All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Horia Geantă" <horia.geanta@nxp.com>
To: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	"David S. Miller" <davem@davemloft.net>,
	Aymen Sghaier <aymen.sghaier@nxp.com>,
	Li Yang <leoyang.li@nxp.com>, Roy Pledge <roy.pledge@nxp.com>,
	Madalin Bucur <madalin.bucur@nxp.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	linux-crypto@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH] crypto: caam/qi - simplify CGR allocation, freeing
Date: Mon,  8 Oct 2018 14:09:37 +0300	[thread overview]
Message-ID: <20181008110937.29198-1-horia.geanta@nxp.com> (raw)
In-Reply-To: <20181005125443.dfhd2asqktm22ney@linutronix.de>

CGRs (Congestion Groups) have to be freed by the same CPU that
initialized them.
This is why currently the driver takes special measures; however, using
set_cpus_allowed_ptr() is incorrect - as reported by Sebastian.

Instead of the generic solution of replacing set_cpus_allowed_ptr() with
work_on_cpu_safe(), we use the qman_delete_cgr_safe() QBMan API instead
of qman_delete_cgr() - which internally takes care of proper CGR
deletion.

Link: https://lkml.kernel.org/r/20181005125443.dfhd2asqktm22ney@linutronix.de
Reported-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Horia Geantă <horia.geanta@nxp.com>
---
 drivers/crypto/caam/qi.c | 43 ++++---------------------------------------
 drivers/crypto/caam/qi.h |  2 +-
 2 files changed, 5 insertions(+), 40 deletions(-)

diff --git a/drivers/crypto/caam/qi.c b/drivers/crypto/caam/qi.c
index 67f7f8c42c93..b84e6c8b1e13 100644
--- a/drivers/crypto/caam/qi.c
+++ b/drivers/crypto/caam/qi.c
@@ -83,13 +83,6 @@ EXPORT_SYMBOL(caam_congested);
 static u64 times_congested;
 #endif
 
-/*
- * CPU from where the module initialised. This is required because QMan driver
- * requires CGRs to be removed from same CPU from where they were originally
- * allocated.
- */
-static int mod_init_cpu;
-
 /*
  * This is a a cache of buffers, from which the users of CAAM QI driver
  * can allocate short (CAAM_QI_MEMCACHE_SIZE) buffers. It's faster than
@@ -492,12 +485,11 @@ void caam_drv_ctx_rel(struct caam_drv_ctx *drv_ctx)
 }
 EXPORT_SYMBOL(caam_drv_ctx_rel);
 
-int caam_qi_shutdown(struct device *qidev)
+void caam_qi_shutdown(struct device *qidev)
 {
-	int i, ret;
+	int i;
 	struct caam_qi_priv *priv = dev_get_drvdata(qidev);
 	const cpumask_t *cpus = qman_affine_cpus();
-	struct cpumask old_cpumask = current->cpus_allowed;
 
 	for_each_cpu(i, cpus) {
 		struct napi_struct *irqtask;
@@ -510,26 +502,12 @@ int caam_qi_shutdown(struct device *qidev)
 			dev_err(qidev, "Rsp FQ kill failed, cpu: %d\n", i);
 	}
 
-	/*
-	 * QMan driver requires CGRs to be deleted from same CPU from where they
-	 * were instantiated. Hence we get the module removal execute from the
-	 * same CPU from where it was originally inserted.
-	 */
-	set_cpus_allowed_ptr(current, get_cpu_mask(mod_init_cpu));
-
-	ret = qman_delete_cgr(&priv->cgr);
-	if (ret)
-		dev_err(qidev, "Deletion of CGR failed: %d\n", ret);
-	else
-		qman_release_cgrid(priv->cgr.cgrid);
+	qman_delete_cgr_safe(&priv->cgr);
+	qman_release_cgrid(priv->cgr.cgrid);
 
 	kmem_cache_destroy(qi_cache);
 
-	/* Now that we're done with the CGRs, restore the cpus allowed mask */
-	set_cpus_allowed_ptr(current, &old_cpumask);
-
 	platform_device_unregister(priv->qi_pdev);
-	return ret;
 }
 
 static void cgr_cb(struct qman_portal *qm, struct qman_cgr *cgr, int congested)
@@ -718,22 +696,11 @@ int caam_qi_init(struct platform_device *caam_pdev)
 	struct device *ctrldev = &caam_pdev->dev, *qidev;
 	struct caam_drv_private *ctrlpriv;
 	const cpumask_t *cpus = qman_affine_cpus();
-	struct cpumask old_cpumask = current->cpus_allowed;
 	static struct platform_device_info qi_pdev_info = {
 		.name = "caam_qi",
 		.id = PLATFORM_DEVID_NONE
 	};
 
-	/*
-	 * QMAN requires CGRs to be removed from same CPU+portal from where it
-	 * was originally allocated. Hence we need to note down the
-	 * initialisation CPU and use the same CPU for module exit.
-	 * We select the first CPU to from the list of portal owning CPUs.
-	 * Then we pin module init to this CPU.
-	 */
-	mod_init_cpu = cpumask_first(cpus);
-	set_cpus_allowed_ptr(current, get_cpu_mask(mod_init_cpu));
-
 	qi_pdev_info.parent = ctrldev;
 	qi_pdev_info.dma_mask = dma_get_mask(ctrldev);
 	qi_pdev = platform_device_register_full(&qi_pdev_info);
@@ -795,8 +762,6 @@ int caam_qi_init(struct platform_device *caam_pdev)
 		return -ENOMEM;
 	}
 
-	/* Done with the CGRs; restore the cpus allowed mask */
-	set_cpus_allowed_ptr(current, &old_cpumask);
 #ifdef CONFIG_DEBUG_FS
 	debugfs_create_file("qi_congested", 0444, ctrlpriv->ctl,
 			    &times_congested, &caam_fops_u64_ro);
diff --git a/drivers/crypto/caam/qi.h b/drivers/crypto/caam/qi.h
index bc421da5ad8b..f93c9c7ed430 100644
--- a/drivers/crypto/caam/qi.h
+++ b/drivers/crypto/caam/qi.h
@@ -173,7 +173,7 @@ int caam_drv_ctx_update(struct caam_drv_ctx *drv_ctx, u32 *sh_desc);
 void caam_drv_ctx_rel(struct caam_drv_ctx *drv_ctx);
 
 int caam_qi_init(struct platform_device *pdev);
-int caam_qi_shutdown(struct device *dev);
+void caam_qi_shutdown(struct device *dev);
 
 /**
  * qi_cache_alloc - Allocate buffers from CAAM-QI cache
-- 
2.16.2

  reply	other threads:[~2018-10-08 11:09 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-10-05 12:54 set_cpus_allowed_ptr() usage in FREESCALE CAAM Sebastian Andrzej Siewior
2018-10-08 11:09 ` Horia Geantă [this message]
2018-10-09 17:11   ` [PATCH] crypto: caam/qi - simplify CGR allocation, freeing Sebastian Andrzej Siewior
2018-10-25 14:05     ` Horia Geanta
2018-10-29  8:56       ` Sebastian Andrzej Siewior
2018-10-17  6:21   ` Herbert Xu

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=20181008110937.29198-1-horia.geanta@nxp.com \
    --to=horia.geanta@nxp.com \
    --cc=aymen.sghaier@nxp.com \
    --cc=bigeasy@linutronix.de \
    --cc=davem@davemloft.net \
    --cc=herbert@gondor.apana.org.au \
    --cc=leoyang.li@nxp.com \
    --cc=linux-crypto@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=madalin.bucur@nxp.com \
    --cc=peterz@infradead.org \
    --cc=roy.pledge@nxp.com \
    --cc=tglx@linutronix.de \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.