Re: [PATCH] i2c: aspeed: Add multi-master use case support

From: Jae Hyun Yoo
Date: Mon Feb 11 2019 - 13:09:58 EST


Hmmm, the #ifdeffery quite increases with this patch. Maybe it is easier
to select I2C_SLAVE in Kconfig? I'll let you decide.

My code checkers found this:

CPPCHECK
drivers/i2c/busses/i2c-aspeed.c:694:10: warning: Variable 'ret' is assigned a value that is never used. [unreadVariable]


Hi Wolfram,

Yes, cppcheck pointed it out correctly. The assignment on the variable
'ret' is not needed at there. Actually, the code should not be wrapped
using I2C_SLAVE because a bus can also be in multi-master environment
even when slave is not enabled. So the warning can be removed by below
fix.

---
diff --git a/drivers/i2c/busses/i2c-aspeed.c b/drivers/i2c/busses/i2c-aspeed.c
index 9253ebccb008..882c4ab1d18a 100644
--- a/drivers/i2c/busses/i2c-aspeed.c
+++ b/drivers/i2c/busses/i2c-aspeed.c
@@ -655,7 +655,6 @@ static int aspeed_i2c_master_xfer(struct i2c_adapter *adap,
{
struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
unsigned long time_left, flags;
- int ret;

spin_lock_irqsave(&bus->lock, flags);
bus->cmd_err = 0;
@@ -664,6 +663,8 @@ static int aspeed_i2c_master_xfer(struct i2c_adapter *adap,
if (!bus->multi_master &&
(readl(bus->base + ASPEED_I2C_CMD_REG) &
ASPEED_I2CD_BUS_BUSY_STS)) {
+ int ret;
+
spin_unlock_irqrestore(&bus->lock, flags);
ret = aspeed_i2c_recover_bus(bus);
if (ret)
@@ -691,7 +692,7 @@ static int aspeed_i2c_master_xfer(struct i2c_adapter *adap,
if (bus->multi_master &&
(readl(bus->base + ASPEED_I2C_CMD_REG) &
ASPEED_I2CD_BUS_BUSY_STS))
- ret = aspeed_i2c_recover_bus(bus);
+ aspeed_i2c_recover_bus(bus);

return -ETIMEDOUT;
---

The reason of adding changes at line 658 and 667 is to remove below
cppcheck warning when the change at 694 is applied.

[drivers/i2c/busses/i2c-aspeed.c:670]: (style) The scope of the variable 'ret' can be reduced.

Can you please apply above changes on top of this patch? Or I can submit
v2 with it.

Thanks,
Jae