2024-03-07 01:07:39

by Jijie Shao

Subject: [PATCH net 6/8] net: hns3: fix reset timeout under full functions and queues

From: Peiyang Wang <[email protected]>

The cmdq reset command times out when all VFs are enabled and the queue is
full. The hardware processing time exceeds the timeout set by the driver.
In order to avoid the above extreme situations, the driver extends the
reset timeout to 1 second.

Signed-off-by: Peiyang Wang <[email protected]>
Signed-off-by: Jijie Shao <[email protected]>
---
.../net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.c | 2 +-
.../net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.c b/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.c
index d92ad6082d8e..652d71326231 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.c
@@ -351,7 +351,7 @@ static int hclge_comm_cmd_csq_done(struct hclge_comm_hw *hw)
static u32 hclge_get_cmdq_tx_timeout(u16 opcode, u32 tx_timeout)
{
static const struct hclge_cmdq_tx_timeout_map cmdq_tx_timeout_map[] = {
- {HCLGE_OPC_CFG_RST_TRIGGER, HCLGE_COMM_CMDQ_TX_TIMEOUT_500MS},
+ {HCLGE_OPC_CFG_RST_TRIGGER, HCLGE_COMM_CMDQ_CFG_RST_TIMEOUT},
};
u32 i;

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.h b/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.h
index 533c19d25e4f..552396518e08 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_common/hclge_comm_cmd.h
@@ -55,7 +55,7 @@
#define HCLGE_COMM_NIC_CMQ_DESC_NUM_S 3
#define HCLGE_COMM_NIC_CMQ_DESC_NUM 1024
#define HCLGE_COMM_CMDQ_TX_TIMEOUT_DEFAULT 30000
-#define HCLGE_COMM_CMDQ_TX_TIMEOUT_500MS 500000
+#define HCLGE_COMM_CMDQ_CFG_RST_TIMEOUT 1000000

enum hclge_opcode_type {
/* Generic commands */
--
2.30.0

2024-03-07 11:24:50

by Sunil Kovvuri Goutham

Subject: RE: [PATCH net 6/8] net: hns3: fix reset timeout under full functions and queues

> -----Original Message-----
> From: Jijie Shao <[email protected]>
> Sent: Thursday, March 7, 2024 6:31 AM
> To: [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]
> Cc: [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]
> Subject: [EXTERNAL] [PATCH net 6/8] net: hns3: fix reset timeout under full
> functions and queues
>
> From: Peiyang Wang <[email protected]>
>
> The cmdq reset command times out when all VFs are enabled and the
> queue is full. The hardware processing time exceeds the timeout set by the
> driver.
> In order to avoid the above extreme situations, the driver extends the reset
> timeout to 1 second.
>

Reviewed-by: Sunil Goutham <[email protected]>

But one observation, loop of udelay(1) for 1sec seems a lot, probably better to use usleep_range().

2024-03-08 09:38:37

by Jijie Shao

Subject: Re: [PATCH net 6/8] net: hns3: fix reset timeout under full functions and queues

on 2024/3/7 19:24, Sunil Kovvuri Goutham wrote:
>> In order to avoid the above extreme situations, the driver extends the reset
>> timeout to 1 second.
>>
> Reviewed-by: Sunil Goutham <[email protected]>
>
> But one observation, loop of udelay(1) for 1sec seems a lot, probably better to use usleep_range().

but this is protected by splock, so cannot use usleep_range().