Thread (6 messages) 6 messages, 2 authors, 2021-11-17

Re: [PATCH v2, 1/1] mailbox: cmdq: add instruction time-out interrupt support

From: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Date: 2021-10-08 14:56:14
Also in: linux-devicetree, linux-mediatek, lkml

Hi, Yongqiang:

yongqiang.niu [off-list ref] 於 2021年10月8日 週五 上午9:49寫道:
On Tue, 2021-10-05 at 07:41 +0800, Chun-Kuang Hu wrote:
quoted
Hi, Yongqiang:

Yongqiang Niu [off-list ref] 於 2021年9月30日 週四 下午9:18寫道:
quoted
add time-out cycle setting to make sure time-out interrupt irq
will happened when instruction time-out for wait and poll

Signed-off-by: Yongqiang Niu <redacted>
---
 drivers/mailbox/mtk-cmdq-mailbox.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
diff --git a/drivers/mailbox/mtk-cmdq-mailbox.c
b/drivers/mailbox/mtk-cmdq-mailbox.c
index 64175a893312..197b03222f94 100644
--- a/drivers/mailbox/mtk-cmdq-mailbox.c
+++ b/drivers/mailbox/mtk-cmdq-mailbox.c
@@ -36,6 +36,7 @@
 #define CMDQ_THR_END_ADDR              0x24
 #define CMDQ_THR_WAIT_TOKEN            0x30
 #define CMDQ_THR_PRIORITY              0x40
+#define CMDQ_THR_INSTN_TIMEOUT_CYCLES  0x50

 #define GCE_GCTL_VALUE                 0x48
@@ -54,6 +55,15 @@
 #define CMDQ_JUMP_BY_OFFSET            0x10000000
 #define CMDQ_JUMP_BY_PA                        0x10000001

+/*
+ * instruction time-out
+ * cycles to issue instruction time-out interrupt for wait and
poll instructions
+ * GCE axi_clock 156MHz
+ * 1 cycle = 6.41ns
+ * instruction time out 2^22*2*6.41ns = 53ms
For different clients, the timeout value would be different, and each
client could use timer to detect timeout, so it's not necessary to
enable timeout in cmdq driver.

Regards,
Chun-Kuang.
if we do not set cmdq hardware timeout, this condition will never
happen
cmdq_thread_irq_handler
if (irq_flag & CMDQ_THR_IRQ_ERROR)
                err = true;

and no error callback
else if (err) {
                        cmdq_task_exec_done(task, -ENOEXEC);
                        cmdq_task_handle_error(curr_task);
                        kfree(task);
                }
the client will never received the error callback, cmdq hardware will
poll the event for ever and no report timeout
I think there are two way to implement the timeout mechanism. The
first way is to use the GCE hardware to detect timeout. The second way
is that client driver use timer to detect timeout, when it's timeout,
use mbox_flush() to clean up the packets in mtk cmdq driver, and
remove the error handle in irq handler.
If you think the first way is better, I think you should pass the
timeout value from client driver to cmdq driver because each client
driver has different timeout value. And the GCE clock may be different
in each SoC, so use clk_get_rate() to get the clock frequency for
different SoC.

Regards,
Chun-Kuang.
quoted
quoted
+ */
+#define CMDQ_INSTN_TIMEOUT_CYCLES      22
+
 struct cmdq_thread {
        struct mbox_chan        *chan;
        void __iomem            *base;
@@ -376,6 +386,7 @@ static int cmdq_mbox_send_data(struct mbox_chan
*chan, void *data)
                writel((task->pa_base + pkt->cmd_buf_size) >> cmdq-
quoted
shift_pa,
                       thread->base + CMDQ_THR_END_ADDR);

+               writel(CMDQ_INSTN_TIMEOUT_CYCLES, thread->base +
CMDQ_THR_INSTN_TIMEOUT_CYCLES);
                writel(thread->priority, thread->base +
CMDQ_THR_PRIORITY);
                writel(CMDQ_THR_IRQ_EN, thread->base +
CMDQ_THR_IRQ_ENABLE);
                writel(CMDQ_THR_ENABLED, thread->base +
CMDQ_THR_ENABLE_TASK);
--
2.25.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help