[Devel] Re: [PATCH] mark read_crX() asm code as volatile

[Kirill Korotaev]

Arjan,

I can experiment with any constraints if you suggest which one.

>From our experiments with gcc, it compares asm strings (sic!!!) to find matches
to be merged! Sigh...
Below are 2 programs which differ in one space in read_cr3_b() asm statement.
The first one compiles incorrectly, while 2nd one - correctly.

My personal feeling is that comparing asm strings is simply a "misfeature".

-------------------------- cut ----------------------------
static inline unsigned long read_cr3_a(void)
{
        unsigned long cr3;
        asm("movq %%cr3,%0" : "=r" (cr3));
        return cr3;
}

static inline unsigned long read_cr3_b(void)
{
        unsigned long cr3;
        asm("movq %%cr3,%0" : "=r" (cr3));
        return cr3;
}

static inline void write_cr3(unsigned long val)
{
        asm volatile("movq %0,%%cr3" :: "r" (val) : "memory");
}

void main()
{
        unsigned long c;
        c = read_cr3_a();
        write_cr3(c | 0x80);
        c = read_cr3_b();
        write_cr3(c | 0x100);
}
-------------------------- cut ----------------------------


-------------------------- cut ----------------------------
static inline unsigned long read_cr3_a(void)
{
        unsigned long cr3;
        asm("movq %%cr3,%0" : "=r" (cr3));
        return cr3;
}

static inline unsigned long read_cr3_b(void)
{
        unsigned long cr3;
        asm("movq  %%cr3,%0" : "=r" (cr3));
        return cr3;
}

static inline void write_cr3(unsigned long val)
{
        asm volatile("movq %0,%%cr3" :: "r" (val) : "memory");
}

void main()
{
        unsigned long c;
        c = read_cr3_a();
        write_cr3(c | 0x80);
        c = read_cr3_b();
        write_cr3(c | 0x100);
}
-------------------------- cut ----------------------------

Kirill

Arjan van de Ven wrote:
> On Tue, 02 Oct 2007 18:08:32 +0400
> Kirill Korotaev <dev at openvz.org> wrote:
> 
> 
>>Some gcc versions (I checked at least 4.1.1 from RHEL5 & 4.1.2 from
>>gentoo) can generate incorrect code with read_crX()/write_crX()
>>functions mix up, due to cached results of read_crX().
>>
> 
> 
> I'm not so sure volatile is the right answer, as compared to giving the
> asm more strict contraints....
> 
> asm volatile tends to mean something else than "the result has
> changed"....
> 
>