Re: [PATCH v4 08/16] xen: Add limited support of VMware's hyper-call rpc

["Jan Beulich" <JBeulich@suse.com>]

>>> On 12.09.14 at 15:37, <boris.ostrovsky@oracle.com> wrote:
> On 09/11/2014 02:36 PM, Don Slutz wrote:
>> +static inline uint16_t get_low_bits(uint32_t bits)
>> +{
>> +    return bits & 0xffff;
>> +}
>> +
>> +static inline uint16_t get_high_bits(uint32_t bits)
>> +{
>> +    return bits >> 16;
>> +}
>> +
>> +static inline uint32_t set_high_bits(uint32_t b, uint32_t val)
>> +{
>> +    return (val << 16) | get_low_bits(b);
>> +}

The names of all three functions (at least one of which I think I saw
in a reply to an earlier patch - code duplication?) are bogus: How
many high or low bits (and for the "high" case out of how many) are
we talking about here?

>> +static inline void set_status(struct cpu_user_regs *ur, uint16_t val)
>> +{
>> +    /* VMware defines this to be only 32 bits */
>> +    ur->rcx = (val << 16) | (ur->rcx & 0xffff);
> 
> Are you shifting val out of its size? Or will compiler automatically 
> promote this to uint32_t? (It does for me but I don't know if this is 
> something we can rely on).

All operands more narrow than int get promoted to int.

Jan