1. Patchew
  2. linux
  3. View series

[PATCH v6] nvme: Skip trace complete_rq on host path error

전민식 posted 1 patch 1 month, 3 weeks ago 
drivers/nvme/host/core.c | 11 ++++++++---
1 file changed, 8 insertions(+), 3 deletions(-)
[PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by 전민식 1 month, 3 weeks ago 
From cfd386c2d769552c4dd32ec8ff5233ef34f9830d Mon Sep 17 00:00:00 2001
From: Minsik Jeon <hmi.jeon@samsung.com>
Date: Thu, 16 Apr 2026 10:43:00 +0900
Subject: [PATCH v6] nvme: Skip trace complete_rq on host path error

we were checking host_pathing_error before calling nvme_setup_cmd().
This is caused the command setup to be skipped entirely when a pathing
error occurred, making it impossible to trace the nvme command via
trace_cmd nvme_complete_rq().

As a result, when nvme_complete_rq() logged a completion with cmdid=0,
it was impossible to correlate the completion with the nvme command
request.

This patch Skip trace_nvme_complete_rq() on NVMe host path error.

Co-authored-by: Beomsoo Kim <beomsooo.kim@samsung.com>
Co-authored-by: Eunsoo Lee <euns212.lee@samsung.com>
Co-authored-by: Steven Seungcheol Lee <sc108.lee@samsung.com>
Signed-off-by: Minsik Jeon <hmi.jeon@samsung.com>
---
 drivers/nvme/host/core.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 1e33af94c24b..e92c97fad305 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -454,11 +454,10 @@ void nvme_end_req(struct request *req)
 	blk_mq_end_request(req, status);
 }
 
-void nvme_complete_rq(struct request *req)
+void __nvme_complete_rq(struct request *req)
 {
 	struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
 
-	trace_nvme_complete_rq(req);
 	nvme_cleanup_cmd(req);
 
 	/*
@@ -493,6 +492,12 @@ void nvme_complete_rq(struct request *req)
 		return;
 	}
 }
+
+void nvme_complete_rq(struct request *req)
+{
+	trace_nvme_complete_rq(req);
+	__nvme_complete_rq(req);
+}
 EXPORT_SYMBOL_GPL(nvme_complete_rq);
 
 void nvme_complete_batch_req(struct request *req)
@@ -513,7 +518,7 @@ blk_status_t nvme_host_path_error(struct request *req)
 {
 	nvme_req(req)->status = NVME_SC_HOST_PATH_ERROR;
 	blk_mq_set_request_complete(req);
-	nvme_complete_rq(req);
+	__nvme_complete_rq(req);
 	return BLK_STS_OK;
 }
 EXPORT_SYMBOL_GPL(nvme_host_path_error);
-- 
2.52.0
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by kernel test robot 1 month, 1 week ago 
Hi 전민식,

kernel test robot noticed the following build warnings:

[auto build test WARNING on axboe/for-next]
[also build test WARNING on linux-nvme/for-next linus/master v7.1-rc1]
[cannot apply to next-20260429]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/nvme-Skip-trace-complete_rq-on-host-path-error/20260416-180151
base:   https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux.git for-next
patch link:    https://lore.kernel.org/r/20260416010713epcms2p78575bd400f93f6edb7fe3f3ead5bf930%40epcms2p7
patch subject: [PATCH v6] nvme: Skip trace complete_rq on host path error
config: x86_64-rhel-9.4 (https://download.01.org/0day-ci/archive/20260430/202604300750.YpUGtmek-lkp@intel.com/config)
compiler: gcc-14 (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260430/202604300750.YpUGtmek-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202604300750.YpUGtmek-lkp@intel.com/

All warnings (new ones prefixed by >>):

>> drivers/nvme/host/core.c:457:6: warning: no previous prototype for '__nvme_complete_rq' [-Wmissing-prototypes]
     457 | void __nvme_complete_rq(struct request *req)
         |      ^~~~~~~~~~~~~~~~~~


vim +/__nvme_complete_rq +457 drivers/nvme/host/core.c

   456	
 > 457	void __nvme_complete_rq(struct request *req)
   458	{
   459		struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
   460	
   461		nvme_cleanup_cmd(req);
   462	
   463		/*
   464		 * Completions of long-running commands should not be able to
   465		 * defer sending of periodic keep alives, since the controller
   466		 * may have completed processing such commands a long time ago
   467		 * (arbitrarily close to command submission time).
   468		 * req->deadline - req->timeout is the command submission time
   469		 * in jiffies.
   470		 */
   471		if (ctrl->kas &&
   472		    req->deadline - req->timeout >= ctrl->ka_last_check_time)
   473			ctrl->comp_seen = true;
   474	
   475		switch (nvme_decide_disposition(req)) {
   476		case COMPLETE:
   477			nvme_end_req(req);
   478			return;
   479		case RETRY:
   480			nvme_retry_req(req);
   481			return;
   482		case FAILOVER:
   483			nvme_failover_req(req);
   484			return;
   485		case AUTHENTICATE:
   486	#ifdef CONFIG_NVME_HOST_AUTH
   487			queue_work(nvme_wq, &ctrl->dhchap_auth_work);
   488			nvme_retry_req(req);
   489	#else
   490			nvme_end_req(req);
   491	#endif
   492			return;
   493		}
   494	}
   495	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
RE: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by 전민식 1 month, 1 week ago 
On Thu, Apr 17, 2026 at 01:00 UTC, 전민식 wrote:
> On Thu, Apr 16, 2026 at 22:00 UTC, Keith Busch wrote:
> > You're missing the "static" declaration, which was present in the
> > original suggestion:
> >
> > https://lore.kernel.org/linux-nvme/acVCnozG1WKPkq1L@kbusch-mbp/
> >
> > Should I just apply my local copy? I think it's an original patch from
> > me at this point.
>
> Please go ahead and apply your local copy.
> I appreciate your help in finalizing the patch.

Hi Keith,

Are you planning to apply your local fix directly?

Thanks,
Minsik Jeon
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by Keith Busch 1 month, 1 week ago 
On Thu, Apr 30, 2026 at 02:34:01PM +0900, 전민식 wrote:
> Are you planning to apply your local fix directly?

It's already committed to the block tree and should land upstream this
week.
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by kernel test robot 1 month, 1 week ago 
Hi 전민식,

kernel test robot noticed the following build warnings:

[auto build test WARNING on axboe/for-next]
[also build test WARNING on linux-nvme/for-next hch-configfs/for-next linus/master v7.1-rc1]
[cannot apply to next-20260429]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/nvme-Skip-trace-complete_rq-on-host-path-error/20260416-180151
base:   https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux.git for-next
patch link:    https://lore.kernel.org/r/20260416010713epcms2p78575bd400f93f6edb7fe3f3ead5bf930%40epcms2p7
patch subject: [PATCH v6] nvme: Skip trace complete_rq on host path error
config: sh-allyesconfig (https://download.01.org/0day-ci/archive/20260430/202604301237.jC9FNHxS-lkp@intel.com/config)
compiler: sh4-linux-gcc (GCC) 15.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260430/202604301237.jC9FNHxS-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202604301237.jC9FNHxS-lkp@intel.com/

All warnings (new ones prefixed by >>):

>> drivers/nvme/host/core.c:457:6: warning: no previous prototype for '__nvme_complete_rq' [-Wmissing-prototypes]
     457 | void __nvme_complete_rq(struct request *req)
         |      ^~~~~~~~~~~~~~~~~~


vim +/__nvme_complete_rq +457 drivers/nvme/host/core.c

   456	
 > 457	void __nvme_complete_rq(struct request *req)
   458	{
   459		struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
   460	
   461		nvme_cleanup_cmd(req);
   462	
   463		/*
   464		 * Completions of long-running commands should not be able to
   465		 * defer sending of periodic keep alives, since the controller
   466		 * may have completed processing such commands a long time ago
   467		 * (arbitrarily close to command submission time).
   468		 * req->deadline - req->timeout is the command submission time
   469		 * in jiffies.
   470		 */
   471		if (ctrl->kas &&
   472		    req->deadline - req->timeout >= ctrl->ka_last_check_time)
   473			ctrl->comp_seen = true;
   474	
   475		switch (nvme_decide_disposition(req)) {
   476		case COMPLETE:
   477			nvme_end_req(req);
   478			return;
   479		case RETRY:
   480			nvme_retry_req(req);
   481			return;
   482		case FAILOVER:
   483			nvme_failover_req(req);
   484			return;
   485		case AUTHENTICATE:
   486	#ifdef CONFIG_NVME_HOST_AUTH
   487			queue_work(nvme_wq, &ctrl->dhchap_auth_work);
   488			nvme_retry_req(req);
   489	#else
   490			nvme_end_req(req);
   491	#endif
   492			return;
   493		}
   494	}
   495	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by kernel test robot 1 month, 1 week ago 
Hi 전민식,

kernel test robot noticed the following build warnings:

[auto build test WARNING on axboe/for-next]
[also build test WARNING on linux-nvme/for-next linus/master v7.1-rc1]
[cannot apply to next-20260429]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/nvme-Skip-trace-complete_rq-on-host-path-error/20260416-180151
base:   https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux.git for-next
patch link:    https://lore.kernel.org/r/20260416010713epcms2p78575bd400f93f6edb7fe3f3ead5bf930%40epcms2p7
patch subject: [PATCH v6] nvme: Skip trace complete_rq on host path error
config: x86_64-kexec (https://download.01.org/0day-ci/archive/20260430/202604300644.6qVuhQQC-lkp@intel.com/config)
compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260430/202604300644.6qVuhQQC-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202604300644.6qVuhQQC-lkp@intel.com/

All warnings (new ones prefixed by >>):

>> drivers/nvme/host/core.c:457:6: warning: no previous prototype for function '__nvme_complete_rq' [-Wmissing-prototypes]
     457 | void __nvme_complete_rq(struct request *req)
         |      ^
   drivers/nvme/host/core.c:457:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
     457 | void __nvme_complete_rq(struct request *req)
         | ^
         | static 
   1 warning generated.


vim +/__nvme_complete_rq +457 drivers/nvme/host/core.c

   456	
 > 457	void __nvme_complete_rq(struct request *req)
   458	{
   459		struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
   460	
   461		nvme_cleanup_cmd(req);
   462	
   463		/*
   464		 * Completions of long-running commands should not be able to
   465		 * defer sending of periodic keep alives, since the controller
   466		 * may have completed processing such commands a long time ago
   467		 * (arbitrarily close to command submission time).
   468		 * req->deadline - req->timeout is the command submission time
   469		 * in jiffies.
   470		 */
   471		if (ctrl->kas &&
   472		    req->deadline - req->timeout >= ctrl->ka_last_check_time)
   473			ctrl->comp_seen = true;
   474	
   475		switch (nvme_decide_disposition(req)) {
   476		case COMPLETE:
   477			nvme_end_req(req);
   478			return;
   479		case RETRY:
   480			nvme_retry_req(req);
   481			return;
   482		case FAILOVER:
   483			nvme_failover_req(req);
   484			return;
   485		case AUTHENTICATE:
   486	#ifdef CONFIG_NVME_HOST_AUTH
   487			queue_work(nvme_wq, &ctrl->dhchap_auth_work);
   488			nvme_retry_req(req);
   489	#else
   490			nvme_end_req(req);
   491	#endif
   492			return;
   493		}
   494	}
   495	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by Keith Busch 1 month, 3 weeks ago 
On Thu, Apr 16, 2026 at 10:07:13AM +0900, 전민식 wrote:
> Co-authored-by: Beomsoo Kim <beomsooo.kim@samsung.com>
> Co-authored-by: Eunsoo Lee <euns212.lee@samsung.com>
> Co-authored-by: Steven Seungcheol Lee <sc108.lee@samsung.com>

...

> -void nvme_complete_rq(struct request *req)
> +void __nvme_complete_rq(struct request *req)

You're missing the "static" declaration, which was present in the
original suggestion:

https://lore.kernel.org/linux-nvme/acVCnozG1WKPkq1L@kbusch-mbp/

Should I just apply my local copy? I think it's an original patch from
me at this point.
RE: Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by 전민식 1 month, 3 weeks ago 
On Thu, Apr 16, 2026 at 22:00 UTC, Keith Busch wrote:
> You're missing the "static" declaration, which was present in the
> original suggestion:
>
> https://lore.kernel.org/linux-nvme/acVCnozG1WKPkq1L@kbusch-mbp/
>
> Should I just apply my local copy? I think it's an original patch from
> me at this point.

Please go ahead and apply your local copy.
I appreciate your help in finalizing the patch.

Thanks,
Minsik Jeon
Re: [PATCH v6] nvme: Skip trace complete_rq on host path error
Posted by Chaitanya Kulkarni 1 month, 3 weeks ago 
On 4/15/26 18:07, 전민식 wrote:
>  From cfd386c2d769552c4dd32ec8ff5233ef34f9830d Mon Sep 17 00:00:00 2001
> From: Minsik Jeon<hmi.jeon@samsung.com>
> Date: Thu, 16 Apr 2026 10:43:00 +0900
> Subject: [PATCH v6] nvme: Skip trace complete_rq on host path error
>
> we were checking host_pathing_error before calling nvme_setup_cmd().
> This is caused the command setup to be skipped entirely when a pathing
> error occurred, making it impossible to trace the nvme command via
> trace_cmd nvme_complete_rq().
>
> As a result, when nvme_complete_rq() logged a completion with cmdid=0,
> it was impossible to correlate the completion with the nvme command
> request.
>
> This patch Skip trace_nvme_complete_rq() on NVMe host path error.
>
> Co-authored-by: Beomsoo Kim<beomsooo.kim@samsung.com>
> Co-authored-by: Eunsoo Lee<euns212.lee@samsung.com>
> Co-authored-by: Steven Seungcheol Lee<sc108.lee@samsung.com>
> Signed-off-by: Minsik Jeon<hmi.jeon@samsung.com>
> ---

Is it possible to send this patch using git send-email ?
Also, do you see any performance difference after this change ?
I don't expect any but it's better to document if any since
this is in the fast path.

-ck

Patchew 2.1-devel-ea86937

© 2016 - 2026 Red Hat, Inc.