From: Serge E. Hallyn <serue@xxxxxxxxxx>
This patch implements the s390 hook for sys_eclone.
The user-space clone-with-pids glue for s390 (clone_s390x.c
from the user-cr package) is now:
struct pid_set {
int num_pids;
pid_t *pids;
};
#define do_eclone(flags, pids, args, sz) \
( { \
register unsigned long int __r1 asm ("1") = (unsigned long int)(__NR_eclone); \
register unsigned long int __r2 asm ("2") = (unsigned long int)(flags); \
register unsigned long int __r3 asm ("3") = (unsigned long int)(args); \
register unsigned long int __r4 asm ("4") = (unsigned long int)(sz); \
register unsigned long int __r5 asm ("5") = (unsigned long int)(pids); \
register long int __result asm ("2"); \
__asm__ __volatile__( \
" svc 0\n" /* do __NR_eclone syscall */ \
" ltgr %%r2,%%r2\n" /* returned 0? */ \
" jnz 1f\n" /* if not goto label 1 */ \
" lg %%r3,0(%%r15)\n" /* get fnarg off stack into arg 1 */ \
" lg %%r2,8(%%r15)\n" /* get fn off stack int r3 basr*/ \
" lgr %%r1,%%r15\n" /* tmp store old stack pointer */ \
" aghi %%r15,-160\n" /* move the stack */ \
" stg %%r1,0(%%r15)\n" /* and save old stack pointer */ \
" basr %%r14,%%r3\n" /* call fn(arg) */ \
" svc 1\n" /* call exit */ \
" 1:\n" \
: "=d" (__result) \
: "d" (__r1), "0" (__r2), "d" (__r3), "d" (__r4), "d" (__r5) \
: "memory"); \
__result; \
} )
int clone_with_pids(int (*fn)(void *), void *child_stack, int flags,
struct pid_set *target_pids, void *arg)
{
struct clone_args clone_args, *ca = &clone_args;
u64 *s;
memset(ca, 0, sizeof(struct clone_args));
ca->nr_pids = target_pids->num_pids;
if (!child_stack) {
/* we could pass in null and then in eclone not
* call exit if child_stack was null, but we'll
* just malloc here */
int sz = 4*getpagesize();
child_stack = malloc(sz);
if (!child_stack)
return -ENOMEM;
child_stack += sz; /* we'll decrement before assigning */
}
ca->child_stack = (u64) child_stack;
s = (u64 *) ca->child_stack;
*--s = (u64) arg;
*--s = (u64) fn;
ca->child_stack -= 16;
return do_eclone(flags, target_pids->pids, ca,
sizeof(struct clone_args));
}
Changelog:
Nov 12: switch to latest (Nov 12) eclone format
Nov 10: use orig_gpr2, not gprs[2] for input arg 1
Signed-off-by: Serge E. Hallyn <serue@xxxxxxxxxx>
---
arch/s390/include/asm/unistd.h | 3 +-
arch/s390/kernel/compat_linux.c | 47 ++++++++++++++++++++++++++++++++++++++
arch/s390/kernel/process.c | 48 +++++++++++++++++++++++++++++++++++++++
arch/s390/kernel/syscalls.S | 1 +
4 files changed, 98 insertions(+), 1 deletions(-)
diff --git a/arch/s390/include/asm/unistd.h b/arch/s390/include/asm/unistd.h
index cb5232d..cbf6c7c 100644
--- a/arch/s390/include/asm/unistd.h
+++ b/arch/s390/include/asm/unistd.h
@@ -269,7 +269,8 @@
#define __NR_pwritev 329
#define __NR_rt_tgsigqueueinfo 330
#define __NR_perf_event_open 331
-#define NR_syscalls 332
+#define __NR_eclone 332
+#define NR_syscalls 333
/*
* There are some system calls that are not present on 64 bit, some
diff --git a/arch/s390/kernel/compat_linux.c b/arch/s390/kernel/compat_linux.c
index 0debcec..c3dc6bd 100644
--- a/arch/s390/kernel/compat_linux.c
+++ b/arch/s390/kernel/compat_linux.c
@@ -762,6 +762,53 @@ asmlinkage long sys32_write(unsigned int fd, char __user * buf, size_t count)
return sys_write(fd, buf, count);
}
+asmlinkage long sys32_eclone(void)
+{
+ int rc;
+ struct pt_regs *regs = task_pt_regs(current);
+ int args_size;
+ struct clone_args kca;
+ unsigned long flags;
+ int __user *parent_tid_ptr;
+ int __user *child_tid_ptr;
+ unsigned long __user child_stack;
+ unsigned long stack_size;
+ unsigned int flags_low;
+ struct clone_args __user *uca;
+ pid_t __user *pids;
+
+ flags_low = regs->orig_gpr2 & 0xffffffffUL;
+ uca = compat_ptr(regs->gprs[3]);
+ args_size = regs->gprs[4] & 0xffffffffUL;
+ pids = compat_ptr(regs->gprs[5]);
+
+ rc = fetch_clone_args_from_user(uca, args_size, &kca);
+ if (rc)
+ return rc;
+
+ if (kca.clone_flags_high)
+ return -EINVAL;
+ flags = flags_low;
+ parent_tid_ptr = (int __user *) kca.parent_tid_ptr;
+ child_tid_ptr = (int __user *) kca.child_tid_ptr;
+
+ stack_size = (unsigned long) kca.child_stack_size;
+ if (stack_size)
+ return -EINVAL;
+
+ child_stack = (unsigned long) kca.child_stack;
+ if (!child_stack)
+ child_stack = regs->gprs[15];
+
+ /*
+ * TODO: On 32-bit systems, clone_flags is passed in as 32-bit value
+ * to several functions. Need to convert clone_flags to 64-bit.
+ */
+ return do_fork_with_pids(flags, child_stack, regs, stack_size,
+ parent_tid_ptr, child_tid_ptr, kca.nr_pids,
+ pids);
+}
+
/*
* 31 bit emulation wrapper functions for sys_fadvise64/fadvise64_64.
* These need to rewrite the advise values for POSIX_FADV_{DONTNEED,NOREUSE}
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index 5417eb5..51f11a1 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -241,6 +241,54 @@ SYSCALL_DEFINE4(clone, unsigned long, newsp, unsigned long, clone_flags,
parent_tidptr, child_tidptr);
}
+SYSCALL_DEFINE0(eclone)
+{
+ int rc;
+ struct pt_regs *regs = task_pt_regs(current);
+ int args_size;
+ struct clone_args kca;
+ unsigned long flags;
+ int __user *parent_tid_ptr;
+ int __user *child_tid_ptr;
+ unsigned long __user child_stack;
+ unsigned long stack_size;
+ unsigned int flags_low;
+ struct clone_args __user *uca;
+ pid_t __user *pids;
+
+ flags_low = regs->orig_gpr2;
+ uca = (struct clone_args __user *) regs->gprs[3];
+ args_size = regs->gprs[4];
+ pids = (pid_t __user *) regs->gprs[5];
+
+ rc = fetch_clone_args_from_user(uca, args_size, &kca);
+ if (rc)
+ return rc;
+
+ if (kca.clone_flags_high)
+ return -EINVAL;
+
+ flags = flags_low;
+ parent_tid_ptr = (int __user *) kca.parent_tid_ptr;
+ child_tid_ptr = (int __user *) kca.child_tid_ptr;
+
+ stack_size = (unsigned long) kca.child_stack_size;
+ if (stack_size)
+ return -EINVAL;
+
+ child_stack = (unsigned long) kca.child_stack;
+ if (!child_stack)
+ child_stack = regs->gprs[15];
+
+ /*
+ * TODO: On 32-bit systems, clone_flags is passed in as 32-bit value
+ * to several functions. Need to convert clone_flags to 64-bit.
+ */
+ return do_fork_with_pids(flags, child_stack, regs, stack_size,
+ parent_tid_ptr, child_tid_ptr, kca.nr_pids,
+ pids);
+}
+
/*
* This is trivial, and on the face of it looks like it
* could equally well be done in user mode.
diff --git a/arch/s390/kernel/syscalls.S b/arch/s390/kernel/syscalls.S
index 30eca07..fb8708d 100644
--- a/arch/s390/kernel/syscalls.S
+++ b/arch/s390/kernel/syscalls.S
@@ -340,3 +340,4 @@ SYSCALL(sys_preadv,sys_preadv,compat_sys_preadv_wrapper)
SYSCALL(sys_pwritev,sys_pwritev,compat_sys_pwritev_wrapper)
SYSCALL(sys_rt_tgsigqueueinfo,sys_rt_tgsigqueueinfo,compat_sys_rt_tgsigqueueinfo_wrapper) /* 330 */
SYSCALL(sys_perf_event_open,sys_perf_event_open,sys_perf_event_open_wrapper)
+SYSCALL(sys_eclone,sys_eclone,sys_eclone_wrapper)
--
1.6.1
_______________________________________________
Containers mailing list
Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/containers
