Development discussion of WireGuard
 help / color / mirror / Atom feed
From: Justin Tee <justintee8345@gmail.com>
To: Vernon Yang <vernon2gm@gmail.com>
Cc: torvalds@linux-foundation.org, tytso@mit.edu, Jason@zx2c4.com,
	 davem@davemloft.net, edumazet@google.com, kuba@kernel.org,
	pabeni@redhat.com,  jejb@linux.ibm.com,
	martin.petersen@oracle.com, yury.norov@gmail.com,
	 andriy.shevchenko@linux.intel.com, linux@rasmusvillemoes.dk,
	 james.smart@broadcom.com, dick.kennedy@broadcom.com,
	 linux-kernel@vger.kernel.org, wireguard@lists.zx2c4.com,
	 netdev@vger.kernel.org, linux-scsi@vger.kernel.org,
	 Justin Tee <justin.tee@broadcom.com>
Subject: Re: [PATCH v2 3/4] scsi: lpfc: fix lpfc_cpu_affinity_check() if no further cpus set
Date: Tue, 7 Mar 2023 16:40:28 -0800	[thread overview]
Message-ID: <CABPRKS-DRvmZpoXBTogjEf5Q+oL5GR71nSVRgmVo5vWx-9OQow@mail.gmail.com> (raw)
In-Reply-To: <20230306210312.2614988-4-vernon2gm@gmail.com>

Hi Vernon,

Is it possible to move the new helper routine lpfc_next_present_cpu
into the lpfc.h header file around where lpfc_next_online_cpu is
defined?

Also, with slight modifications we could use the new
lpfc_next_present_cpu helper routine for the
lpfc_nvmet_setup_io_context() patch as well so that we can contain all
lpfc changes within one patch.

---
 drivers/scsi/lpfc/lpfc.h       | 20 ++++++++++++++++++++
 drivers/scsi/lpfc/lpfc_init.c  | 31 +++++++------------------------
 drivers/scsi/lpfc/lpfc_nvmet.c |  5 +----
 3 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/drivers/scsi/lpfc/lpfc.h b/drivers/scsi/lpfc/lpfc.h
index cf55f8e3bd9f..f342d6bc5726 100644
--- a/drivers/scsi/lpfc/lpfc.h
+++ b/drivers/scsi/lpfc/lpfc.h
@@ -1737,6 +1737,26 @@ lpfc_next_online_cpu(const struct cpumask
*mask, unsigned int start)

     return cpu_it;
 }
+
+/**
+ * lpfc_next_present_cpu - Finds next present CPU after n
+ * @n: the cpu prior to search
+ *
+ * Note: If no next present cpu, then fallback to first present cpu.
+ *
+ **/
+static inline unsigned int lpfc_next_present_cpu(int n)
+{
+    unsigned int cpu;
+
+    cpu = cpumask_next(n, cpu_present_mask);
+
+    if (cpu >= nr_cpu_ids)
+        cpu = cpumask_first(cpu_present_mask);
+
+    return cpu;
+}
+
 /**
  * lpfc_sli4_mod_hba_eq_delay - update EQ delay
  * @phba: Pointer to HBA context object.
diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
index 3e1e1d17b2b4..f28af338341f 100644
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -12560,10 +12560,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba
*phba, int vectors)
                     (new_cpup->eq != LPFC_VECTOR_MAP_EMPTY) &&
                     (new_cpup->phys_id == cpup->phys_id))
                     goto found_same;
-                new_cpu = cpumask_next(
-                    new_cpu, cpu_present_mask);
-                if (new_cpu == nr_cpumask_bits)
-                    new_cpu = first_cpu;
+                new_cpu = lpfc_next_present_cpu(new_cpu);
             }
             /* At this point, we leave the CPU as unassigned */
             continue;
@@ -12575,9 +12572,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba,
int vectors)
              * chance of having multiple unassigned CPU entries
              * selecting the same IRQ.
              */
-            start_cpu = cpumask_next(new_cpu, cpu_present_mask);
-            if (start_cpu == nr_cpumask_bits)
-                start_cpu = first_cpu;
+            start_cpu = lpfc_next_present_cpu(new_cpu);

             lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
                     "3337 Set Affinity: CPU %d "
@@ -12610,10 +12605,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba
*phba, int vectors)
                 if (!(new_cpup->flag & LPFC_CPU_MAP_UNASSIGN) &&
                     (new_cpup->eq != LPFC_VECTOR_MAP_EMPTY))
                     goto found_any;
-                new_cpu = cpumask_next(
-                    new_cpu, cpu_present_mask);
-                if (new_cpu == nr_cpumask_bits)
-                    new_cpu = first_cpu;
+                new_cpu = lpfc_next_present_cpu(new_cpu);
             }
             /* We should never leave an entry unassigned */
             lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
@@ -12629,9 +12621,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba,
int vectors)
              * chance of having multiple unassigned CPU entries
              * selecting the same IRQ.
              */
-            start_cpu = cpumask_next(new_cpu, cpu_present_mask);
-            if (start_cpu == nr_cpumask_bits)
-                start_cpu = first_cpu;
+            start_cpu = lpfc_next_present_cpu(new_cpu);

             lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
                     "3338 Set Affinity: CPU %d "
@@ -12702,9 +12692,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba,
int vectors)
                 new_cpup->core_id == cpup->core_id) {
                 goto found_hdwq;
             }
-            new_cpu = cpumask_next(new_cpu, cpu_present_mask);
-            if (new_cpu == nr_cpumask_bits)
-                new_cpu = first_cpu;
+            new_cpu = lpfc_next_present_cpu(new_cpu);
         }

         /* If we can't match both phys_id and core_id,
@@ -12716,10 +12704,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba
*phba, int vectors)
             if (new_cpup->hdwq != LPFC_VECTOR_MAP_EMPTY &&
                 new_cpup->phys_id == cpup->phys_id)
                 goto found_hdwq;
-
-            new_cpu = cpumask_next(new_cpu, cpu_present_mask);
-            if (new_cpu == nr_cpumask_bits)
-                new_cpu = first_cpu;
+            new_cpu = lpfc_next_present_cpu(new_cpu);
         }

         /* Otherwise just round robin on cfg_hdw_queue */
@@ -12728,9 +12713,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba,
int vectors)
         goto logit;
  found_hdwq:
         /* We found an available entry, copy the IRQ info */
-        start_cpu = cpumask_next(new_cpu, cpu_present_mask);
-        if (start_cpu == nr_cpumask_bits)
-            start_cpu = first_cpu;
+        start_cpu = lpfc_next_present_cpu(new_cpu);
         cpup->hdwq = new_cpup->hdwq;
  logit:
         lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
diff --git a/drivers/scsi/lpfc/lpfc_nvmet.c b/drivers/scsi/lpfc/lpfc_nvmet.c
index 7517dd55fe91..2d8ac2ceb6f3 100644
--- a/drivers/scsi/lpfc/lpfc_nvmet.c
+++ b/drivers/scsi/lpfc/lpfc_nvmet.c
@@ -1620,10 +1620,7 @@ lpfc_nvmet_setup_io_context(struct lpfc_hba *phba)
             cpu = cpumask_first(cpu_present_mask);
             continue;
         }
-        cpu = cpumask_next(cpu, cpu_present_mask);
-        if (cpu == nr_cpu_ids)
-            cpu = cpumask_first(cpu_present_mask);
-
+        cpu = lpfc_next_present_cpu(cpu);
     }

     for_each_present_cpu(i) {
-- 
2.38.0

Thanks,
Justin

On Mon, Mar 6, 2023 at 1:10 PM Vernon Yang <vernon2gm@gmail.com> wrote:
>
> When cpumask_next() the return value is greater than or equal to
> nr_cpu_ids, it indicates invalid.
>
> Before commit 596ff4a09b89 ("cpumask: re-introduce constant-sized cpumask
> optimizations"), when cpumask_next() returned an invalid cpu, the driver
> used the judgment equal to nr_cpu_ids to indicate the invalid cpu, so it
> happened to work normally, but this is the wrong approach.
>
> After commit 596ff4a09b89 ("cpumask: re-introduce constant-sized cpumask
> optimizations"), these incorrect practices actively buggy, so fix it to
> correctly.
>
> Signed-off-by: Vernon Yang <vernon2gm@gmail.com>
> ---
>  drivers/scsi/lpfc/lpfc_init.c | 43 ++++++++++++++++-------------------
>  1 file changed, 20 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
> index 61958a24a43d..acfffdbe9ba1 100644
> --- a/drivers/scsi/lpfc/lpfc_init.c
> +++ b/drivers/scsi/lpfc/lpfc_init.c
> @@ -12473,6 +12473,16 @@ lpfc_hba_eq_hdl_array_init(struct lpfc_hba *phba)
>         }
>  }
>
> +static inline int lpfc_next_present_cpu(int n, int first_cpu)
> +{
> +       n = cpumask_next(n, cpu_present_mask);
> +
> +       if (n >= nr_cpu_ids)
> +               n = first_cpu;
> +
> +       return n;
> +}
> +
>  /**
>   * lpfc_cpu_affinity_check - Check vector CPU affinity mappings
>   * @phba: pointer to lpfc hba data structure.
> @@ -12561,10 +12571,8 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                                     (new_cpup->eq != LPFC_VECTOR_MAP_EMPTY) &&
>                                     (new_cpup->phys_id == cpup->phys_id))
>                                         goto found_same;
> -                               new_cpu = cpumask_next(
> -                                       new_cpu, cpu_present_mask);
> -                               if (new_cpu == nr_cpumask_bits)
> -                                       new_cpu = first_cpu;
> +
> +                               new_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>                         }
>                         /* At this point, we leave the CPU as unassigned */
>                         continue;
> @@ -12576,9 +12584,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                          * chance of having multiple unassigned CPU entries
>                          * selecting the same IRQ.
>                          */
> -                       start_cpu = cpumask_next(new_cpu, cpu_present_mask);
> -                       if (start_cpu == nr_cpumask_bits)
> -                               start_cpu = first_cpu;
> +                       start_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>
>                         lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
>                                         "3337 Set Affinity: CPU %d "
> @@ -12611,10 +12617,8 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                                 if (!(new_cpup->flag & LPFC_CPU_MAP_UNASSIGN) &&
>                                     (new_cpup->eq != LPFC_VECTOR_MAP_EMPTY))
>                                         goto found_any;
> -                               new_cpu = cpumask_next(
> -                                       new_cpu, cpu_present_mask);
> -                               if (new_cpu == nr_cpumask_bits)
> -                                       new_cpu = first_cpu;
> +
> +                               new_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>                         }
>                         /* We should never leave an entry unassigned */
>                         lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
> @@ -12630,9 +12634,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                          * chance of having multiple unassigned CPU entries
>                          * selecting the same IRQ.
>                          */
> -                       start_cpu = cpumask_next(new_cpu, cpu_present_mask);
> -                       if (start_cpu == nr_cpumask_bits)
> -                               start_cpu = first_cpu;
> +                       start_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>
>                         lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
>                                         "3338 Set Affinity: CPU %d "
> @@ -12703,9 +12705,8 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                             new_cpup->core_id == cpup->core_id) {
>                                 goto found_hdwq;
>                         }
> -                       new_cpu = cpumask_next(new_cpu, cpu_present_mask);
> -                       if (new_cpu == nr_cpumask_bits)
> -                               new_cpu = first_cpu;
> +
> +                       new_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>                 }
>
>                 /* If we can't match both phys_id and core_id,
> @@ -12718,9 +12719,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                             new_cpup->phys_id == cpup->phys_id)
>                                 goto found_hdwq;
>
> -                       new_cpu = cpumask_next(new_cpu, cpu_present_mask);
> -                       if (new_cpu == nr_cpumask_bits)
> -                               new_cpu = first_cpu;
> +                       new_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>                 }
>
>                 /* Otherwise just round robin on cfg_hdw_queue */
> @@ -12729,9 +12728,7 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
>                 goto logit;
>   found_hdwq:
>                 /* We found an available entry, copy the IRQ info */
> -               start_cpu = cpumask_next(new_cpu, cpu_present_mask);
> -               if (start_cpu == nr_cpumask_bits)
> -                       start_cpu = first_cpu;
> +               start_cpu = lpfc_next_present_cpu(new_cpu, first_cpu);
>                 cpup->hdwq = new_cpup->hdwq;
>   logit:
>                 lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
> --
> 2.34.1
>

  reply	other threads:[~2023-03-10 10:46 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-06 21:03 [PATCH v2 0/4] fix call cpumask_next() " Vernon Yang
2023-03-06 21:03 ` [PATCH v2 1/4] random: fix try_to_generate_entropy() " Vernon Yang
2023-03-06 21:03 ` [PATCH v2 2/4] wireguard: fix wg_cpumask_choose_online() " Vernon Yang
2023-03-06 21:03 ` [PATCH v2 3/4] scsi: lpfc: fix lpfc_cpu_affinity_check() " Vernon Yang
2023-03-08  0:40   ` Justin Tee [this message]
2023-03-06 21:03 ` [PATCH v2 4/4] scsi: lpfc: fix lpfc_nvmet_setup_io_context() " Vernon Yang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CABPRKS-DRvmZpoXBTogjEf5Q+oL5GR71nSVRgmVo5vWx-9OQow@mail.gmail.com \
    --to=justintee8345@gmail.com \
    --cc=Jason@zx2c4.com \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=davem@davemloft.net \
    --cc=dick.kennedy@broadcom.com \
    --cc=edumazet@google.com \
    --cc=james.smart@broadcom.com \
    --cc=jejb@linux.ibm.com \
    --cc=justin.tee@broadcom.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=linux@rasmusvillemoes.dk \
    --cc=martin.petersen@oracle.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=torvalds@linux-foundation.org \
    --cc=tytso@mit.edu \
    --cc=vernon2gm@gmail.com \
    --cc=wireguard@lists.zx2c4.com \
    --cc=yury.norov@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).