Re: [PATCH v2 4/6] x86: Add clear_page_nocache

[Borislav Petkov <bp@alien8.de>]

On Mon, Aug 13, 2012 at 02:43:34PM +0300, Kirill A. Shutemov wrote:
> $ cat test.c
> #include <stdio.h>
> #include <sys/mman.h>
> 
> #define SIZE 1024*1024*1024
> 
> void clear_page_nocache_sse2(void *page) __attribute__((regparm(1)));
> 
> int main(int argc, char** argv)
> {
>         char *p;
>         unsigned long i, j;
> 
>         p = mmap(NULL, SIZE, PROT_WRITE|PROT_READ,
>                         MAP_PRIVATE|MAP_ANONYMOUS|MAP_POPULATE, -1, 0);
>         for(j = 0; j < 100; j++) {
>                 for(i = 0; i < SIZE; i += 4096) {
>                         clear_page_nocache_sse2(p + i);
>                 }
>         }
> 
>         return 0;
> }
> $ cat clear_page_nocache_unroll32.S
> .globl clear_page_nocache_sse2
> .align 4,0x90
> clear_page_nocache_sse2:
> .cfi_startproc
>         mov    %eax,%edx
>         xorl   %eax,%eax
>         movl   $4096/32,%ecx
>         .p2align 4
> .Lloop_sse2:
>         decl    %ecx
> #define PUT(x) movnti %eax,x*4(%edx)
>         PUT(0)
>         PUT(1)
>         PUT(2)
>         PUT(3)
>         PUT(4)
>         PUT(5)
>         PUT(6)
>         PUT(7)
> #undef PUT
>         lea     32(%edx),%edx
>         jnz     .Lloop_sse2
>         nop
>         ret
> .cfi_endproc
> .type clear_page_nocache_sse2, @function
> .size clear_page_nocache_sse2, .-clear_page_nocache_sse2
> $ cat clear_page_nocache_unroll64.S
> .globl clear_page_nocache_sse2
> .align 4,0x90
> clear_page_nocache_sse2:
> .cfi_startproc
>         mov    %eax,%edx

This must still be the 32-bit version becaue it segfaults here. Here's
why:

mmap above gives a ptr which, on 64-bit, is larger than 32-bit, i.e. it
looks like 0x7fffxxxxx000, i.e. starting from top of userspace.

Now, the mov above truncates that ptr and the thing segfaults.

Doing s/edx/rdx/g fixes it though.

Thanks.

-- 
Regards/Gruss,
Boris.