-29 15:44 ` [PATCH for-next 8/8] IB/hfi1: Always perform offline transition Dennis Dalessandro
2017-07-31 19:27 ` [PATCH for-next 0/8] IB/hfi1, rdmavt, qib: patches for next 07/29/2017 Doug Ledford
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:43 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Dean Luick, Sebastian Sanchez
From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
When link interrupts occur, multiple link down requests
could be queued up when only one is needed. This could get
the hfi1 out of sync with its link partner during LNI.
Only allow one link down request to be queued at any one time.
Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 4 +++-
drivers/infiniband/hw/hfi1/hfi.h | 1 +
2 files changed, 4 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 8443d41..a3af46c 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -7045,6 +7045,7 @@ void handle_link_down(struct work_struct *work)
/* Go offline first, then deal with reading/writing through 8051 */
was_up = !!(ppd->host_link_state & HLS_UP);
set_link_state(ppd, HLS_DN_OFFLINE);
+ xchg(&ppd->is_link_down_queued, 0);
if (was_up) {
lcl_reason = 0;
@@ -7805,10 +7806,11 @@ static void handle_8051_interrupt(struct hfi1_devdata *dd, u32 unused, u64 reg)
*/
if ((ppd->host_link_state &
(HLS_GOING_OFFLINE | HLS_LINK_COOLDOWN)) ||
- ppd->link_enabled == 0) {
+ ppd->link_enabled == 0 || ppd->is_link_down_queued) {
dd_dev_info(dd, "%s: not queuing link down\n",
__func__);
} else {
+ xchg(&ppd->is_link_down_queued, 1);
queue_work(ppd->link_wq, &ppd->link_down_work);
}
}
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index ee6c389..fb5f839 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -644,6 +644,7 @@ struct hfi1_pportdata {
/* placeholders for IB MAD packet settings */
u8 overrun_threshold;
u8 phy_error_threshold;
+ unsigned int is_link_down_queued;
/* Used to override LED behavior for things like maintenance beaconing*/
/*
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH for-next 8/8] IB/hfi1: Always perform offline transition
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (6 preceding siblings ...)
2017-07-29 15:43 ` [PATCH for-next 7/8] IB/hfi1: Prevent link down request double queuing Dennis Dalessandro
@ 2017-07-29 15:44 ` Dennis Dalessandro
2017-07-31 19:27 ` [PATCH for-next 0/8] IB/hfi1, rdmavt, qib: patches for next 07/29/2017 Doug Ledford
8 siblings, 0 replies; 11+ messages in thread
From: Dennis Dalessandro @ 2017-07-29 15:44 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Dean Luick, Sebastian Sanchez
From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Always initiate an offline transition request
when a link down occurs. The firmware will
use this request to confirm that the driver
has seen the link down message. A host version
is set to indicate this driver behavior to the
firmware.
Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 66 +++++++++++++++++----------------
drivers/infiniband/hw/hfi1/chip.h | 5 +++
drivers/infiniband/hw/hfi1/firmware.c | 8 ++++
3 files changed, 46 insertions(+), 33 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index a3af46c..101fbbb 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -8832,6 +8832,20 @@ static void read_remote_device_id(struct hfi1_devdata *dd, u16 *device_id,
& REMOTE_DEVICE_REV_MASK;
}
+int write_host_interface_version(struct hfi1_devdata *dd, u8 version)
+{
+ u32 frame;
+ u32 mask;
+
+ mask = (HOST_INTERFACE_VERSION_MASK << HOST_INTERFACE_VERSION_SHIFT);
+ read_8051_config(dd, RESERVED_REGISTERS, GENERAL_CONFIG, &frame);
+ /* Clear, then set field */
+ frame &= ~mask;
+ frame |= ((u32)version << HOST_INTERFACE_VERSION_SHIFT);
+ return load_8051_config(dd, RESERVED_REGISTERS, GENERAL_CONFIG,
+ frame);
+}
+
void read_misc_status(struct hfi1_devdata *dd, u8 *ver_major, u8 *ver_minor,
u8 *ver_patch)
{
@@ -10262,49 +10276,35 @@ static void force_logical_link_state_down(struct hfi1_pportdata *ppd)
static int goto_offline(struct hfi1_pportdata *ppd, u8 rem_reason)
{
struct hfi1_devdata *dd = ppd->dd;
- u32 pstate, previous_state;
+ u32 previous_state;
int ret;
- int do_transition;
- int do_wait;
update_lcb_cache(dd);
previous_state = ppd->host_link_state;
ppd->host_link_state = HLS_GOING_OFFLINE;
- pstate = read_physical_state(dd);
- if (pstate == PLS_OFFLINE) {
- do_transition = 0; /* in right state */
- do_wait = 0; /* ...no need to wait */
- } else if ((pstate & 0xf0) == PLS_OFFLINE) {
- do_transition = 0; /* in an offline transient state */
- do_wait = 1; /* ...wait for it to settle */
- } else {
- do_transition = 1; /* need to move to offline */
- do_wait = 1; /* ...will need to wait */
- }
- if (do_transition) {
- ret = set_physical_link_state(dd,
- (rem_reason << 8) | PLS_OFFLINE);
+ /* start offline transition */
+ ret = set_physical_link_state(dd, (rem_reason << 8) | PLS_OFFLINE);
- if (ret != HCMD_SUCCESS) {
- dd_dev_err(dd,
- "Failed to transition to Offline link state, return %d\n",
- ret);
- return -EINVAL;
- }
- if (ppd->offline_disabled_reason ==
- HFI1_ODR_MASK(OPA_LINKDOWN_REASON_NONE))
- ppd->offline_disabled_reason =
- HFI1_ODR_MASK(OPA_LINKDOWN_REASON_TRANSIENT);
+ if (ret != HCMD_SUCCESS) {
+ dd_dev_err(dd,
+ "Failed to transition to Offline link state, return %d\n",
+ ret);
+ return -EINVAL;
}
+ if (ppd->offline_disabled_reason ==
+ HFI1_ODR_MASK(OPA_LINKDOWN_REASON_NONE))
+ ppd->offline_disabled_reason =
+ HFI1_ODR_MASK(OPA_LINKDOWN_REASON_TRANSIENT);
- if (do_wait) {
- /* it can take a while for the link to go down */
- ret = wait_physical_linkstate(ppd, PLS_OFFLINE, 10000);
- if (ret < 0)
- return ret;
- }
+ /*
+ * Wait for offline transition. It can take a while for
+ * the link to go down.
+ */
+ ret = wait_physical_linkstate(ppd, PLS_OFFLINE, 10000);
+ if (ret < 0)
+ return ret;
/*
* Now in charge of LCB - must be after the physical state is
diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h
index bef6301..6a0c691 100644
--- a/drivers/infiniband/hw/hfi1/chip.h
+++ b/drivers/infiniband/hw/hfi1/chip.h
@@ -384,6 +384,7 @@
#define VERIFY_CAP_LOCAL_FABRIC 0x08
#define VERIFY_CAP_LOCAL_LINK_WIDTH 0x09
#define LOCAL_DEVICE_ID 0x0a
+#define RESERVED_REGISTERS 0x0b
#define LOCAL_LNI_INFO 0x0c
#define REMOTE_LNI_INFO 0x0d
#define MISC_STATUS 0x0e
@@ -506,6 +507,9 @@
#define DOWN_REMOTE_REASON_SHIFT 16
#define DOWN_REMOTE_REASON_MASK 0xff
+#define HOST_INTERFACE_VERSION_SHIFT 16
+#define HOST_INTERFACE_VERSION_MASK 0xff
+
/* verify capability PHY power management bits */
#define PWRM_BER_CONTROL 0x1
#define PWRM_BANDWIDTH_CONTROL 0x2
@@ -704,6 +708,7 @@ bool check_chip_resource(struct hfi1_devdata *dd, u32 resource,
/* chip.c */
void read_misc_status(struct hfi1_devdata *dd, u8 *ver_major, u8 *ver_minor,
u8 *ver_patch);
+int write_host_interface_version(struct hfi1_devdata *dd, u8 version);
void read_guid(struct hfi1_devdata *dd);
int wait_fm_ready(struct hfi1_devdata *dd, u32 mstimeout);
void set_link_down_reason(struct hfi1_pportdata *ppd, u8 lcl_reason,
diff --git a/drivers/infiniband/hw/hfi1/firmware.c b/drivers/infiniband/hw/hfi1/firmware.c
index 83efd6b..885714b 100644
--- a/drivers/infiniband/hw/hfi1/firmware.c
+++ b/drivers/infiniband/hw/hfi1/firmware.c
@@ -69,6 +69,7 @@
#define ALT_FW_FABRIC_NAME "hfi1_fabric_d.fw"
#define ALT_FW_SBUS_NAME "hfi1_sbus_d.fw"
#define ALT_FW_PCIE_NAME "hfi1_pcie_d.fw"
+#define HOST_INTERFACE_VERSION 1
static uint fw_8051_load = 1;
static uint fw_fabric_serdes_load = 1;
@@ -1087,6 +1088,13 @@ static int load_8051_firmware(struct hfi1_devdata *dd,
dd_dev_info(dd, "8051 firmware version %d.%d.%d\n",
(int)ver_major, (int)ver_minor, (int)ver_patch);
dd->dc8051_ver = dc8051_ver(ver_major, ver_minor, ver_patch);
+ ret = write_host_interface_version(dd, HOST_INTERFACE_VERSION);
+ if (ret != HCMD_SUCCESS) {
+ dd_dev_err(dd,
+ "Failed to set host interface version, return 0x%x\n",
+ ret);
+ return -EIO;
+ }
return 0;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 11+ messages in thread* Re: [PATCH for-next 0/8] IB/hfi1, rdmavt, qib: patches for next 07/29/2017
[not found] ` <20170729154124.21487.53119.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (7 preceding siblings ...)
2017-07-29 15:44 ` [PATCH for-next 8/8] IB/hfi1: Always perform offline transition Dennis Dalessandro
@ 2017-07-31 19:27 ` Doug Ledford
8 siblings, 0 replies; 11+ messages in thread
From: Doug Ledford @ 2017-07-31 19:27 UTC (permalink / raw)
To: Dennis Dalessandro
Cc: Mike Marciniszyn, Jan Sokolowski, Jakub Byczkowski,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, Alex Estrin, Michael J. Ruhl,
Easwar Hariharan, Dean Luick, Sebastian Sanchez
On Sat, 2017-07-29 at 08:43 -0700, Dennis Dalessandro wrote:
> Hi Doug,
>
> Here is the next set of patches our our drivers for 4.14. There are
> a couple
> patches for link bugs and a bug fix from Mike for SGE compression.
> This fixes an
> issue in a previous commit that we sent for 4.14, since it's already
> in your
> tree on kernel.org I have included a fixes line. There are three
> other minor
> fixes as well.
>
> Patches can can also be found in my GitHub repo at:
> https://github.com/ddalessa/kernel/tree/for-4.14
>
> ---
>
> Alex Estrin (1):
> IB/hfi1: Harden state transition to Armed and Active
>
> Jan Sokolowski (2):
> IB/hfi1: Do not enable disabled port on cable insert
> IB/hfi1: Disambiguate corruption and uninitialized error cases
>
> Michael J. Ruhl (1):
> IB/hfi1: Only set fd pointer when base context is completely
> initialized
>
> Mike Marciniszyn (1):
> IB/{rdmavt, hfi1, qib}: Fix panic with post receive and SGE
> compression
>
> Sebastian Sanchez (3):
> IB/hfi1: Create workqueue for link events
> IB/hfi1: Prevent link down request double queuing
> IB/hfi1: Always perform offline transition
>
>
> drivers/infiniband/hw/hfi1/chip.c | 112 +++++++++++++++--
> ------------
> drivers/infiniband/hw/hfi1/chip.h | 5 +
> drivers/infiniband/hw/hfi1/driver.c | 2 -
> drivers/infiniband/hw/hfi1/file_ops.c | 55 ++++++++------
> drivers/infiniband/hw/hfi1/firmware.c | 8 ++
> drivers/infiniband/hw/hfi1/hfi.h | 2 +
> drivers/infiniband/hw/hfi1/init.c | 26 +++++++
> drivers/infiniband/hw/hfi1/pio.c | 2 -
> drivers/infiniband/hw/hfi1/platform.c | 14 ++--
> drivers/infiniband/hw/hfi1/ruc.c | 6 +-
> drivers/infiniband/hw/hfi1/sdma.c | 2 -
> drivers/infiniband/hw/hfi1/user_exp_rcv.c | 4 +
> drivers/infiniband/hw/hfi1/user_exp_rcv.h | 3 +
> drivers/infiniband/hw/hfi1/user_sdma.c | 4 +
> drivers/infiniband/hw/hfi1/user_sdma.h | 3 +
> drivers/infiniband/hw/qib/qib_ruc.c | 6 +-
> drivers/infiniband/sw/rdmavt/mr.c | 8 +-
> drivers/infiniband/sw/rdmavt/qp.c | 16 ++--
> 18 files changed, 166 insertions(+), 112 deletions(-)
>
Hi Denny,
Series applied, thanks.
--
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
GPG KeyID: B826A3330E572FDD
Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 11+ messages in thread