提交 b0338080 编写于 作者: M Marek Vasut 提交者: Heiko Schocher

i2c: designware: Avoid overwriting the cmd_data register

Make sure the driver writes the cmd_data register only once per
read transfer instead of doing so potentially repeatedly.

In case the read transfer didn't finish quickly enough, the loop
in the driver code would spin fast enough to write the same value
into the cmd_data register again before re-checking whether the
transfer completed, which would cause another spurious read transfer
on the bus.
Signed-off-by: NMarek Vasut <marex@denx.de>
Cc: Stefan Roese <sr@denx.de>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Heiko Schocher <hs@denx.de>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
Cc: Chin Liang See <clsee@altera.com>
上级 5ac5861c
......@@ -249,6 +249,7 @@ static int __dw_i2c_read(struct i2c_regs *i2c_base, u8 dev, uint addr,
int alen, u8 *buffer, int len)
{
unsigned long start_time_rx;
unsigned int active = 0;
#ifdef CONFIG_SYS_I2C_EEPROM_ADDR_OVERFLOW
/*
......@@ -274,18 +275,28 @@ static int __dw_i2c_read(struct i2c_regs *i2c_base, u8 dev, uint addr,
start_time_rx = get_timer(0);
while (len) {
if (len == 1)
writel(IC_CMD | IC_STOP, &i2c_base->ic_cmd_data);
else
writel(IC_CMD, &i2c_base->ic_cmd_data);
if (!active) {
/*
* Avoid writing to ic_cmd_data multiple times
* in case this loop spins too quickly and the
* ic_status RFNE bit isn't set after the first
* write. Subsequent writes to ic_cmd_data can
* trigger spurious i2c transfer.
*/
if (len == 1)
writel(IC_CMD | IC_STOP, &i2c_base->ic_cmd_data);
else
writel(IC_CMD, &i2c_base->ic_cmd_data);
active = 1;
}
if (readl(&i2c_base->ic_status) & IC_STATUS_RFNE) {
*buffer++ = (uchar)readl(&i2c_base->ic_cmd_data);
len--;
start_time_rx = get_timer(0);
active = 0;
} else if (get_timer(start_time_rx) > I2C_BYTE_TO) {
return 1;
return 1;
}
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册