Coindicentally I also got it to run. The prototype is in agreement with the usual clone() prototype. #if defined(__i386__) && defined(__NR_clone_with_pids) /* * libc doesn't support clone_with_pid() yet... * (see: http://lkml.indiana.edu/hypermail/linux/kernel/9604.3/0204.html) */ static int clone_with_pids(int (*fn)(void *), void *child_stack, int flags, struct target_pid_set *target_pids, void *arg) { long retval; void **newstack; /* * Set up the stack for child: * - the (void *) arg will be the argument for the child function * - the fn pointer will be loaded into ebx after the clone */ newstack = (void **) child_stack; *--newstack = arg; *--newstack = fn; __asm__ __volatile__( "movl %0, %%ebx\n\t" /* flags -> 1st (ebx) */ "movl %1, %%ecx\n\t" /* newstack -> 2nd (ecx)*/ "xorl %%edi, %%edi\n\t" /* 0 -> edi */ "xorl %%edx, %%edx\n\t" /* 0 -> edx */ "pushl %%ebp\n\t" /* save value of ebp */ "movl %2, %%ebp\n\t" /* flags -> 6th (ebp) */ : :"b" (flags), "c" (newstack), "r" (target_pids) ); __asm__ __volatile__( "int $0x80\n\t" /* Linux/i386 system call */ "testl %0,%0\n\t" /* check return value */ "jne 1f\n\t" /* jump if parent */ "popl %%ebx\n\t" /* get subthread function */ "call *%%ebx\n\t" /* start subthread function */ "movl %2,%0\n\t" "int $0x80\n" /* exit system call: exit subthread */ "1:\n\t" "popl %%ebp\t" /* restore parent's ebp */ :"=a" (retval) :"0" (__NR_clone_with_pids), "i" (__NR_exit) :"ebx", "ecx" ); if (retval < 0) { errno = -retval; retval = -1; } return retval; } #endif I added it to mktree.c and commited to user-cr.git :: ckpt-v16-dev, I also added the clone_with_pids patches to linux-cr.git :: ckpt-v16-dev It works well when not in a new namespace (mktree --pids < ckpt.image) assuming the the pids are avaiable. It doesn't work yet when in a new namespace, as the kernel code needs to be adjusted (for the restarting tasks to find the coordintor in an ancestor namespace). Oren. On Fri, 5 Jun 2009, Sukadev Bhattiprolu wrote: > > Attached are two files - > > cwp.c - implements clone_with_pids() library interface > cwp-test.c - a simple program to test the interface > > There maybe more optimal ways of implementing it though :-) > > If it makes sense, will submit as a patch to user-cr tree. > > Sukadev > _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers
