ACPI throttling: fix endian bug in acpi_read_throttling_status()
authorDan Carpenter <dan.carpenter@oracle.com>
Wed, 7 Mar 2012 11:57:36 +0000 (14:57 +0300)
committerLen Brown <len.brown@intel.com>
Fri, 30 Mar 2012 20:06:12 +0000 (16:06 -0400)
Using a u64 here creates an endian bug.  We store a u32 number in the
top byte which is a larger number than intended on big endian systems.
There is no reason to use a 64 bit data type here, I guess it was just
an oversight.

I removed the initialization to zero as well.  It's needed with a u64
but with a u32, the variable gets initialized properly inside the call
to acpi_os_read_port().

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Len Brown <len.brown@intel.com>
drivers/acpi/processor_throttling.c

index 605a2954ef175e563db7d212d2184c924b7b5013..1d02b7b5ade094dd386cabdfa8a39c959e829ea7 100644 (file)
@@ -769,7 +769,7 @@ static int acpi_read_throttling_status(struct acpi_processor *pr,
                                        u64 *value)
 {
        u32 bit_width, bit_offset;
-       u64 ptc_value;
+       u32 ptc_value;
        u64 ptc_mask;
        struct acpi_processor_throttling *throttling;
        int ret = -1;
@@ -777,12 +777,11 @@ static int acpi_read_throttling_status(struct acpi_processor *pr,
        throttling = &pr->throttling;
        switch (throttling->status_register.space_id) {
        case ACPI_ADR_SPACE_SYSTEM_IO:
-               ptc_value = 0;
                bit_width = throttling->status_register.bit_width;
                bit_offset = throttling->status_register.bit_offset;
 
                acpi_os_read_port((acpi_io_address) throttling->status_register.
-                                 address, (u32 *) &ptc_value,
+                                 address, &ptc_value,
                                  (u32) (bit_width + bit_offset));
                ptc_mask = (1 << bit_width) - 1;
                *value = (u64) ((ptc_value >> bit_offset) & ptc_mask);