Re: (E)GLIBC: inline syscalls

[Petr Salinger <Petr.Salinger@seznam.cz>]

> > There will be also significant slowdown, as we have to prepare aguments
> > on stack, alter stack to get exactly the same stack layout as the
> > standard syscall would be used. We will lose correct unwinding info, et 
> > cetera, et cetera.

> After some discussions on IRC, it seems the easiest way do to it, is to
> fill all the arguments in a table, replace %esp by the address of
> this table in the asm inline code, and restore the old value just after
> the syscall.

It is not async signal safe. We cannot gurantee the space under esp 
will not be used by a signal handler, just try execute code bellow.

In theory, we could prepare all the arguments in a table,
decrement esp by appropriate amount and copy the original table just 
above new esp, perform syscall and restore previous esp.

Petr

-------------------------------------------------------------------------------

#include <signal.h>
#include <stdio.h>

volatile int a,b;

#define gsp(a)                          \
({                                              \
  register long int result;                     \
  asm volatile (                                \
                "mov %%esp,%0\n\t"              \
                : "=a" (result));               \
  result;                                       \
})


void xsig(void)
{
  a = gsp(t);
}

void stack(int i)
{
 char c[4096];
 if (i) return stack(--i);
 b = gsp(p);
 raise(SIGUSR1);
 printf("%08x %08x \n", b, a);
}

int main(void)
{
  signal(SIGUSR1,xsig);
  b = gsp(p);
  raise(SIGUSR1);
  printf("%08x %08x \n", b, a);

  xsig();
  printf("%08x %08x \n", b, a);

  stack(15);

  b = gsp(p);
  raise(SIGUSR1);
  printf("%08x %08x \n", b, a);
}