[PATCH 0/3] CAAM job ring driver cleanups

[PATCH 0/3] CAAM job ring driver cleanups

[Vakul Garg]

This patchset cleans up caam job ring driver.
The patch series needs to be applied after merging below mentioned
patch.

https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg37068.html

Vakul Garg (3):
  crypto: caam/jr - Remove spinlock for output job ring
  crypto: caam/jr - Remove redundant vars from job ring private data
  crypto: caam/jr - Remove extra memory barrier during job ring enqueue

 drivers/crypto/caam/intern.h |  3 ---
 drivers/crypto/caam/jr.c     | 17 +++++------------
 2 files changed, 5 insertions(+), 15 deletions(-)

-- 
2.13.6

[PATCH 1/3] crypto: caam/jr - Remove spinlock for output job ring

[Vakul Garg]

For each job ring pair, the output ring is processed exactly by one cpu
at a time under a tasklet context (one per ring). Therefore, there is no
need to protect a job ring's access & its private data structure using a
lock. Hence the lock can be removed.

Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
---
 drivers/crypto/caam/intern.h | 1 -
 drivers/crypto/caam/jr.c     | 5 -----
 2 files changed, 6 deletions(-)

diff --git a/drivers/crypto/caam/intern.h b/drivers/crypto/caam/intern.h
index 30d5b6c5892f..48d62e020599 100644
--- a/drivers/crypto/caam/intern.h
+++ b/drivers/crypto/caam/intern.h
@@ -56,7 +56,6 @@ struct caam_drv_private_jr {
 	u32 inpring_avail;	/* Number of free entries in input ring */
 	int head;			/* entinfo (s/w ring) head index */
 	dma_addr_t *inpring;	/* Base of input ring, alloc DMA-safe */
-	spinlock_t outlock ____cacheline_aligned; /* Output ring index lock */
 	int out_ring_read_index;	/* Output index "tail" */
 	int tail;			/* entinfo (s/w ring) tail index */
 	struct jr_outentry *outring;	/* Base of output ring, DMA-safe */
diff --git a/drivers/crypto/caam/jr.c b/drivers/crypto/caam/jr.c
index b9caa95755d1..89de7297d099 100644
--- a/drivers/crypto/caam/jr.c
+++ b/drivers/crypto/caam/jr.c
@@ -177,8 +177,6 @@ static void caam_jr_dequeue(unsigned long devarg)
 
 		head = READ_ONCE(jrp->head);
 
-		spin_lock(&jrp->outlock);
-
 		sw_idx = tail = jrp->tail;
 		hw_idx = jrp->out_ring_read_index;
 
@@ -234,8 +232,6 @@ static void caam_jr_dequeue(unsigned long devarg)
 			jrp->tail = tail;
 		}
 
-		spin_unlock(&jrp->outlock);
-
 		/* Finally, execute user's callback */
 		usercall(dev, userdesc, userstatus, userarg);
 		outring_used--;
@@ -452,7 +448,6 @@ static int caam_jr_init(struct device *dev)
 	jrp->inpring_avail = JOBR_DEPTH;
 
 	spin_lock_init(&jrp->inplock);
-	spin_lock_init(&jrp->outlock);
 
 	/* Select interrupt coalescing parameters */
 	clrsetbits_32(&jrp->rregs->rconfig_lo, 0, JOBR_INTC |
-- 
2.13.6

Re: [PATCH 1/3] crypto: caam/jr - Remove spinlock for output job ring

[Chris Spencer]

Hi Vakul,

On Fri, 15 Mar 2019 at 04:05, Vakul Garg <vakul.garg@nxp.com> wrote:
> For each job ring pair, the output ring is processed exactly by one cpu
> at a time under a tasklet context (one per ring). Therefore, there is no
> need to protect a job ring's access & its private data structure using a
> lock. Hence the lock can be removed.
>
> Signed-off-by: Vakul Garg <vakul.garg@nxp.com>

This comment in caam_jr_dequeue doesn't really make sense anymore:
/* Stash callback params for use outside of lock */

Since the lock has been removed you could probably get away with just
invoking the user callback immediately which would simplify the code a
little.

Thanks,
Chris

[PATCH 2/3] crypto: caam/jr - Remove redundant vars from job ring private data

[Vakul Garg]

For each job ring, the variable 'ringsize' is initialised but never
used. Similarly variables 'inp_ring_read_index' and 'head' always track
the same value and instead of 'inp_ring_read_index', caam_jr_enqueue()
can use 'head' itself. Both these variables have been removed.

Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
---
 drivers/crypto/caam/intern.h | 2 --
 drivers/crypto/caam/jr.c     | 6 +-----
 2 files changed, 1 insertion(+), 7 deletions(-)

diff --git a/drivers/crypto/caam/intern.h b/drivers/crypto/caam/intern.h
index 48d62e020599..3392615dc91b 100644
--- a/drivers/crypto/caam/intern.h
+++ b/drivers/crypto/caam/intern.h
@@ -49,10 +49,8 @@ struct caam_drv_private_jr {
 	atomic_t tfm_count ____cacheline_aligned;
 
 	/* Job ring info */
-	int ringsize;	/* Size of rings (assume input = output) */
 	struct caam_jrentry_info *entinfo;	/* Alloc'ed 1 per ring entry */
 	spinlock_t inplock ____cacheline_aligned; /* Input ring index lock */
-	int inp_ring_write_index;	/* Input index "tail" */
 	u32 inpring_avail;	/* Number of free entries in input ring */
 	int head;			/* entinfo (s/w ring) head index */
 	dma_addr_t *inpring;	/* Base of input ring, alloc DMA-safe */
diff --git a/drivers/crypto/caam/jr.c b/drivers/crypto/caam/jr.c
index 89de7297d099..02f02fa996da 100644
--- a/drivers/crypto/caam/jr.c
+++ b/drivers/crypto/caam/jr.c
@@ -358,7 +358,7 @@ int caam_jr_enqueue(struct device *dev, u32 *desc,
 	head_entry->cbkarg = areq;
 	head_entry->desc_addr_dma = desc_dma;
 
-	jrp->inpring[jrp->inp_ring_write_index] = cpu_to_caam_dma(desc_dma);
+	jrp->inpring[head] = cpu_to_caam_dma(desc_dma);
 
 	/*
 	 * Guarantee that the descriptor's DMA address has been written to
@@ -367,8 +367,6 @@ int caam_jr_enqueue(struct device *dev, u32 *desc,
 	 */
 	smp_wmb();
 
-	jrp->inp_ring_write_index = (jrp->inp_ring_write_index + 1) &
-				    (JOBR_DEPTH - 1);
 	jrp->head = (head + 1) & (JOBR_DEPTH - 1);
 
 	/*
@@ -434,7 +432,6 @@ static int caam_jr_init(struct device *dev)
 		jrp->entinfo[i].desc_addr_dma = !0;
 
 	/* Setup rings */
-	jrp->inp_ring_write_index = 0;
 	jrp->out_ring_read_index = 0;
 	jrp->head = 0;
 	jrp->tail = 0;
@@ -444,7 +441,6 @@ static int caam_jr_init(struct device *dev)
 	wr_reg32(&jrp->rregs->inpring_size, JOBR_DEPTH);
 	wr_reg32(&jrp->rregs->outring_size, JOBR_DEPTH);
 
-	jrp->ringsize = JOBR_DEPTH;
 	jrp->inpring_avail = JOBR_DEPTH;
 
 	spin_lock_init(&jrp->inplock);
-- 
2.13.6

[PATCH 3/3] crypto: caam/jr - Remove extra memory barrier during job ring enqueue

[Vakul Garg]

In caam_jr_enqueue(), a write barrier is needed to order stores to job
ring slot before declaring addition of new job into input job ring.
The register write is done using wr_reg32() which internally uses
iowrite32() for write operation. The api iowrite32() issues a write
barrier before issuing write operation. Therefore, the wmb() preceding
wr_reg32() can be safely removed.

Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
---
 drivers/crypto/caam/jr.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/crypto/caam/jr.c b/drivers/crypto/caam/jr.c
index 02f02fa996da..3bd28386eb07 100644
--- a/drivers/crypto/caam/jr.c
+++ b/drivers/crypto/caam/jr.c
@@ -371,9 +371,11 @@ int caam_jr_enqueue(struct device *dev, u32 *desc,
 
 	/*
 	 * Ensure that all job information has been written before
-	 * notifying CAAM that a new job was added to the input ring.
+	 * notifying CAAM that a new job was added to the input ring
+	 * using a memory barrier. The wr_reg32() uses api iowrite32()
+	 * to do the register write. iowrite32() issues a memory barrier
+	 * before the write operation.
 	 */
-	wmb();
 
 	wr_reg32(&jrp->rregs->inpring_jobadd, 1);
 
-- 
2.13.6