Running TCRYPT with LRW compiled causes spinlock recursion:

    testing speed of async lrw(aes) (lrw(ecb-aes-s5p)) encryption
    tcrypt: test 0 (256 bit key, 16 byte blocks): 19007 operations in 1 seconds 
(304112 bytes)
    tcrypt: test 1 (256 bit key, 64 byte blocks): 15753 operations in 1 seconds 
(1008192 bytes)
    tcrypt: test 2 (256 bit key, 256 byte blocks): 14293 operations in 1 
seconds (3659008 bytes)
    tcrypt: test 3 (256 bit key, 1024 byte blocks): 11906 operations in 1 
seconds (12191744 bytes)
    tcrypt: test 4 (256 bit key, 8192 byte blocks):
    BUG: spinlock recursion on CPU#1, irq/84-10830000/89
     lock: 0xeea99a68, .magic: dead4ead, .owner: irq/84-10830000/89, 
.owner_cpu: 1
    CPU: 1 PID: 89 Comm: irq/84-10830000 Not tainted 
4.11.0-rc1-00001-g897ca6d0800d #559
    Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
    [<c010e1ec>] (unwind_backtrace) from [<c010ae1c>] (show_stack+0x10/0x14)
    [<c010ae1c>] (show_stack) from [<c03449c0>] (dump_stack+0x78/0x8c)
    [<c03449c0>] (dump_stack) from [<c015de68>] (do_raw_spin_lock+0x11c/0x120)
    [<c015de68>] (do_raw_spin_lock) from [<c0720110>] 
(_raw_spin_lock_irqsave+0x20/0x28)
    [<c0720110>] (_raw_spin_lock_irqsave) from [<c0572ca0>] 
(s5p_aes_crypt+0x2c/0xb4)
    [<c0572ca0>] (s5p_aes_crypt) from [<bf1d8aa4>] (do_encrypt+0x78/0xb0 [lrw])
    [<bf1d8aa4>] (do_encrypt [lrw]) from [<bf1d8b00>] (encrypt_done+0x24/0x54 
[lrw])
    [<bf1d8b00>] (encrypt_done [lrw]) from [<c05732a0>] 
(s5p_aes_complete+0x60/0xcc)
    [<c05732a0>] (s5p_aes_complete) from [<c0573440>] 
(s5p_aes_interrupt+0x134/0x1a0)
    [<c0573440>] (s5p_aes_interrupt) from [<c01667c4>] (irq_thread_fn+0x1c/0x54)
    [<c01667c4>] (irq_thread_fn) from [<c0166a98>] (irq_thread+0x12c/0x1e0)
    [<c0166a98>] (irq_thread) from [<c0136a28>] (kthread+0x108/0x138)
    [<c0136a28>] (kthread) from [<c0107778>] (ret_from_fork+0x14/0x3c)

Interrupt handling routine was calling req->base.complete() under
spinlock.  In most cases this wasn't fatal but when combined with some
of the cipher modes (like LRW) this caused recursion - starting the new
encryption (s5p_aes_crypt()) while still holding the spinlock from
previous round (s5p_aes_complete()).

Beside that, the s5p_aes_interrupt() error handling path could execute
two completions in case of error for RX and TX blocks.

Rewrite the interrupt handling routine and the completion by:

1. Splitting the operations on scatterlist copies from
   s5p_aes_complete() into separate s5p_sg_done(). This still should be
   done under lock.
   The s5p_aes_complete() now only calls req->base.complete() and it has
   to be called outside of lock.

2. Moving the s5p_aes_complete() out of spinlock critical sections.
   In interrupt service routine s5p_aes_interrupts(), it appeared in few
   places, including error paths inside other functions called from ISR.
   This code was not so obvious to read so simplify it by putting the
   s5p_aes_complete() only within ISR level.

Reported-by: Nathan Royce <nroycea+ker...@gmail.com>
Cc: <sta...@vger.kernel.org> # v4.10.x: 07de4bc88c crypto: s5p-sss - Fix 
completing
Cc: <sta...@vger.kernel.org> # v4.10.x
Signed-off-by: Krzysztof Kozlowski <k...@kernel.org>

---

I think, along with 07de4bc88c this should be backported to v4.10 as it
happens there.

Signed-off-by: Krzysztof Kozlowski <k...@kernel.org>
---
 drivers/crypto/s5p-sss.c | 127 ++++++++++++++++++++++++++++++-----------------
 1 file changed, 82 insertions(+), 45 deletions(-)

diff --git a/drivers/crypto/s5p-sss.c b/drivers/crypto/s5p-sss.c
index a668286d62cb..1b9da3dc799b 100644
--- a/drivers/crypto/s5p-sss.c
+++ b/drivers/crypto/s5p-sss.c
@@ -270,7 +270,7 @@ static void s5p_sg_copy_buf(void *buf, struct scatterlist 
*sg,
        scatterwalk_done(&walk, out, 0);
 }
 
-static void s5p_aes_complete(struct s5p_aes_dev *dev, int err)
+static void s5p_sg_done(struct s5p_aes_dev *dev)
 {
        if (dev->sg_dst_cpy) {
                dev_dbg(dev->dev,
@@ -281,8 +281,11 @@ static void s5p_aes_complete(struct s5p_aes_dev *dev, int 
err)
        }
        s5p_free_sg_cpy(dev, &dev->sg_src_cpy);
        s5p_free_sg_cpy(dev, &dev->sg_dst_cpy);
+}
 
-       /* holding a lock outside */
+/* Calls the completion. Cannot be called with dev->lock hold. */
+static void s5p_aes_complete(struct s5p_aes_dev *dev, int err)
+{
        dev->req->base.complete(&dev->req->base, err);
        dev->busy = false;
 }
@@ -368,51 +371,44 @@ static int s5p_set_indata(struct s5p_aes_dev *dev, struct 
scatterlist *sg)
 }
 
 /*
- * Returns true if new transmitting (output) data is ready and its
- * address+length have to be written to device (by calling
- * s5p_set_dma_outdata()). False otherwise.
+ * Returns -ERRNO on error (mapping of new data failed).
+ * On success returns:
+ *  - 0 if there is no more data,
+ *  - 1 if new transmitting (output) data is ready and its address+length
+ *     have to be written to device (by calling s5p_set_dma_outdata()).
  */
-static bool s5p_aes_tx(struct s5p_aes_dev *dev)
+static int s5p_aes_tx(struct s5p_aes_dev *dev)
 {
-       int err = 0;
-       bool ret = false;
+       int ret = 0;
 
        s5p_unset_outdata(dev);
 
        if (!sg_is_last(dev->sg_dst)) {
-               err = s5p_set_outdata(dev, sg_next(dev->sg_dst));
-               if (err)
-                       s5p_aes_complete(dev, err);
-               else
-                       ret = true;
-       } else {
-               s5p_aes_complete(dev, err);
-
-               dev->busy = true;
-               tasklet_schedule(&dev->tasklet);
+               ret = s5p_set_outdata(dev, sg_next(dev->sg_dst));
+               if (!ret)
+                       ret = 1;
        }
 
        return ret;
 }
 
 /*
- * Returns true if new receiving (input) data is ready and its
- * address+length have to be written to device (by calling
- * s5p_set_dma_indata()). False otherwise.
+ * Returns -ERRNO on error (mapping of new data failed).
+ * On success returns:
+ *  - 0 if there is no more data,
+ *  - 1 if new receiving (input) data is ready and its address+length
+ *     have to be written to device (by calling s5p_set_dma_indata()).
  */
-static bool s5p_aes_rx(struct s5p_aes_dev *dev)
+static int s5p_aes_rx(struct s5p_aes_dev *dev/*, bool *set_dma*/)
 {
-       int err;
-       bool ret = false;
+       int ret = 0;
 
        s5p_unset_indata(dev);
 
        if (!sg_is_last(dev->sg_src)) {
-               err = s5p_set_indata(dev, sg_next(dev->sg_src));
-               if (err)
-                       s5p_aes_complete(dev, err);
-               else
-                       ret = true;
+               ret = s5p_set_indata(dev, sg_next(dev->sg_src));
+               if (!ret)
+                       ret = 1;
        }
 
        return ret;
@@ -422,33 +418,73 @@ static irqreturn_t s5p_aes_interrupt(int irq, void 
*dev_id)
 {
        struct platform_device *pdev = dev_id;
        struct s5p_aes_dev *dev = platform_get_drvdata(pdev);
-       bool set_dma_tx = false;
-       bool set_dma_rx = false;
+       int err_dma_tx = 0;
+       int err_dma_rx = 0;
+       bool tx_end = false;
        unsigned long flags;
        uint32_t status;
+       int err;
 
        spin_lock_irqsave(&dev->lock, flags);
 
+       /*
+        * Handle rx or tx interrupt. If there is still data (scatterlist did 
not
+        * reach end), then map next scatterlist entry.
+        * In case of such mapping error, s5p_aes_complete() should be called.
+        *
+        * If there is no more data in tx scatter list, call s5p_aes_complete()
+        * and schedule new tasklet.
+        */
        status = SSS_READ(dev, FCINTSTAT);
        if (status & SSS_FCINTSTAT_BRDMAINT)
-               set_dma_rx = s5p_aes_rx(dev);
-       if (status & SSS_FCINTSTAT_BTDMAINT)
-               set_dma_tx = s5p_aes_tx(dev);
+               err_dma_rx = s5p_aes_rx(dev);
+
+       if (status & SSS_FCINTSTAT_BTDMAINT) {
+               if (sg_is_last(dev->sg_dst))
+                       tx_end = true;
+               err_dma_tx = s5p_aes_tx(dev);
+       }
 
        SSS_WRITE(dev, FCINTPEND, status);
 
-       /*
-        * Writing length of DMA block (either receiving or transmitting)
-        * will start the operation immediately, so this should be done
-        * at the end (even after clearing pending interrupts to not miss the
-        * interrupt).
-        */
-       if (set_dma_tx)
-               s5p_set_dma_outdata(dev, dev->sg_dst);
-       if (set_dma_rx)
-               s5p_set_dma_indata(dev, dev->sg_src);
+       if (err_dma_rx < 0) {
+               err = err_dma_rx;
+               goto error;
+       }
+       if (err_dma_tx < 0) {
+               err = err_dma_tx;
+               goto error;
+       }
+
+       if (tx_end) {
+               s5p_sg_done(dev);
+
+               spin_unlock_irqrestore(&dev->lock, flags);
+
+               s5p_aes_complete(dev, 0);
+               dev->busy = true;
+               tasklet_schedule(&dev->tasklet);
+       } else {
+               /*
+                * Writing length of DMA block (either receiving or
+                * transmitting) will start the operation immediately, so this
+                * should be done at the end (even after clearing pending
+                * interrupts to not miss the interrupt).
+                */
+               if (err_dma_tx == 1)
+                       s5p_set_dma_outdata(dev, dev->sg_dst);
+               if (err_dma_rx == 1)
+                       s5p_set_dma_indata(dev, dev->sg_src);
 
+               spin_unlock_irqrestore(&dev->lock, flags);
+       }
+
+       return IRQ_HANDLED;
+
+error:
+       s5p_sg_done(dev);
        spin_unlock_irqrestore(&dev->lock, flags);
+       s5p_aes_complete(dev, err);
 
        return IRQ_HANDLED;
 }
@@ -597,8 +633,9 @@ static void s5p_aes_crypt_start(struct s5p_aes_dev *dev, 
unsigned long mode)
        s5p_unset_indata(dev);
 
 indata_error:
-       s5p_aes_complete(dev, err);
+       s5p_sg_done(dev);
        spin_unlock_irqrestore(&dev->lock, flags);
+       s5p_aes_complete(dev, err);
 }
 
 static void s5p_tasklet_cb(unsigned long data)
-- 
2.9.3

Reply via email to