Visa Inquirer wrote:

> If I need to make a system call with 6 arguments I can use the code:
> "       push    %esi\n"
> "       push    %edi\n"
> "       push    %ebx\n"
> "       push    %ebp\n"
> "       movl    16+ 4(%esp),%eax\n" // syscall_num
> "       movl    16+ 8(%esp),%ebx\n" // arg1
> "       movl    16+12(%esp),%ecx\n" // arg2
> "       movl    16+16(%esp),%edx\n" // arg3
> "       movl    16+20(%esp),%esi\n" // arg4
> "       movl    16+24(%esp),%edi\n" // arg5
> "       movl    16+28(%esp),%ebp\n" // arg6
> "       int     $0x80\n"
> "       popl    %ebp\n"
> "       popl    %ebx\n"
> "       popl    %edi\n"
> "       popl    %esi\n"
> "       ret\n"
>
> But what happens for system calls like 'mmap'?
> void *mmap(void *start, size_t length, int prot, int flags,
>   int fd, off_t offset);
> Its last argument is off_t which is 64-bit.

Are you sure the offset really is 64-bits wide?
http://lxr.linux.no/source/arch/i38.../sys_i386.c#L43
asmlinkage long sys_mmap2(unsigned long addr, unsigned long len,
unsigned long prot, unsigned long flags,
unsigned long fd, unsigned long pgoff)

> Where should the high
> part of this offset fit in? There is not enough registers...
> Same for lseek:
> off_t lseek(int fildes, off_t offset, int whence);
> how will the high part of 'offset' argument be passed?

Are you sure the offset really is 64-bits wide?
asmlinkage off_t sys_lseek(unsigned int fd, off_t offset,
unsigned int origin)
typedef __kernel_off_t off_t;
typedef long __kernel_off_t;

llseek explicitly breaks a 64-bit argument into two 32-bit pieces:
http://lxr.linux.no/source/fs/read_write.c#L154
asmlinkage long sys_llseek(unsigned int fd, unsigned long offset_high,
unsigned long offset_low, loff_t __user * result,
unsigned int origin)

[ Post a follow-up to this message ]