Isolation: Remove the syscall(SYS_getpid) wrapper.

When using SYS_clone we used the getpid kernel system call directly via
syscall(SYS_getpid) to avoid issues with cached pids.

However since we are now only using fork(2) (+ unshare(2) for
namespaces) we no longer need to call the kernel getpid directly as the
fork(2) will ensure the cached pid is invalidated.

Reviewed-by: Alejandro Colomar <[email protected]>
Signed-off-by: Andrew Clayton <[email protected]>

Author: ac000

src/nxt_process.h

@@ -13,17 +13,8 @@
 #endif
 
 
-#if (NXT_HAVE_LINUX_NS)
-/*
- * Old glibc wrapper for getpid(2) returns a cached pid invalidated only by
- * fork(2) calls.  As we use clone(2) for container, it returns the wrong pid.
- */
-#define nxt_getpid()                                                          \
-    syscall(SYS_getpid)
-#else
 #define nxt_getpid()                                                          \
     getpid()
-#endif
 
 typedef pid_t            nxt_pid_t;