Hello Hans Verkuil,
The patch fb7b37cf913c: "V4L/DVB (7535): saa717x: add new audio/video
decoder i2c driver" from Apr 9, 2008, leads to the following static
checker warning:
drivers/media/i2c/saa717x.c:155 saa717x_read()
warn: right shifting more than type allows
drivers/media/i2c/saa717x.c
133 static u32 saa717x_read(struct v4l2_subdev *sd, u32 reg)
134 {
135 struct i2c_client *client = v4l2_get_subdevdata(sd);
136 struct i2c_adapter *adap = client->adapter;
137 int fw_addr = (reg >= 0x404 && reg <= 0x4b8) || reg == 0x528;
138 unsigned char mm1[2];
139 unsigned char mm2[4] = { 0, 0, 0, 0 };
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
140 struct i2c_msg msgs[2];
141 u32 value;
142
143 msgs[0].flags = 0;
144 msgs[1].flags = I2C_M_RD;
145 msgs[0].addr = msgs[1].addr = client->addr;
146 mm1[0] = (reg >> 8) & 0xff;
147 mm1[1] = reg & 0xff;
148 msgs[0].len = 2;
149 msgs[0].buf = mm1;
150 msgs[1].len = fw_addr ? 3 : 1; /* Multibyte Registers contains
*only* 3 bytes */
151 msgs[1].buf = mm2;
152 i2c_transfer(adap, msgs, 2);
153
154 if (fw_addr)
155 value = (mm2[2] & 0xff) | ((mm2[1] & 0xff) >> 8) |
((mm2[0] & 0xff) >> 16);
The mask and shift doesn't make sense for chars (also it's reversed).
156 else
157 value = mm2[0] & 0xff;
My guess is the intention was:
if (fw_addr)
value = mm2[2] | mm2[1] | mm2[0];
else
value = mm2[0];
158
159 v4l2_dbg(2, debug, sd, "read: reg 0x%03x=0x%08x\n", reg,
value);
160 return value;
161 }
regards,
dan carpenter
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
